Published 5:00 am Monday, August 22, 2022
By Staff Reports
NATCHEZ — The 31st season of the Natchez Festival of Music brought us the music of Queen, an acrobatic opera, and the world premiere of a harp and saxophone composition by the Kontra Duo. We said goodbye to the Impromptu Piano Trio whose violinist retired after the event. Concert goers were awed by the sounds of ancient songs from China played by pianist Zhaolei Xie and enjoyed music by Stephen Sondheim and the songs of Carole King and Judy Garland.
Natchez Festival of Music organizers rocked them into planning a new season of concerts that starts in October.
The 2022-2023 Natchez Festival of Music, the 32nd season, will include 14 concerts of jazz, rock, opera, Broadway songs, chamber music, gospel and more.
Here are some important dates coming up, just around the corner.
October 6, 2022, Red Wine, Rossini and Rampal
The Festival, in conjunction with Fall Pilgrimage, will present Red Wine, Rossini and Rampal, celebrating the centennial of French flutist Jean-Pierre Rampal. The renowned musician brought the flute to new prominence as a solo instrument adaptable to a wide range of music styles. Performers that night will be flutists Fabio Angelo Colajanni and Danilo Mezzadri, vocalist Susan Ruggiero and pianist Brian Murphy. The Carriage House is the venue for this gala chamber concert.
October 27, 2022, Gianni Schicchi and The Medium
Opera fans won’t want to miss this one.
The Festival will present two short operas: Giacomo Puccini’s comic opera Gianni Schicchi and Gian Carlo Menotti’s The Medium. Gianni Schicchi involves an untimely death, a fabricated will, thieving relatives and a swindler. The Medium, first performed in 1946, deals with the occult, guilt and madness.
Of course, there is always romance lurking in the wings. Puccini set his opera, first performed in 1918, in Renaissance Florence, but the Natchez’s Festival of Music sets this production in contemporary Gulfport, Mississippi. Both operas will be in English. Menotti is best known for his Christmas opera, “Amahl and the Night Visitors.” For those who would like a preview of music from these operas, try a web search of “O mio bambino caro” from Gianni Schicchi and “Monica’s Waltz” from The Medium.
November 6, 2022, Handel’s Messiah
Jointly sponsored with Trinity Episcopal Church to celebrate the congregation’s 200th anniversary, this concert of excerpts from George Frederic Handel’s Messiah will be performed by a 40-member choir and orchestra.
The grand oratorio with its rousing Hallelujah Chorus was composed in 1741, and first performed in Dublin, Ireland, on Easter, 1742, but today the Messiah is more often presented during the Christmas season.
December 2, 2022, Elvis Christmas
For this Festival’s Christmas Cabaret, Elvis—Victor Trevino—will be in the room.
Trevino is the winner of the 2022 Ultimate Elvis Competition.
Coming up in 2023
Concerts in 2023 will include a commemoration of Martin Luther King weekend with Voices of Freedom on Jan. 15. The Festival’s Valentine gala is scheduled for Feb. 10, featuring “Be My Love,” a tribute to Mario Lanza, performed by Peter Lake, Stacey Trenteseaux and Tyler Kemp. A March chamber concert by the Herrington Jazz Quintet is planned for Spring Pilgrimage.
On April 29, the Johann Strauss operetta Die Fledermaus will usher in the 32nd season of May music.
Watch for ticket information for each concert at natchezfestivalofmusic.com.
The Natchez Festival of Music is a 501(c)3 nonprofit corporation supported by donations, grants and ticket sales.
Do you plan to attend the Blues and Soul Superbowl on the Natchez Bluff Oct. 7 and 8?
View Results
May 2022 season went so well, Natchez Festival of Music is coming back for an encore this fall – Mississippi's Best Community Newspaper | Mississippi's Best Community Newspaper – Natchez Democrat
Understanding How Mindfulness Meditation Reduces Pain – Breastcancer.org
Mindfulness meditation seems to ease pain by allowing people to separate pain from things they’re experiencing, according to a small study.
The research was published online on July 7, 2022, by the journal Pain. Read the abstract of “Disentangling self from pain: mindfulness meditation-induced pain relief is driven by thalamic-default mode network decoupling.”
Jon Kabat-Zinn, PhD, is internationally known for his work as a scientist, writer, and meditation teacher who had a goal of bringing mindfulness into the mainstream of medicine and society. He is professor of medicine emeritus at the University of Massachusetts Medical School, where he founded its world-renowned Mindfulness-Based Stress Reduction Clinic in 1979.
Although definitions of mindfulness vary slightly (depending on who you’re talking to), Dr. Kabat-Zinn describes it as an awareness that arises through paying attention, on purpose, in the present moment, non-judgmentally, in the service of self-understanding and wisdom. Essentially, mindfulness is a special type of awareness.
During mindfulness meditation, you don’t ignore distracting thoughts and feelings. Instead, you acknowledge and observe them without judgment. By disconnecting yourself from your thoughts and feelings, you can look at them objectively — as if they were not yours — and gain insights about them.
Research has suggested that mindfulness meditation can ease the following side effects of breast cancer treatment:
pain
depression
fatigue
Still, researchers haven’t been able to figure out how mindfulness meditation helps to ease these side effects.
The small study included 40 people who were considered healthy, pain-free, and had never meditated. Half the people were men and half were women, and their average age was 30.
On the study’s first day, all the people had a brain scan while researchers used a probe to apply painful heat to the backs of their right calves. The people could move away from the probe and stop the pain at any time. The researchers applied painful heat to the people several times.
The people had to rate their average pain levels.
The researchers then split the people into two even groups:
20 people completed four 20-minute mindfulness meditation training sessions. They were instructed to focus on their breath and detach from themselves by acknowledging their thoughts, sensations, and emotions, but then letting them go without judgment or reacting to them.
20 people spent four 20-minute sessions listening to an audio recording of The Natural History and Antiquities of Selborne. The people were not allowed to sleep, use their phones, or talk during the sessions.
On the study’s last day, both groups had a brain scan again while researchers applied painful heat to the same spot on their legs. The researchers told the people in the mindfulness meditation group to meditate while the heat was applied. The researchers told the people in the audio book group to rest with their eyes closed.
After asking all the people to rate their average pain levels again, the researchers found that people in the meditation group had a:
32% reduction in pain intensity
33% reduction in pain unpleasantness
After analyzing the people’s brain activity during the heat application, the researchers also found that people who meditated had less connection between two areas of the brain: the thalamus and the default mode network.
The thalamus delivers sensory information, such as heat, cold, and pain, to the rest of the brain.
The default mode network is a collection of brain areas that are most active when a person is daydreaming or processing their own thoughts and feelings as opposed to processing information from the outside world.
One of the areas in the default mode network is the precuneus, which is involved in self-awareness and is one of the first areas to stop functioning when a person loses consciousness. Another area is the ventromedial prefrontal cortex, which includes several smaller regions that work together to process how you place value on your experiences.
People who had more pain relief had less connection between these areas, as well as less activity in the default mode network areas.
“We were really excited to confirm that you don’t have to be an expert meditator to experience these analgesic effects,” senior author Fadel Zeidan, PhD, associate professor of anesthesiology at the University of California, San Diego School of Medicine, said in a statement. “This is a really important finding for the millions of people looking for a fast-acting and non-pharmacological treatment for pain.
“For many people struggling with chronic pain, what often affects their quality of life most is not the pain itself, but the mental suffering and frustration that comes along with it,” Dr. Zeidan continued. “Their pain becomes a part of who they are as individuals — something they can’t escape — and this exacerbates their suffering.”
If you’re experiencing pain from breast cancer treatment or from the cancer itself, this study offers promising results and echoes results from other small studies: Mindfulness meditation can help ease pain.
If you’re interested in trying mindfulness meditation but aren’t sure how to start, you can:
Learn more about Meditation.
Listen to a podcast with Laura Cohen Romano about mindfulness.
You also can ask for information at your cancer treatment center or read more about Jon Kabat-Zinn’s guided Mindfulness Meditation practices.
Written by: Jamie DePolo, senior editor
— Last updated on August 19, 2022, 2:40 PM
By submitting my email, I agree to receive newsletters from Breastcancer.org
A confirmation has been sent to your email address.
About us
Get in touch
You can help
Numerical methodology to evaluate unipolar saturation current limit of DC corona discharge in complex geometries | Scientific Reports – Nature.com
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Scientific Reports volume 12, Article number: 14252 (2022)
Metrics details
The unipolar saturation current limit (({I}_{sat})) gives an upper limit to the corona current that can be obtained from a unipolar corona discharge. Therefore, it implies a theoretical limit to the performance of unipolar corona discharge devices. However, it has not been widely used in practice because it is difficult to deal with complex discharge configurations in an analytical way. This study aims to establish and validate a numerical methodology to evaluate the maximum current, which numerically imitates the unipolar saturation current limit. It was shown that the maximum current has the same mathematical definition as the unipolar saturation current. For validation, the maximum current was compared with an analytical solution of the Poisson equation for the coaxial cylinders configuration. The differences between the maximum current and unipolar saturation current limit for the coaxial cylinders, pin-to-plane, and single wire-to-plane configurations were discussed in terms of the assumptions used in the semi-analytical derivation of the unipolar saturation current limit. The validated methodology was applied to a multiple wire-to-plane configuration, for which a semi-analytical expression of the unipolar saturation current limit has not yet been developed. The effects of geometric and operation parameters on the maximum currents of the multiple wire-to-plane configuration were analyzed. The results were regressed into a single formula.
A direct current (DC) corona discharge has various modes depending on the operation conditions such as an electrode configuration, magnitude and polarity of an applied voltage, and composition and velocity of a gas1,2,3,4,5,6. Goldman et al.7 categorized various types of the DC corona discharge into unipolar and bipolar corona discharges, depending on whether one or both ion polarities exist in the inter-electrode space. In a unipolar corona discharge, such as the positive Hermstein’s glow, negative Trichel pulse, and negative pulseless glow corona discharges, ionization of gas molecules occurs only in a very small region near the discharge electrode, namely the ionization region. Only charge carriers with a single polarity (i.e. unipolar) drift through the remaining inter-electrode space, namely the drift region. Charges with opposite polarity are immediately absorbed on the surface of the discharge electrode. Because the drift region is filled only with unipolar ions, the corona current results solely from the movement of the unipolar ions, and this discharge mode is called the unipolar corona discharge. As the corona current increases, however, a discharge mode transition from unipolar to bipolar corona discharge may occur under certain conditions7,8. In a bipolar corona discharge, such as the streamer corona discharge, a conducting plasma is produced so rapidly that charges with opposite polarity cannot be completely absorbed at the discharge electrode7. Consequently, plasma filaments, each of which is called a streamer, grow out of the ionization region toward the ground electrode, resulting in secondary ionization in the drift region. In other words, ions of both polarities (i.e. bipolar) coexist in and drift through the entire inter-electrode space. It should be noted that depending on the context, the terms unipolar and bipolar can refer to not only the polarity of ions as used above, but also the polarity of the applied voltage; the bipolar corona often means the simultaneous operation of the positive and negative (i.e. bipolar) corona discharges, whereas the unipolar corona refers to the operation using only one of the positive or negative (i.e. unipolar) applied voltage9.
The unipolar-to-bipolar discharge mode transition is of great interest in various applications of DC corona discharges because the steady, uniform, and homogeneous characteristics of the unipolar mode are utilized in many applications. The unipolar corona discharge devices usually show better performances with higher corona currents, because the amount of unipolar ions to be used increases. However, if the corona current is so high that the discharge mode shifts to the bipolar regime, the uniformity and the homogeneity of the unipolar ion distribution are lost, degrading the performance of the devices. For instance, the collection efficiency of electrostatic precipitators is significantly reduced when the discharge mode shifts from the unipolar to the bipolar regime because the charges acquired by particulate matters are neutralized10. Kim and Hwang11 showed that the discharge mode transition from the unipolar to the bipolar lowered the electric-to-kinetic energy conversion efficiency and the thrust performance of electrohydrodynamic flow generators. Recently, Wang et al.12 utilized corona discharges to achieve a fast electrostatic printing of multifunctional materials to be attached on the human skin. A higher applied voltage may be preferred to achieve a faster printing, whereas it is essential to maintain the discharge in the unipolar regime for homogeneous and stable printing. Therefore, it is expected that the maximum performance of the unipolar corona discharge devices is closely related to the unipolar-to-bipolar discharge mode transition, or the maximum amount of the unipolar ions that can be obtained from the discharge conditions of interest.
In this regard, Sigmond8 semi-analytically derived an upper limit to the corona current that can be obtained from a unipolar corona discharge, which is called the unipolar saturation current limit (({I}_{sat})). If the experimentally measured corona current (({I}_{exp})) exceeds ({I}_{sat}), then the corona discharge of interest is, at least partially, bipolar. In this case, ({I}_{exp}) does not result solely from the movement of the unipolar ions; at least a portion of ({I}_{exp}) results from electrons, or secondary ionization in the drift region, or both of them8. In other words, ({I}_{sat}) implies a theoretical limit to the performance of unipolar corona discharge devices.
To date, the evaluation of ({I}_{sat}) has been based on the unipolar charge drift formula, which describes the time-dependent change in unipolar charge density along the paths of ions8. Using this formula, the estimation of ({I}_{sat}) reduces to the estimation of the lengths of electric field lines distributed in the inter-electrode space13. Many studies have analytically calculated the shape and length of electric field lines to obtain expressions for ({I}_{sat}) in the discharge configurations of coaxial cylinders8, pin-to-plane8,14, and single wire-to-plane14,15,16. However, it is difficult to derive analytical solutions for electric field line distributions in complex configurations. As a result, ({I}_{sat}) has not been widely used in the field of corona discharges in spite of its potential usefulness.
Recently, Kim et al.17 suggested an approach to obtain ({I}_{sat}) without using the unipolar charge drift formula, by proposing the concept of maximum current (({I}_{max})). They described the calculation procedure to obtain ({I}_{max}) and showed that ({I}_{max}) had a similar value to ({I}_{sat}) for a single wire-to-plane type corona discharge. Later, Kim and Hwang11 simplified the calculation procedure by omitting the step of obtaining the size of the ionization region, which requires several iterative simulations.
This study aims to establish, validate, and apply a numerical methodology to evaluate ({I}_{max}) for a complex discharge configuration. The calculation procedure to obtain ({I}_{max}) was adopted from the method proposed by Kim and Hwang11. The procedure was validated by comparing the calculated values of ({I}_{max}) with an analytical solution of the Poisson equation for coaxial cylinders configuration. After validation, this methodology was applied to pin-to-plane and single wire-to-plane configurations, and the values of ({I}_{max}) were compared with those of ({I}_{sat}) given by Sigmond8,15. Finally, ({I}_{max}) of a multiple wire-to-plane configuration was evaluated. The dependence of ({I}_{max}) on various geometric parameters and applied voltage was examined for this configuration.
Corona discharge is a complex multi-physical phenomenon, and various processes are underlying, including ionization of neutral molecules by collision with electron, chemical reactions between numerous charged and neutral species, and transports of those discharge-induced species by electric field, ambient flow and ionic wind. Accordingly, numerical modeling of corona discharge in principle requires transport equations for electron, electron energy, and each species, and a set of reactions describing the chemical behavior of each species5,18. For a unipolar corona discharge, however, a drastically simplified model can be employed. The simplified model considers only one species: air ion. The air ion is a fictitious unipolar ion which represents various charged species generated from the unipolar corona discharge. Thus, all electrical behaviors of charged species in the unipolar corona discharge are implied in the behavior of the air ion. Within the simplified model, the generation and transport of neutral species are not considered. Therefore, chemical reactions occurring between various charged and neutral species are also not solved. Instead, the effect of the chemical reactions is implicitly included in the boundary condition of unipolar ion charge density. In the same manner, the behavior of electrons is also not considered. Thus, the potential effects of high gradient of electric field on the electron drift velocity and local field approximations cannot be discussed with the simplified model.
The simplified model consists of the steady-state governing equations for the electric field and distribution of the air ion given as follows:
where (V) is the electric potential, (q) is the space charge density of the unipolar air ions generated by the corona discharge, (overrightarrow{j}) is the unipolar corona current density, (mu) is the electrical mobility of the ions, and (overrightarrow{E}=-nabla V) is the electric field.
For the boundary condition of (V), constant values were imposed to the surfaces of discharge ((V={V}_{a})) and ground ((V=0)) electrodes, respectively. To specify the boundary condition of (q) at the discharge electrode, the ionization region was assumed to be circular and concentric with the center of curvature of the discharge electrode. The radius of the ionization region (({R}_{i})) was assumed to be constant. For a pin-type discharge electrode, the ionization region was assumed to be a sphere with radius ({R}_{i}), concentric with the center of curvature of the tip of the pin. Similarly, for a wire-type discharge electrode, the ionization region was assumed to be a cylinder with radius ({R}_{i}), coaxial with the central axis of the wire. Then, a uniform and constant charge density ({q}_{i}) was specified in the ionization region. At the ground electrode, the zero Neumann condition ((partial qtext{/}partial n=0)) was imposed. After the (V) and (q) fields converged, the corona current was calculated by integrating (overrightarrow{j}) over the ground electrode.
As mentioned in “Unipolar corona discharge simulation” section, in the simplified model, chemical aspects of corona discharge such as ionization are implicitly included in the boundary value of (q) at the discharge electrode, ({q}_{i}). Therefore, the problem of solving the unipolar corona discharge using the simplified model is equivalent to the problem of finding an appropriate value of ({q}_{i}); the remaining is to solve the transport equations (Eqs. (1) and (2)) fluid-mechanically using the boundary condition. Accordingly, the unipolar corona current calculated by numerical simulation with the simplified model (({I}_{num})) for a given applied voltage ({V}_{a}) can be expressed as a function of ({q}_{i}) as follows:
If a value of ({q}_{i}^{^{prime}}) is appropriately determined to represent a real corona discharge, (f) is expected to be close to the experimentally measured corona current, that is, ({I}_{num}^{^{prime}}=fleft({q}_{i}^{^{prime}};{V}_{a}right)approx {I}_{exp}).
The unipolar saturation current ({I}_{sat}) is an upper limit to the corona current obtainable from the unipolar corona discharge for a given ({V}_{a}). Intuitively, ({I}_{num}) is expected to increase with increasing ({q}_{i}). Therefore, in the viewpoint of Eq. (3), ({I}_{sat}) can be expressed as follows:
Kim and Hwang11 numerically solved the simplified model (Eqs. (1) and (2)) of the unipolar corona discharge. They fixed the value of ({R}_{i}) and calculated ({I}_{num}) by increasing ({q}_{i}) stepwise by (Delta {q}_{i}). Their results showed that when ({q}_{i}) was sufficiently high, ({I}_{num}) saturated at a certain value and did not increase even though ({q}_{i}) increased further. Kim and Hwang11 defined the saturated value of ({I}_{num}) as the maximum current, ({I}_{max}). This description is exactly identical to Eq. (4). Therefore, ({I}_{max}) can also be written as follows:
In other words, ({I}_{sat}) and ({I}_{max}) are the semi-analytical and numerical implementations of the same expression, respectively. Thus, ({I}_{sat}) and ({I}_{max}) are expected to be equal to each other. In this study, ({I}_{max}) was calculated by following the calculation procedure suggested by Kim and Hwang11. The flowchart of the calculation procedure is given in Supplementary Fig. S1.
Sigmond8,15 proposed the semi-analytical expressions of ({I}_{sat}) for the coaxial cylinders (Eq. (6)), pin-to-plane (Eq. (7)), and single wire-to-plane (Eq. (8)) configurations as follows:
Feng19 reported a fully-analytical solution of the Poisson equation for corona discharge for the cylinders configuration using one-dimensional symmetricity. For other geometries, fully-analytical solutions are not available. Therefore, in this study, the calculated value of ({I}_{max}) was validated using the current–voltage ((I)–(V)) relationship of the coaxial cylinders configuration. In “Analytical current–voltage curve for coaxial cylinders configuration” section, the fully-analytical (I)–(V) relationship of the coaxial cylinders configuration reported in Feng19 was introduced. In “Maximum current calculation for coaxial cylinders configurations” section, ({I}_{max}) was calculated for the coaxial cylinders configuration. After that, the calculated values of ({I}_{max}) were compared with ({I}_{sat}) and fully-analytically computed corona current, which will be denoted by ({I}_{analyt}).
Based on the historic works of Townsend20 and Thomson and Thomson21, Feng19 obtained a fully-analytical solution of the Poisson equation to express the (I)–(V) characteristics for the coaxial cylinders configuration. Within the simplified model (Eqs. (1) and (2)), the analytical (I)–(V) relationship for the coaxial cylinders configuration is given as follows19:
where ({R}_{w}) and ({R}_{o}) are the radii of the inner wire and outer cylinder electrodes, respectively, and ({I}_{analyt}) is the analytical corona current per unit length along the axial direction given as follows:
({C}_{1}) is a non-negative integral constant of the Poisson equation and expressed as follows:
where ({E}_{w}) is the electric field strength at the surface of the inner wire electrode. Several studies employed the same approach with more details on the behaviors of electrons and ions such as ionization and attachment, and presented similar expressions to Eq. (9)22,23.
Because all other parameters are supposed to be given, Eq. (9) calculates the value of ({V}_{a}) required to generate the corona current ({I}_{analyt}) if the integral constant ({C}_{1}) is given. To calculate ({C}_{1}), the value of ({E}_{w}) should first be determined. The simplest and most common method to determine ({E}_{w}) is to apply the Kaptzov hypothesis, which states that for all ({V}_{a}) greater than the corona onset voltage, the value of ({E}_{w}) remains constant at the onset value ({E}_{on}) that is usually calculated by the Peek equation. The Peek equation for a cylindrical discharge electrode at the standard temperature and pressure is given as follows:
where (f) is the roughness factor of the electrode surface.
Two coaxial cylinders configurations were adopted for validation: ({R}_{w}) = 0.1 mm and ({R}_{o}) = 25.5 mm24, and ({R}_{w}) = 3.19 mm and ({R}_{o}) = 290.5 mm25 (see Fig. 1a). Figure 2 shows the comparisons between ({I}_{exp}) and ({I}_{analyt}) in the coaxial cylinders configurations. The value of (mu) was set to 2.0 (times) 10–4 m2 V−1 s−1 as reported24,25. The fully-analytical (I)–(V) curve given by Feng19 matched with the experimental values24,25 with (f) = 0.9 and 0.6, respectively. The value of (f) is 1.0 for the polished clean surface and 0.5–0.7 for dirty scratched wires26. Figure 2 also shows the ({I}_{sat}) curves for each case. For the case of ({R}_{w}) = 0.1 mm and ({R}_{o}) = 25.5 mm, ({I}_{exp}) was lower than ({I}_{sat}). It should be noted that ({I}_{exp}) lower than ({I}_{sat}) does not mean a unipolar corona discharge. For the case of ({R}_{w}) = 3.19 mm and ({R}_{o}) = 25.5 mm, ({I}_{exp}) exceeded ({I}_{sat}) at high ({V}_{a}). Therefore, it can be said that for the operation condition at which ({I}_{exp}) was higher than ({I}_{sat}), at least a portion of ({I}_{exp}) resulted from a bipolar corona discharge.
Computational domains and schematic diagrams of the ionization region: a coaxial cylinders, b pin-to-plane, c single wire-to-plane, and d multiple wire-to-plane configurations.
({I}_{exp}), ({I}_{analyt}), and ({I}_{sat}) of the coaxial cylinders configurations: ({R}_{i}) = 0.1 mm and ({R}_{o}) = 25.5 mm, and ({R}_{i}) = 3.19 mm and ({R}_{o}) = 290.5 mm.
Following the procedure described in “Maximum current calculation” section, ({I}_{max}) was calculated for the coaxial cylinders configurations. Figure 3a shows the saturation profiles of ({I}_{num}) versus ({q}_{i}) for positive discharge under the conditions ({R}_{w}) = 0.1 mm, ({R}_{o}) = 25.5 mm, and ({R}_{i}) = 3 ({R}_{w}). For any value of ({V}_{a}), ({I}_{num}) became saturated as ({q}_{i}) increased (see square markers in Fig. 3a), as in Kim and Hwang11. Thus, the value of ({I}_{num}) corresponding to the square marker can be accepted as ({I}_{max}) for a given ({V}_{a}). Figure 3a also shows (right vertical axis) the profiles of the reduced electric field strength at the surface of the discharge electrode (({E}_{w}text{/}N), where (N) is the number density of neutral gas molecules) versus ({q}_{i}). When ({q}_{i}) increased, ({E}_{w}text{/}N) decreased while ({I}_{num}) increased. At any saturation point, ({E}_{w}text{/}N) was below 120 Td (1 Td = 10−21 V m2), which is equivalent to the breakdown electric field strength (({E}_{b})) of air at the standard temperature and pressure. The ionization coefficient is greater than the attachment coefficient where (Etext{/}N>{E}_{b}text{/}N=120 mathrm{Td}) 27,28. Electrons are sufficiently accelerated and energetic enough to ionize the colliding neutral molecules when (E>{E}_{b}). On the other hand, when (E<{E}_{b}), electrons are not sufficiently accelerated. Insufficiently energetic electrons cannot ionize neutral molecules, but they are attached to the molecules. Therefore, it is reasonable to assume that ({E}_{w}) should also remain greater than ({E}_{b}) even if the unipolar corona discharge is in its saturated state. The circular markers in Fig. 3a indicate ({E}_{w}text{/}N) = 120 Td, and the cross markers indicate the corresponding ({I}_{num}). Hereinafter, if ({E}_{w}text{/}N) reached 120 Td, the corresponding ({I}_{num}) was accepted as ({I}_{max}) even though ({I}_{num}) was not yet completely saturated.
Results for the coaxial cylinders configuration: a saturation profiles of ({I}_{num}) and ({E}_{w}text{/}N) for ({R}_{w}) = 0.1 mm, ({R}_{o}) = 25.5 mm, and ({R}_{i}) = 3 ({R}_{w}), and b ({I}_{anlayt,b}), ({I}_{sat}), and ({I}_{max}) under various calculation conditions.
As mentioned in “Analytical current–voltage curve for coaxial cylinders configuration” section, the value of ({E}_{w}) should be first specified to calculate ({I}_{analyt}). As discussed with Fig. 2, ({E}_{on}) was used as ({E}_{w}) for a real corona discharge. However, for the saturated state, the representative value of ({E}_{w}) has not been reported. In other words, it is unclear which value should be used as ({E}_{w}) in Eq. (11) to calculate ({I}_{analyt}) representing ({I}_{sat}) and ({I}_{max}). Obviously, the Kaptzov hypothesis cannot be employed for the saturated state, as shown in Fig. 2; ({I}_{analyt}) calculated by using ({E}_{w}) = ({E}_{on}) was significantly different from ({I}_{sat}) for the case ({R}_{w}) = 0.1 mm and ({R}_{o}) = 25.5 mm. Sigmond8 derived the semi-analytical expression of ({I}_{sat}) without using information on ({E}_{w}). Instead, an average electric field strength (e.g., (overline{E }={V}_{a}text{/}{R}_{o}) for the coaxial cylinders configuration) was used. Therefore, without a better option, the breakdown electric field strength ({E}_{b}) was used as ({E}_{w}) for the saturated state, as also used for ({I}_{max}). ({I}_{analyt,b}) given in Fig. 3b indicate the value of ({I}_{analyt}) calculated using ({E}_{w}) = ({E}_{b}). As shown in Fig. 3b, ({I}_{max}) was the same as ({I}_{analyt,b}) for all ({V}_{a}).
The dashed lines in Fig. 3b indicate the ({I}_{sat}) curves. ({I}_{max}) and ({I}_{sat}) had the similar magnitudes as in Kim et al.17, but with some deviations. The semi-analytical expressions for ({I}_{sat}) given in Eqs. (6)–(8) were originally derived under various assumptions such as the Deutsch assumption, which states that the space charge does not distort the shape of the space-charge-free (i.e. Laplacian) electric field lines, but only scale the strength of the electric field; and the small discharge electrode assumption, which can be expressed as ({R}_{w}text{/}{R}_{o}ll 1) 8. Because the Deutsch assumption is valid for the coaxial cylinders configuration, the deviation between ({I}_{sat}) and ({I}_{max}) might result from the invalidity of the small discharge electrode assumption. For simple comparison, ({I}_{max}) was assumed to have the same form as ({I}_{sat}), i.e. ({I}_{max}={C}_{2}{V}_{a}^{gamma }), where ({C}_{2}) is an arbitrary constant. Then, the exponent (gamma) was obtained from the slope of the least-squares fit line on the logarithmic scale. For the cases ({R}_{w}text{/}{R}_{o}=0.1/25.5=0.004) and ({R}_{w}text{/}{R}_{o}=3.19/290.5=0.011), the value of (gamma) was 2.37 and 2.86, respectively. In other words, the smaller ({R}_{w}text{/}{R}_{o}) was, the closer (gamma) was to the semi-analytical value, 2 (Eq. (6)), which was consistent with the small discharge electrode assumption.
The effects of three different ({R}_{i}) values (1.5 ({R}_{w}), 3 ({R}_{w}), and 5 ({R}_{w})) on ({I}_{max}) were tested. For all conditions, ({R}_{i}) had a negligible effect on ({I}_{max}), in agreement with Kim and Hwang11. For a real corona discharge, the shape and size of the ionization region are probably not constant, and the distribution of charges in the ionization region is also not uniform. Thus, numerical modeling of the real corona discharge should consider those effects. However, ({I}_{max}) is an attempt to find a limit value of corona current obtainable under a certain applied voltage from a certain discharge configuration, not a real physical current. In the viewpoint of Eq. (5), ({I}_{max}) is a current calculated with an infinitely high ({q}_{i}). When the value of ({q}_{i}) is infinitely high, the way in which the boundary condition of ({q}_{i}) is imposed should not affect the final value of ({I}_{max}).
The validated calculation procedure was applied to the pin-to-plane and single wire-to-plane configurations to calculate ({I}_{max}). The results were compared with ({I}_{sat}) curves. Then, the same methodology was applied to the multiple wire-to-plane configuration to examine the effects of various geometric parameters and applied voltage on ({I}_{max}). The value of (mu) was set to 1.4 (times) 10–4 and 1.6 (times) 10–4 m2 V-1 s-1 for positive and negative discharges, respectively29.
The computational domain for the positive corona discharge in the pin-to-plane configuration was adopted from Adamiak and Atten30 (see Fig. 1b). Figure 4a shows that the values of ({I}_{max}) were approximately the same as those of ({I}_{sat}) in various pin-to-plane configurations. With assuming ({I}_{max}={C}_{2}{V}_{a}^{gamma }) as in Section Maximum current calculation for coaxial cylinders configurations, the values of (gamma) were given by 2.047, 2.089, and 2.090 for ({R}_{t}text{/}L) = 0.0011, 0.0031, and 0.0048, respectively, which were very close to the semi-analytical value, 2 (Eq. (7)). As mentioned in Section Analyticalcurrent voltage curve for coaxial cylinders configuration Based and Maximum current calculation for coaxial cylinders configurations, the semi-analytical expressions of ({I}_{sat}) given in Eqs. (6)–(8) were derived under the small discharge electrode assumption. Therefore, it can be said that the pin electrodes with ({R}_{t}text{/}L) = 0.0011 to 0.0048 used in this study were small enough to neglect the effect of ({R}_{t}) on ({I}_{max}). It is noteworthy that unlike the coaxial cylinders configurations, the Deutsch assumption is basically not valid for the non-symmetric geometry such as the pin-to-plane and wire-to-plane configurations31,32. Jones and Davies31 concluded that the use of the Deutsch assumption to the pin-to-plane configuration should be avoided because it can cause a significant error. Nevertheless, ({I}_{max}) was in good agreement with ({I}_{sat}) for the considered conditions. It is unclear and has not been reported in the literature whether and how the validity of the Deutsch assumption affects ({I}_{sat}). However, the analysis of the theoretical background of ({I}_{sat}) is not in the scope of this study, and thus, will not be further discussed. The change in ({R}_{i}) (1.5 ({R}_{t}), 3 ({R}_{t}), and 5 ({R}_{t})) had little effect on ({I}_{max}) of the pin-to-plane configuration as shown in Supplementary Fig. S2.
({I}_{sat}) and ({I}_{max}) for a the pin-to-plane configuration under ({R}_{i}) = 3 ({R}_{t}), and b the single wire-to-plane configurations under ({R}_{i}) = 3 ({R}_{w}).
The computational domain for the negative corona discharge in the single wire-to-plane configuration was adopted from Ohkubo et al.33 (see Fig. 1c). Figure 4b shows ({I}_{max}) versus ({V}_{a}) under (D) = 100 mm and various ({R}_{w}). The values of (gamma) were greater than 2 for all ({R}_{w}) and increased to greater than 3 as ({R}_{w}) increased. Because corona discharge would be less intensive for thicker wires, ({I}_{max}) is also expected to be smaller for thicker wires. In addition, this result was consistent with the small discharge electrode assumption; the smaller ({R}_{w}) was, the closer ({I}_{max}) was to ({I}_{sat}). As in other configurations, ({R}_{i}) had little effect on ({I}_{max}) of the single wire-to-plane configuration (see Supplementary Fig. S3). It is noted that both ({I}_{sat}) given in Eq. (8) and ({I}_{max}) discussed within the single wire-to-plane configuration are the current values for one plane.
Figure 1d shows the multiple wire-to-plane configuration. (D) and (S) are the wire-to-ground distance and half of the wire-to-wire spacing, respectively. The normalized position along the ground electrode is represented by (xtext{/}S), and (theta) is the angle of incidence. To observe the effects of various combinations of (S) and (D), the following parameter was used:
As mentioned in Section Single wire to plane configuration, ({I}_{sat}=1.62mu varepsilon {V}_{a}^{2}text{/}{D}^{2}) for the single wire-to-plane configuration given in Eq. (8) describes the saturation current per one plane. For the multiple wire-to-plane configuration, this is equivalent to one cell (green shading in Fig. 1d). Assuming that ({I}_{max}) in the multiple wire-to-plane configuration exhibits similar behavior to that of the single wire-to-plane configuration, the expression of ({I}_{max}) for one cell of the multiple wire-to-plane configuration will have the similar form with that for the single wire-to-plane, that is, ({I}_{max,cell}={C}_{3}mu varepsilon {V}_{a}^{gamma }text{/}{D}^{beta }), where ({C}_{3}), (gamma), and (beta) are constants. Then, ({I}_{max}) of the entire multiple wire-to-plane configuration can be expressed as ({I}_{max,multiple}=2{n}_{w}{I}_{max,cell}=2{n}_{w}{C}_{3}mu varepsilon {V}_{a}^{gamma }text{/}{D}^{beta }), where ({n}_{w}) is the number of wires. In addition, the wire-to-wire spacing (S) would affect the magnitude of ({I}_{max}) by an electrical shielding. Therefore, ({I}_{max,multiple}) will have the following form:
where (gleft(Sright)) is a correlation function that describes the dependence of ({I}_{max}) on (S). The actual simulation was conducted over half of one cell, or a unit (blue shading in Fig. 1d), using a symmetry boundary condition. ({I}_{max}) for one unit of the multiple wire-to-plane configuration (({I}_{max,unit})) was obtained by integrating the current density over (S) (see Fig. 1d). Then, ({I}_{max}) for one cell was calculated by ({I}_{max,cell}=2{I}_{max,unit}). Hereinafter, ({I}_{max}) for the multiple wire-to-plane configuration refers to ({I}_{max,cell}) to directly compare the maximum current with the saturation current. Thus, ({I}_{max}) for the multiple wire-to-plane configuration is expressed as follows:
where ({C}_{4}) is a constant. The ranges of (S), (D), and ({V}_{a}) considered were 50–1000 mm, 50–250 mm, and 45–70 kV, respectively. For simplicity, the dependence of ({I}_{max}) on ({R}_{w}) was not considered, and the value of ({R}_{w}) was fixed at 1.59 mm.
Figure 5a shows ({I}_{max}) calculated for various values of (S) and ({V}_{a}) with (D) = 114.3 mm. When (S) was small so that ({theta }_{Stext{/}D}) was less than approximately 65°, ({I}_{max}) increased with increasing (S). However, as (S) increased further so that ({theta }_{Stext{/}D}) became greater than 65°, ({I}_{max}) converged to ({I}_{sat}) of the single wire-to-plane configuration. When ({theta }_{Stext{/}D}) > 65°, two adjacent wires were too far apart to affect each other. They then behaved as two independent single wire-to-plane configurations. For (D) = 114.3 mm, (S) = 250, 500, and 1000 mm gave ({theta }_{Stext{/}D}) = 65.4, 77.1, and 83.5°, respectively. In other words, for (D) = 114.3 mm, the multiple wire-to-plane configurations with (S) = 250, 500, and 1000 mm were equivalent to simple series of ({n}_{w}) single wire-to-plane configurations. Figure 5a also shows ({I}_{sat}) of the single wire-to-plane configuration for (D) = 114.3 mm. For ({V}_{a}) = 70 kV, ({I}_{max}) for the multiple wire-to-plane configuration converged closely to ({I}_{sat}) for the single wire-to-plane configuration as (S) increased. On the other hand, for lower ({V}_{a}), the converged value of ({I}_{max}) was smaller than ({I}_{sat}). This is consistent with the results for the single wire-to-plane configuration with thick wires demonstrated in Fig. 4b.
Simulation results of the multiple wire-to-plane configuration: a ({I}_{max}) for various (S) and ({I}_{sat}) for the single wire-to-plane configuration under (D) = 114.3 mm and various ({V}_{a}), and b normalized current density distributions for ({V}_{a}) = 60 kV and (D) = 114.3 mm under various (S).
The magnitude of the corona current density ((j)) on the ground electrode is known to follow the Warburg distribution, which is given as follows8:
where (theta ={mathrm{tan}}^{-1}left(xtext{/}Dright)) is the angle of incidence (see Fig. 1d), and (m) is an empirical constant. Kim et al.17 stated that for the single wire-to-plane configuration, the distributions of (j) corresponding to ({I}_{exp}) and ({I}_{num}), namely ({j}_{exp}) and ({j}_{num}), followed the Warburg distribution with (m) = 4.2. Figure 5b shows that for (S) larger than 250 mm, the distribution of (j) corresponding to ({I}_{max}) (that is, ({j}_{max})) also followed the Warburg distribution with (m) = 4.2. Figure 5b also shows that for (S) larger than 250 mm, ({j}_{max}) dropped to zero before reaching the midpoint between two wires ((xtext{/}S) = 1). In this situation, the two wires did not influence each other, and each cell in the multiple wire-to-plane configuration converged to the single wire-to-plane configuration, as shown in Fig. 5a. For the single wire-to-plane configuration, ions generated from a discharge electrode do not drift outside (theta) > 65°; hence, (j) drops to zero at approximately (theta) = 65°8,32. For (S) less than 250 mm, however, ({j}_{max}) did not fall to zero before reaching (xtext{/}S) = 1 because of the electrical shielding by the neighboring wires34. Therefore, in this range, ({I}_{max}) was smaller by the undrawn area of the ({j}_{max}) distribution, resulting in a decrease in ({I}_{max}) with decreasing (S), as shown in Fig. 5a. In addition, for (S) less than 250 mm, ({j}_{max}) itself deviated from the Warburg distribution. When (S) = 50 mm, ({j}_{max}) at (xtext{/}S) = 1 was approximately 33% smaller than its value in the Warburg distribution. This further reduced the value of ({I}_{max}).
Figure 6 shows the values of ({I}_{max}) normalized by the value of ({I}_{max}) at (S) = 1000 mm for various ({V}_{a}). Note that the values of ({I}_{max}) calculated for (S) = 1000 mm approximate the values of ({I}_{max}) of the single wire-to-plane configuration for the considered range of (D) (50–250 mm). Figure 5a showed that the magnitude of ({I}_{max}) increased with increasing ({V}_{a}). However, Fig. 6 revealed that the normalized magnitude of ({I}_{max}) did not depend on ({V}_{a}); ({I}_{max}) for different ({V}_{a}) collapsed into a single profile. For ({theta }_{Stext{/}D}) < 45°, the normalized ({I}_{max}) was linearly proportional to ({theta }_{Stext{/}D}). Regression analysis in the linear region (Regime I) yielded ({I}_{max}text{/}{I}_{max}left(S=1000right)=1.18{theta }_{Stext{/}D}-0.065), where ({theta }_{Stext{/}D}) is in radians (solid line in Fig. 6). For ({theta }_{Stext{/}D}) > 65° (Regime III), ({I}_{max}) was the same as that in the single wire-to-plane configuration, as also shown in Figs. 5a and b. Then, replacing ({I}_{max}left(S=1000right)) using Eq. (15), the expression for ({I}_{max}) is given by
where ({theta }_{Stext{/}D}) is in radians in the regression equation. Regime II (45° < ({theta }_{Stext{/}D}) < 65°) represents a transition between Regimes I and III. Using regression, the three regimes can also be represented in a continuous manner as follows (dashed curve in Fig. 6):
where ({theta }_{Stext{/}D}) is in radians.
Normalized ({I}_{max}) for various ({theta }_{Stext{/}D}) and ({V}_{a}). Regime I: multiple wire, II: transition, and III: single-wire-behaved (red: (D) = 50 mm, yellow: 80, green: 114.3, blue: 150, purple: 200, black: 250; ○: ({V}_{a}) = 45 kV, □: 50, ◇: 55, △: 60, (times): 65, + : 70).
Figure 7 shows the dependence of ({I}_{max}) on (D) for various values of (S) and ({V}_{a}). Taking the logarithm with respect to (D) of both sides of Eq. (15) gave the slope of the least-squares fit line as (beta). For a given (S), (beta) was approximately independent of ({V}_{a}). The average values of (beta) were 3.06, 2.76, 2.66, 2.54, 2.40, 2.34, and 2.24 for (S) = 50, 80, 114.3, 150, 200, 250, and 500 mm, respectively. That is, the dependence of ({I}_{max}) on (D) was affected by the value of (S); the value of (beta) decreased and converged to that of the single wire-to-plane configuration as (S) increased.
({I}_{max}) for various (D), (S), and ({V}_{a}).
Taking the logarithm with respect to ({V}_{a}) of both sides of Eq. (15) gave the slope of the least-squares fit line as (gamma). Figure 8 shows a plot of (gamma) versus ({theta }_{Stext{/}D}) for given (D). Note that for each value of (D), an increase in ({theta }_{Stext{/}D}) corresponds to an increase in (S). Overall, the value of (gamma) ranged in approximately 2.5–3.0 and decreased slightly with increasing ({theta }_{Stext{/}D}) or (S).
(gamma) for various ({theta }_{Stext{/}D}) and (D).
In this study, a numerical methodology to evaluate the unipolar saturation current of DC corona discharges was proposed using the concept of maximum current. It was shown that the maximum current has the same mathematical definition as the unipolar saturation current. For validation, the maximum current of the unipolar corona discharge for the coaxial cylinders configuration was compared with the analytical solution of the Poisson equation. For the coaxial cylinders and single wire-to-plane configurations, the difference between the maximum current and the unipolar saturation current decreased as the radius of the wire electrode decreased, confirming the effect of small discharge electrode assumption. For the pin-to-plane configuration, the maximum current was approximately the same as the unipolar saturation current. The calculation procedure was applied to a multiple wire-to-plane configuration, for which a semi-analytical expression of the saturation current has not yet been reported. The results showed that the maximum current was influenced by the electrical shielding from the neighboring discharge electrodes. The effects of geometric and operation parameters on the maximum currents of the multiple wire-to-plane configuration were examined, showing a potential usage for complex discharge configurations.
The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.
Giao, T. N. & Jordan, J. B. Modes of corona discharges in air. IEEE Trans. Power App. Syst. 87, 1207–1215 (1968).
ADS Article Google Scholar
Akishev, Y. et al. The influence of electrode geometry and gas flow on corona-to-glow and glow-to-spark threshold currents in air. J. Phys. D Appl. Phys. 34, 2875 (2001).
ADS CAS Article Google Scholar
Akishev, Y. et al. Negative corona, glow and spark discharges in ambient air and transitions between them. Plasma Sources Sci. Technol. 14, S18 (2005).
Article Google Scholar
Callebaut, T., Kochetov, I., Akishev, Y., Napartovich, A. & Leys, C. Numerical simulation and experimental study of the corona and glow regime of a negative pin-to-plate discharge in flowing ambient air. Plasma Sources Sci. Technol. 13, 245–250 (2004).
ADS Article Google Scholar
Liu, L. & Becerra, M. On the transition from stable positive glow corona to streamers. J. Phys. D Appl. Phys. 49, 225202 (2016).
ADS Article Google Scholar
Khun, J., Scholtz, V., Hozák, P., Fitl, P. & Julák, J. Various DC-driven point-to-plain discharges as non-thermal plasma sources and their bactericidal effects. Plasma Sources Sci. Technol. 27, 065002 (2018).
ADS Article Google Scholar
Goldman, M., Goldman, A. & Sigmond, R. S. The corona discharge, its properties and specific uses. Pure Appl. Chem. 57, 1353–1362 (1985).
CAS Article Google Scholar
Sigmond, R. S. Simple approximate treatment of unipolar space-charge-dominated coronas: The Warburg law and the saturation current. J. Appl. Phys. 53, 891–898 (1982).
ADS CAS Article Google Scholar
Dinh, T. X., Terebessy, T. & Bui, T. T. Bipolar corona discharge based air flow generation with low net charge. Sens. Actuators A Phys. 244, 146–155 (2016).
Article Google Scholar
Ni, M. et al. Development of back corona discharge in a wire-cylinder electrostatic precipitator at high temperatures. Powder Technol. 286, 789–797 (2015).
CAS Article Google Scholar
Kim, C. & Hwang, J. The effect of charge mode transition on electrohydrodynamic flow in a multistage negative air corona discharge. J. Phys. D Appl. Phys. 45, 465204 (2012).
ADS CAS Article Google Scholar
Wang, L. et al. Corona-enabled electrostatic printing for ultra-fast manufacturing of binder-free multifunctional E-Skins. ACS Appl. Mater. Interfaces 13, 45966–45976 (2021).
CAS Article Google Scholar
Moronis, A. X., Simou, N., Kiousis, K. N. & Fylladitakis, E. D. A model for determining the unipolar ionic saturation current in parallel wire-cylinder electrodes during corona discharge. IEEE Trans. Dielectr. Electr. Insul. 21, 1035–1043 (2014).
Article Google Scholar
Fylladitakis, E. D., Moronis, A. X. & Theodoridis, M. P. A mathematical model for determining an electrohydrodynamic accelerator’s monopolar flow limit during positive corona discharge. IEEE T. Plasma Sci. 45, 432–440 (2017).
ADS Article Google Scholar
Sigmond, R. S. The unipolar corona space charge flow problem. J. Electrostat. 18, 249–272 (1986).
Article Google Scholar
Fylladitakis, E. D., Moronis, A. X. & Theodoridis, M. Analytical model for determination of the unipolar ionic saturation current during positive corona discharge for geometries comprising cylindrical emitter electrodes. IEEE Trans. Dielectr. Electr. Insul. 24, 852–860 (2017).
Article Google Scholar
Kim, C., Noh, K. C. & Hwang, J. Numerical investigation of corona plasma region in negative wire-to-duct corona discharge. Aerosol Air Qual. Res. 10, 446–455 (2010).
Article Google Scholar
Komuro, A., Ono, R. & Oda, T. Behaviour of OH radicals in an atmospheric-pressure streamer discharge studied by two-dimensional numerical simulation. J. Phys. D Appl. Phys. 46, 175206 (2013).
ADS Article Google Scholar
Feng, J. Q. An analysis of corona currents between two concentric cylindrical electrodes. J. Electrostat. 46, 37–48 (1999).
Article Google Scholar
Townsend, J. S. X. I. The potentials required to maintain currents between coaxial cylinders. Philos. Mag. 28, 83–90 (1914).
Article Google Scholar
Thomson, J. J. & Thomson, G. P. Conduction of Electricity through Gases (Cambridge University Press, 1933).
Yanallah, K., Pontiga, F., Meslem, Y. & Castellanos, A. An analytical approach to wire-to-cylinder corona discharge. J. Electrostat. 70, 374–383 (2012).
CAS Article Google Scholar
Monrolin, N., Praud, O. & Plouraboué, F. Revisiting the positive DC corona discharge theory: Beyond Peek’s and Townsend’s law. Phys. Plasmas 25, 063503 (2018).
ADS Article Google Scholar
Jones, J. E., Dupuy, J., Schreiber, G. O. S. & Waters, R. T. Boundary conditions for the positive direct-current corona in a coaxial system. J. Phys. D Appl. Phys. 21, 322 (1988).
ADS CAS Article Google Scholar
Waters, R. T. & Stark, W. B. Characteristics of the stabilized glow discharge in air. J. Phys. D Appl. Phys. 8, 416 (1975).
ADS Article Google Scholar
McDonald, J. R., Smith, W. B., Spencer, H. W. III. & Sparks, L. E. A mathematical model for calculating electrical conditions in wire-duct electrostatic precipitation devices. J. Appl. Phys. 48, 2231–2243 (1977).
ADS Article Google Scholar
Lowke, J. J. & Morrow, R. Theory of electric corona including the role of plasma chemistry. Pure Appl. Chem. 66, 1287–1294 (1994).
CAS Article Google Scholar
Chen, J. & Davidson, J. H. Electron density and energy distributions in the positive DC corona: Interpretation for corona-enhanced chemical reactions. Plasma Chem. Plasma Process. 22, 199–224 (2002).
CAS Article Google Scholar
Hinds, W. C. Aerosol technology: Properties, behavior, and measurement of airborne particles (Wiley, 1999).
Adamiak, K. & Atten, P. Simulation of corona discharge in point–plane configuration. J. Electrostat. 61, 85–98 (2004).
Article Google Scholar
Jones, J. E. & Davies, M. A critique of the Deutsch assumption. J. Phys. D Appl. Phys. 25, 1749 (1992).
ADS Article Google Scholar
Oussalah, N. & Zebboudj, Y. Finite-element analysis of positive and negative corona discharge in wire-to-plane system. Eur. Phys. J. Appl. Phys. 34, 215–223 (2006).
ADS Article Google Scholar
Ohkubo, T., Nomoto, Y., Adachi, T. & McLean, K. J. Electric field in a wire-duct type electrostatic precipitator. J. Electrostat. 18, 289–303 (1986).
Article Google Scholar
Arif, S., Branken, D. J., Everson, R. C., Neomagus, H. W. J. P. & Arif, A. The influence of design parameters on the occurrence of shielding in multi-electrode ESPs and its effect on performance. J. Electrostat. 93, 17–30 (2018).
Article Google Scholar
Download references
This work was supported by a grant from the Subway Fine Dust Reduction Technology Development Project of the Ministry of Land Infrastructure and Transport of the Republic of Korea (21QPPW-B152306-03).
School of Mechanical Engineering, Yonsei University, 134 Sinchon-dong, Seodaemun-gu, Seoul, 03722, Republic of Korea
Sangwoo Kim & Jungho Hwang
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
S.K. conceptualized and designed the study, performed simulations and validation, analyzed the data, prepared figures, and wrote the manuscript. J.H. supervised the research project, analyzed the data, reviewed the formulation for regression analysis, and wrote the manuscript. All authors reviewed and approved the final version of the manuscript.
Correspondence to Jungho Hwang.
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Reprints and Permissions
Kim, S., Hwang, J. Numerical methodology to evaluate unipolar saturation current limit of DC corona discharge in complex geometries. Sci Rep 12, 14252 (2022). https://doi.org/10.1038/s41598-022-18144-5
Download citation
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-022-18144-5
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
Advertisement
Advanced search
© 2022 Springer Nature Limited
Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.
VSP special agents testify in deadly Rockbridge Co. gas station explosion trial – WFXRtv.com
WFXRtv
Please enter a search term.
Please enter a search term.
by: Kelsey Jean-Baptiste
Posted:
Updated:
by: Kelsey Jean-Baptiste
Posted:
Updated:
ROCKBRIDGE COUNTY, Va. (WFXR) — The trial for a Roanoke man charged with four counts of manslaughter in connection with a deadly gas station explosion in Rockbridge County in 2019 entered its fourth day on Thursday.
On May 10, 2019, an explosion at the South River Market killed four people, including the owner of the store, Roger Lee Roberts; two employees, Kevin Tate Roberts and Samantha Gail Lewis; and a patron, Paul Dewayne Ruley.
Phillip Ray Westmoreland, the driver of the truck that delivered the fuel to the gas station, was arrested on March 19, 2021 and charged with four counts of involuntary manslaughter, according to Rockbridge County court records.
Seventeen months after Westmoreland’s arrest — on Thursday, Aug. 18 — the fourth day of his trial got underway, even though Judge Christopher Russel informed the court that one juror had been dismissed due to family issues and was replaced with the alternate juror.
On Thursday morning, Virginia State Police Senior Special Agent Mark Austin went back on the stand to share more facts about his investigation into the explosion, which lasted until November 2019.
The prosecution questioned Austin about his understanding of Westmoreland’s qualifications to be a tanker driver.
Austin says tankers are supposed to have a class A license for combination vehicles — including tankers that weigh 26,000 pounds or more. He adds drivers like Westmoreland would need a special endorsement to drive those vehicles or a tanker.
In Austin’s investigation, he also obtained documents of multiple deliveries made to the South River Market, the amount of fuel that was picked up, the miles driven, the capacity size of the tanks, and more.
During the trial, some of those delivery tickets were noted to the jury as:
Austin also explained that there were multiple safety inspections of Westmoreland’s vehicle. In that report, no issues came up.
“There are minor defects that would not have caused a malfunction of the distribution of gasoline or the tanker,” said Mary E. Lester, Chief Deputy Commonwealth’s Attorney for Rockbridge County and the City of Lexington, as she read out a stipulation of the safety inspection.
In the defense’s cross-examination, Westmoreland’s attorney, Robert Dean, questioned Austin about his investigation, focusing on the “flag, bag, and tag” technique in order to show that other investigators at the scene handled that rather than Austin.
In order to demonstrate that a thorough investigation was not conducted, Dean also questioned Austin about dumpsters that may have been used after the explosion. Defense claimed the dumpsters were dropped off at different locations and that Austin never looked inside of them to see if any important evidence was thrown out.
After Austin left the stand, the prosecution brought Virginia State Police Special Agent Shane Clifton — who interviewed Westmoreland after the explosion — to the stand.
Clifton told the court that Westmoreland had years of experience driving trucks and proceeded to discuss Westmoreland’s routine every time he delivered gasoline.
“He doesn’t question the amount on the order. He just picks up the order and delivers,” said Clifton.
According to Clifton, Westmoreland made seven to 10 deliveries at the South River Market.
Next up, they discussed how Westmoreland measured the tanks before he began pumping fuel into them.
“He remembered the measurement for the conventional tank was 65 inches,” said Clifton.
It was noted that Westmoreland never suggested there was something wrong with the tanks.
After a brief lunch, everyone entered the room to listen to an hour-and-45-minute audio recording of Clifton’s interview with Westmoreland about the morning before Westmoreland delivered the fuel.
Once fully loaded, he told investigators about how he filled the tanks at the South River Market.
Westmoreland was also questioned on various numbers regarding the tank size, how much he needed to fill them up with gas, the amount of space, etc.
“Just tell them you need my driver sheet and anything they need from that store,” said Westmoreland in the recording.
According to Westmoreland, it’s hard for him to remember numbers, so he writes down everything in his driving delivery book.
A few minutes into the interview, Westmoreland willingly took a breathalyzer test. In addition, he was asked if he ever used drugs, drank alcohol, used his cellphone, or did anything that would have distracted him.
Meanwhile, the prosecution is trying to prove that Westmoreland allegedly never looked at the conversion charts, which would have given him the right measurement to put the gasoline in the tanks.
At one point in the interview, Westmoreland discussed a vapor recovery system, which catches vapors from the tank that is refueled.
He also admitted that the South River Market’s system was so old that he didn’t have that system, making the vapors go straight up into the air.
During the the fourth day of the trial, a huge portion of the interview focused on the measurements of the tank, the conversion unit used when delivering, and whether or not Westmoreland was distracted during the delivery.
Thursday’s proceedings ended after the audio recording concluded. The trial is set to resume at 9 a.m. on Friday, Aug. 19.
Copyright 2022 Nexstar Media Inc. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.
Thanks for signing up!
Watch for us in your inbox.
Subscribe Now
UFC 278 Usman vs. Edwards 2: Leon Edwards scores last minute knockout to win welterweight title – ESPN
The Salt Lake City crowd and the announcers can’t believe what they’re seeing after Kamaru Usman is knocked out cold by Leon Edwards. (2:02)
SALT LAKE CITY – Leon Edwards has shocked the MMA world. With a head kick in the final round of a fight he was handily losing, Edwards knocked out Usman to win the UFC welterweight title in the main event of UFC 278 on Saturday night here at Vivint Arena. The finish came at 4:04 of the fifth round. Usman was considered the pound-for-pound best fighter in the world.
Edwards snapped Usman’s 15-fight UFC winning streak, which was a UFC record. It was one of the most stunning knockouts in the history of the sport.
“I can’t put it into words,” Edwards said.
Usman had been dominating the fight after the first round. But Usman leaned down to his right to defend an Edwards left and Edwards caught him flush with a left head kick, sending the more than 17,000 in attendance into a frenzy. Usman, who was out on contact, was less than a minute away from a sixth title defense.
Full recap from Marc Raimondi
Round-by-Round coverage:
Round 1: Edwards lands a few leg kicks, then defends a takedown attempt. He then slips on an attack, and Usman gets him to the canvas. But Edwards gets up, and with two minutes to go, he scores a trip takedown into full mount. Edwards ends up taking the back and threatening a choke until the horn. 10-9, Edwards.
Round 2: Edwards gets his foot caught against the cage, and the brief distraction allows Usman to swarm with punches. Even after Edwards gets away, Usman keeps up the pressure, leading to two takedowns. Edwards is starting to look just a bit worn down. 10-9, Usman. (19-19)
Round 3: Usman resumes his stalking. A minute in, he gets a takedown. Edwards gets right back up, but Usman is all over him. Edwards lands a couple of shots, but Usman again closes in, gets another takedown and takes back control. Then he is on top, in full mount. Edwards looks drained, and Usman is rolling. 10-9, Usman. (29-28 Usman.)
Round 4: Edwards stands his ground, lands a few strikes and is close to a takedown. But Usman defends well, turns the tables and midway through the round gets a takedown of his own. Usman is in full control, all over Edwards. There’s a pause as referee Herb Dean sees Edwards grab the cage to get up from a takedown. Dean puts them back on the canvas, and even after Edwards gets back up, he is sagging. 10-9, Usman. (39-37 Usman.)
Round 5: Usman keeps stalking, and Edwards is not threatening. The challenger needs to do something big, and he has nothing for Usman, who is cruising. Edwards lands a couple of shots, but one is low. Then, just as it looked like Edwards was out of it, his left high kick knocks out Usman.
Paulo Costa handily wins his fight vs. Luke Rockhold via unanimous decision in the co-main event of UFC 278.
Paulo Costa and Luke Rockhold went into Saturday not only as two of the best middleweights in the world, but also two of the middleweights in most need of a win. In the end, it was Costa who got it done.
Costa (14-2) out-lasted Rockhold (16-6) over the course of a grueling three-round affair at UFC 278 on Saturday, collecting unanimous judges’ scores of 30-27. Both middleweights looked absolutely exhausted in the high altitude of Vivint Arena, particularly Rockhold, who repeatedly put his hands on his hips between rounds and was forced to breathe out of his mouth due to blood pouring from his nose early on.
It was a much-needed win for Costa, who began his career with 13 consecutive wins but was coming off back-to-back losses to Israel Adesanya and Marvin Vettori.
For Rockhold, a former UFC and Strikeforce champion, it appeared to be the end of the road. The 37-year-old returned on Saturday from a three-year layoff, and essentially announced his retirement in the Octagon after. Although he didn’t actually say he was done, he was extremely emotional, removed his gloves inside the cage and nodded in affirmation when UFC commentator Joe Rogan mentioned the possibility of it being his last fight.
“I can’t do this s— anymore,” Rockhold said. “I gave it my all. I’m f—ing old.”
Costa, whose last win was in 2019 against Yoel Romero, got off to a hot start by defending Rockhold’s initial takedown attempts. He landed several hard hammerfists after stuffing one of those early attempts, which immediately caused swelling on Rockhold’s face and bloodied his nose. He maintained a steady pressure on Rockhold on the feet, constantly walking him to the fence and throwing combinations.
By the end of the first round, Rockhold’s body language was terrible. He leaned heavily on the fence between the first and second rounds, and repeatedly lowered his hands and looked for rests in the middle of the action. He actually got a rest in the second round when Costa landed an accidental punch below the belt. Even after the break, however, Rockhold continued to breathe heavily.
This fight is madness #UFC278
According to UFC Stats, Costa landed 106 total strikes to 57 for Rockhold.
Rockhold showed heart and will power in the third, when he continued to throw dangerous shots even though he was clearly exhausted. At the end of the fight, he reversed position on Costa on the floor and finished the round on top. The fight ended with Rockhold rubbing his face on Costa, so the blood coming from his nose smeared over the Brazilian’s face.
Despite the two losses coming in, Costa is still considered a top 5 middleweight, although he has expressed an interest in boxing in the near future.
Dvalishvili had landed five or more takedowns in all but two of his previous nine UFC fights. He had hit double figures in three of those bouts. His nickname is “The Machine,” and it was obvious why that is.
Then he stepped in with Aldo and went 0 for 16 on takedown attempts. Was it Dvalishvili’s undoing? No, it was not. The 31-year-old from the Republic of Georgia couldn’t get the Brazilian legend to the canvas, but his relentless attempts against the cage, though foiled one after another, produced a nonstop attack of knees to the leg that, combined with the effort needed to stop takedowns, drained Aldo and led to Dvalishvili’s seventh straight victory (29-28, 29-28, 30-27).
The win streak ties Dvalishvili for the best in the bantamweight division with the champion, but no title showdown is imminent. The champ, Aljamain Sterling, is Dvalishvili’s training partner on Long Island, New York, and Dvalishvili has no intention of fighting him. “He is my brother,” Dvalishvili said.
He also paid homage to Aldo, the 35-year-old former featherweight champ from Brazil. Aldo received a standing ovation during his walkout and the night’s loudest applause during his introduction. He saw a three-fight winning streak come to an end. But along the way, he earned Dvalishvili’s respect.
“He was very technical against the cage,” Dvalishvili said. “Usually, I take down everybody.”
Pudilova was a brawler in her first go-around with the UFC. This time, it’s clear she has improved and is far more versatile.
Using wrestling and violent ground and pound, Pudilova picked up a TKO victory at 4:04 of the second round over Wu Yanan. Pudilova, who is now training at SBG Ireland under coach John Kavanagh, took Wu down, got her back, moved to mount and along the way was hammering down with big punches and elbows. Referee Herb Dean had no choice but to step in and stop the fight.
“I can’t believe I’m back in the UFC,” Pudilova said.
Pudilova also got a takedown in the first round and had a lot of success on the ground. Wu had some decent leg kicks on the feet, but Pudilova got an arm-and-head throw in the second round, got Wu’s back and started firing off violent elbows from top position, leading to the finish. Wu had no answers.
Aye now that’s some violence
Pudilova, 28, went 5-1 with the Oktagon promotion after being released by the UFC in 2020. The Czech Republic native lost four straight before being cut, but two of those were $50,000 Fight of the Night performances.
Wu, a 26-year-old from China fighting out of Albuquerque, New Mexico, has dropped four straight.
It wasn’t all that long ago, Pedro was considered a very promising prospect at 205 pounds. That sentiment has now returned.
Pedro, 30, lost a lot of momentum between 2017 and 2022, when injuries limited him to just six total appearances, in which he held a middling 3-3 record. Since returning to health this year, the Australian has recorded back-to-back finishes against Ike Villanueva and now Hunsucker.
The finish on Saturday came just 65 seconds into the fight. He caught Hunsucker with a clean, counter left jab. Immediately after landing the shot, Pedro smiled, pointed at Hunsucker and moved aggressively forward. He dropped him moments later with a front kick to the midsection and finished the fight with punches on the ground.
Beautiful work @tyson_pedro_ 👏#UFC278
“I trained so hard, it was a very hard camp,” Pedro said. “To get it done that quick, that’s what it’s all about. I haven’t fought in front of fans for four years, so do me just one favor. Let me hear some noise.”
A step up in competition will likely be next for Pedro, as he was an 8-to-1 betting favorite over Hunsucker. He improves to 5-3 overall in the UFC. His losses came against experienced opposition in Ilir Latifi, Ovince St-Preux and Mauricio ‘Shogun’ Rua.
If you watched just the first round of this fight, you would find it inconceivable that Romanov would exit the Octagon no longer undefeated. He took down Tybura less than a minute in and mauled him on the canvas for the entirety of the first five minutes, building a 28-0 advantage in strikes.
But Tybura survived to the horn, and when he came out for Round 2, he took over against a Romanov who appeared to have no energy left. As a result, Tybura won the second and third rounds and was rewarded with a majority decision, with two judges scoring the bout 29-28 in his favor and the third scorecard reading 28-28 because of a 10-8 first round for the 16-1 Romanov.
Tybura, a 36-year-old from Poland, has won six of his last seven fights after persevering through a rough first round, putting his opponent in trouble with a takedown in the second, then peppering him with punches in a slow-motion third round. His experience against higher-level opposition — he’s been in with former UFC champs Fabricio Werdum and Andrei Arlovski, among other top heavyweights — enabled Tybura to pace himself and remain fresh in the 4,300-foot altitude of Salt Lake City.
Romanov, who is 31 and from Moldova, was a force in the first round but sagged after that. He’d been past the midpoint of Round 3 only three times in his career and had scored finishes in 15 of his 16 wins. This time he had nothing in reserve once his gas tank emptied.
Salt Lake City’s elevation is 4,226 feet, making cardio a major storyline at UFC 278. Gordon, obviously knowing that, invested in working the body — and thereby the gas tank — with incredible success.
Gordon completely dominated Santos in a unanimous decision (30-27, 30-27, 30-27) victory, one of the most one-sided of Gordon’s career. Gordon’s cardio held the whole way and he was able to pepper Santos with combinations in every round, including clear and consistent attention paid to body shots.
Santos was bleeding from his nose in the first round and had a cut under his eye by the end of the first. Gordon had several surges of big flurries that did damage. Meanwhile, Santos, a Brazilian jiu-jitsu specialist, was never able to get Gordon to the ground in a dominant position. By the third round, Santos was clearly tired with Gordon still pushing forward and getting in his face.
Gordon, 33, has won four of his last five fights. The New York native, who trains out of Kill Cliff FC in Florida, is now 7-4 in the UFC. Santos, a 42-year-old fighting out of Brazil’s Nova Uniao camp, has dropped three straight.
Luis Saldana celebrates after he thinks he’s won the fight, but referee Mike Beltran deducts a point from him in the first round.
There’s no other way to say it: Saldaña blew multiple chances to cash in an upset victory over Woodson.
Saldaña, who fights out of Phoenix, dropped Woodson with a clean left hook midway through the opening round. Rather than follow him to the floor and look for a finish, Saldaña nonchalantly walked away and allowed Woodson to get up, and didn’t aggressively pursue him even though Woodson looked very unstable.
The real mistake came moments later, however, when Saldaña dropped him again with a straight left. Again, Saldaña elected not to follow him to the floor. As Woodson sat up, Saldaña rocked him with a blatantly illegal knee to the head, which led to a pause in the action and referee Mike Beltran docking Saldaña a point.
The rest of the bout was relatively inactive, with Woodson doing just a little more. He threatened Saldaña with a modified triangle choke in the second round, and worked a steady jab in the third. The result, due in large part to the point deduction, was a split draw. The judges’ scorecards read 29-27 Woodson, 29-27 Saldaña and 28-28.
Draws are garbage it means neither fighter gets their win money and it doesn’t move their career forward so it’s a huge waste of time and expense
Woodson, who fights out of St. Louis, is now 3-1-1 in the UFC, while Saldaña is 2-1-1.
Round 3 played out in slow motion, with Fletcher lying on his back and showing no significant effort to get up, and Loosa maintaining top position but doing little to try to end the fight. But the fans didn’t start booing until there was barely a minute left. How could they have after what these two fighters had delivered in the round right before?
Loosa had spent much of that middle round landing right hand after right hand to the face of Fletcher, who was bloodied and wobbly by the middle of the five minutes. But the high output appeared to tire Loosa, and Fletcher took advantage of that late in the round with a barrage of punches that had his opponent stumbling around the cage, with referee Mike Beltran looming close by in case he needed to jump in.
As the fans roared the horn sounded, both fighters had little to give from that point on. But merely by getting the fight back to the canvas at the start of the third round and keeping it there, Loosa did enough to secure the win (29-27, 29-28, 29-28).
Yooooo this fight is insane !!! #UFC278
Loosa is 28 and from Switzerland, but trains in South Florida. He won for the first time in the Octagon after dropping his UFC debut in April.
Fletcher, a 25-year-old from Lafayette, Louisiana, has lost both of his UFC fights after coming to the promotion earlier this year undefeated.
Amir Albazi celebrates after his submission win vs. Francisco Figueiredo at UFC 278.
Albazi is one of the top prospects to watch at 125 pounds and he did nothing Saturday to change that perception.
Using his wrestling and grappling, Albazi dominated Figueiredo en route to a submission (rear-naked choke) win at 4:34 of the first round. Albazi landed two takedowns and on the second one he worked to Figueiredo’s back slickly and locked in the choke.
“There’s a bunch of killers at flyweight and I’m the best of them,” Albazi said. Albazi, 28, had not fought since January 2021, a unanimous decision victory over Zhalgas Zhumagulov. The Iraq-born England resident has won four straight, including his first three in the UFC.
Amir is a prrrroblem @AmirAlbazi
Figueiredo, the 32-year-old brother of flyweight champion Deiveson fighting out of Brazil, is 2-2 in the UFC.
“The king is back,” Albazi said.
‘The Mongolian Murderer’ held off a late rally by Perrin en route to unanimous 29-28 judges’ scorecards. The fight was very competitive and the Salt Lake City crowd booed the decision, but Perrin himself seemed to accept it and applauded Aoriqileng.
Aoriqileng’s counter striking stood out early in the fight, as he wobbled Perrin with a left hook and regularly caught him with shots as he looked to close distance. Perrin’s wrestling became a factor in the second and third rounds, however.
Some of the best action of the bout took place in the final 10 seconds, when Perrin walked Aoriqileng back to the fence and unloaded combinations. Afterwards, Aoriqileng said he wasn’t hurt by the shots, but it was probably Perrin’s best moment.
Aoriqileng is now 2-2 in the UFC, although he is 2-0 since moving up from flyweight. Perrin, Massachusetts, falls to 0-2.
Two fighters chasing their first UFC victory wasted no time in getting after it. Both came out swinging, and Da Silva landed the first impactful blow, a right hand that dropped Altamirano and had him in trouble in the opening minute. But Altamirano got up and kept coming forward, and after a knee to the body put Da Silva on the canvas, the Mexican swarmed with a barrage of elbows that finished the job at 3:39 of the first round.
Altamirano, a 31-year-old from Mexico City who trains in Fort Worth, Texas, had to withstand the early assault, but once he had the fight going his way, he was relentless in winning for the fifth time in his last six bouts. He lost his UFC debut in February.
Da Silva, who is 26 and from Brazil, looked like the sharper fighter at the start, knocking down his opponent and nearly connecting with spinning attacks. But once he was on his back, he made the mistake of trying for submissions rather than defending against the elbows that opened up a cut on his face and ended the fight. He has lost all three of his UFC bouts.
Which animal has the largest brain relative to its body size? – Livescience.com
Live Science is supported by its audience. When you purchase through links on our site, we may earn an affiliate commission. Here’s why you can trust us.
By published
Smaller animals have larger brains relative to their bodies.
If terms like “bird brain” and “harebrained” are any indication, animal brains present fascinating and diverse examples of one of the most complex organs known to science. Animal brains differ not only in overall size but also in their size relative to the animal’s body mass.
At 18 pounds (8 kilograms), on average, the sperm whale (Physeter macrocephalus) has the largest brain, but has an overall body mass of 45 tons (40 metric tons), giving it a brain-to-body-mass ratio of 1:5,100. But which animal has the biggest brain relative to its body size?
A 2009 study in the journal Brain, Behavior and Evolution (opens in new tab) found that an especially tiny genus of ant has the largest brain for its body size. Brachymyrmex has an average body mass of up to 0.049 milligrams and an average brain mass of 0.006 milligram. That means its brain is roughly 12% of its body mass, giving it a brain-to-body-mass ratio of about 1:8.
Related: Why do our brains have folds?
In absolute terms, animals’ brain size tends to increase with the size of the animal. Bigger brains are typically related to three factors: “maternal investment, complexity of behavior and the sheer size of the body,” said Sophie Scott, a professor of cognitive neuroscience at University College London.
“Having a bigger body means needing to control that more,” Scott told Live Science. “Apex predators tend to be big. And because of a need for more complex behavior, like being able to outwit your prey, they benefit from having a bigger brain.”
But, brain size isn’t a perfect predictor of animal intelligence. The brain of an African elephant (Loxodonta africana) weighs 10 pounds (4.6 kg), on average, according to a 2014 study in the journal Frontiers in Neuroanatomy (opens in new tab), three times larger than the human brain. Their large brain size is due in part to their massive cerebellum, used to coordinate the muscle activity in their trunk and ears, according to Scott.
“If you look at the brain of a rabbit, a cat and a small monkey, they’re not that different in size, but their behavior is going to be very different because of the nature of the brain cells,” Scott said. “By the time you get to the monkey, you’re seeing a primate brain, with proportionally larger frontal lobe areas and more curiosity-driven behavior.”
Scott explained that evolutionary adaptations change the brain’s structure to increase the size of certain areas and favor certain neural connections. In humans, the size of our cerebral cortex and the density of cortical neurons (the number of neurons present there) explain our intelligence more than the size of our brains relative to our bodies. Compared with other animals “we have quite a small body for the size of our brains,” Scott said.
—How many calories can the brain burn by thinking?
—Which animals have the longest arms?
—What animal has the largest ears?
When comparing brains of different species, it’s important to consider brain architecture as well as brain size. Because brain-to-body-mass ratio does not take into account the evolutionary development of the cerebral cortex and the density of neural connections found there, scientists look to encephalization quotient (EQ) as a more accurate measure of animal intelligence. The encephalization quotient is the relative brain size observed in a particular species, compared with the expected brain size of other species of a similar body size. One key factor in EQ is the relative size of the cortex compared with the rest of the brain. Comparing animals based on their EQ provides a more accurate view of their intelligence than brain-to-body-mass-ratio, according to the Encyclopedia of Behavioral Neuroscience (opens in new tab), though not as accurate as measuring the absolute size and interaction of individual brain regions.
Related: Are cats or dogs smarter?
Then, there’s a concept known as Haller’s rule: The bigger the animal is, the smaller the brain-to-body ratio will be. “Because brain size scales relative to body size, the smallest animals have relatively the largest brains,” Wulfila Gronenberg, a professor of neurobiology at the University of Arizona, told Live Science.
For example, ants have relatively small brains compared with other hymenoptera, a class that includes bees, wasps, hornets and sawflies. “We think this is because … ant workers don’t fly,” Gronenberg said. Flying requires a lot of visual processing, so many flying insects typically have large eyes, thus leading to larger optical lobes. “In some insects, like a dragonfly, visual processing is more than half of their entire brain,” Gronenberg said.
Originally published on Live Science.
Martin McGuigan is an Irish writer based in Norwich, England. His work has appeared in The Mays XIX, Cabinet of the Heed and SHE magazine. His writing explores the bizarre questions of everyday life, the mysteries of human psychology, and environmental issues. He studied English literature at the University of Cambridge and creative writing at the University of East Anglia.
Stay up to date on the latest science news by signing up for our Essentials newsletter.
Thank you for signing up to Live Science. You will receive a verification email shortly.
There was a problem. Please refresh the page and try again.
Live Science is part of Future US Inc, an international media group and leading digital publisher. Visit our corporate site (opens in new tab).
© Future US, Inc. Full 7th Floor, 130 West 42nd Street, New York, NY 10036.
In the box: the ultimate beginner's guide to digital audio – MusicRadar
MusicRadar is supported by its audience. When you purchase through links on our site, we may earn an affiliate commission. Here’s why you can trust us.
By published
Let’s find out exactly what’s going on ‘inside the box’ as we go back to first principles
It’s entirely possible to make music on a computer without having the faintest idea of how digital audio works, but a little knowledge can go a long way when it comes to pushing the medium to its limits.
For those unfamiliar with the basics of how sound works, let’s start at square one. Some of the following explanations might seem a little abstract out of the context of actual mixing, but introducing these ideas now is important, as they’ll be referenced when we get more hands-on later in the issue.
Sound is the oscillation (that is, waves) of pressure through a medium – air, for example. When these waves hit our tympanic membranes (eardrums), the air vibration is converted to a vibration in the fluid that fills the channels of the inner ear. These channels also host cells with microscopic ‘hairs’ (called stereocilia) that release chemical neurotransmitters when pushed hard enough by the vibrations in the fluid. It’s these neurotransmitters that tell our brain what we’re listening to.
The process of recording audio works by converting the pressure waves in the air into an electrical signal. For example, when we (used to) record using a microphone feeding into a tape recorder, the transducer in the mic converts the pressure oscillation in the air into an electrical signal, which is fed to the tape head, which polarises the magnetic particles on the tape running over it in direct proportion to the signal.
The movement of sound through air – and, indeed, the signal recorded to tape – is what we’d call an analogue signal. That means that it’s continuous – it moves smoothly from one ‘value’ to the next without ‘stepping’, even under the most microscopic of scrutiny.
The answer is a method called pulse code modulation (PCM), which is the main system employed when working with digital audio information. It starts with the process of analogue-to-digital conversion (ADC), which involves measuring the value of the continuous signal at regular intervals and creating a facsimile of the original waveform. The higher the frequency at which the signal is referenced – or ‘sampled’ – and the greater the precision of the value that’s recorded, the closer to the original waveform the digital recording will be (see Fig. 1).
The number of samples taken per second is called the sampling rate. CD-quality audio is at a sampling rate of 44.1kHz, meaning that the audio signal is sampled a whopping 44,100 times per second. This gives us a pretty smooth representation of even high frequencies – the higher the frequency we want to measure accurately, the higher the sampling rate required.
The theorem states that the sampling rate needs to be double the highest frequency, lest what’s known as aliasing occurs (see Fig. 2). Aliasing causes unmusical artefacts in the sound, as frequencies higher than the sample rate will be “reflected” around the Nyquist frequency, which is half the sample rate. So at 44.1kHz, Nyquist is 22,050Hz, and if we try to sample, say, 25,050Hz (22,050Hz plus 3000Hz), we’ll instead get 19,050Hz (22,050Hz minus 3000Hz). As frequencies above 20kHz or so can’t be perceived by the human ear, they’re not massively useful anyway.
In ADC, what’s known as an anti-aliasing filter is used to band-limit the sound, stopping any frequency higher than half the sampling rate from getting through. So, that 44.1kHz rate gives us double our 20kHz, plus a bit of extra room for the low-pass filter to filter out the unwanted high-end signal. 44.1kHz was the de facto digital playback standard through the CD era, although much higher sample rates are now common.
Once our sample has been taken, it needs to be stored as a number. You probably already know that CD quality audio is 16-bit – but what, exactly, is a bit? A bit (short for binary digit) is the most fundamental unit of information in computing, and can have one of two values – usually represented as 1 or 0. If we add another bit, we double the number of available states: for two bits, the number of available states would be four (two to the power of two) – that is, 00, 01, 10 and 11.
Four bits would allow for 16 states (two to the power of four), eight bits would allow for 256 states (two to the power of eight), and 16 bits a massive 65,536 possible states (two to the power of 16). So, we’ve gone from a primitive ‘on or off’ state at 1-bit to having quite a fine resolution at 16-bit. See Fig. 3 for a visual representation of this.
When an analogue-to-digital converter samples an analogue signal and rounds it up or down to the nearest available level, as dictated by the number of available bits, we get ‘quantisation error’ or ‘quantisation distortion’. This is the difference between the original analogue signal and its digital facsimile, and can be considered an additional stochastic signal known as ‘quantisation noise’. A stochastic signal is non-deterministic, exhibiting different behaviour every time it’s observed – a random signal, in other words.
So, in a digital recording, we get our ‘noise floor’ (the smallest measurement we can take with any certainty) from the number of bits we’re using. This is known as ‘word size’ or ‘word length’. For every bit we add to the word length, we get twice the number of possible states for each sample, doubling the dynamic range with another 6dB or so of headroom over the noise floor.
For example, a 16-bit digital recording at CD quality gives us a hypothetical 96dB of dynamic range. It might surprise you to learn that this is actually higher than the dynamic range of vinyl, which is more like 80dB. If you’re unfamiliar with decibels and dynamic range, don’t worry – we’ll come back to them shortly.
If you’re new to music production and this all seems very abstract, perhaps a practical example will bring things into focus. Install the free audio editor Audacity, launch it and set the input to your computer’s built-in microphone. Click the record button, and repeatedly speak the word “test” into the microphone, quietly at first, then getting louder each time. You should see something like the image below: a series of similar waveforms that increase in size on the vertical axis.
However, the greater the bit depth, the more accurate the waveform, and this is particularly important when dealing with signals that contain both loud and quiet elements – percussive sounds, for example. This is because more bits give us more dynamic range.
Dynamic range is the ratio of the largest and smallest possible values of a variable such as amplitude, and it can be expressed in terms of decibels (dB), which we’ll get into in more detail later. In simple terms, it’s the difference in dB between the loudest and softest parts of the signal.
Dynamic range is the ratio of the largest and smallest possible values of a variable such as amplitude
Here’s an example of dynamic range in action that most of us will be familiar with: you’re watching a movie late at night, trying to keep the volume at a level that isn’t going to disturb anyone. The trouble is, the scenes in which the characters talk to each other are too quiet, while the scenes with guns and explosions are too loud! Wouldn’t it be better if the movie’s soundtrack was at a consistent volume level?
Well, for late-night home viewing, maybe; but in a cinema, it’s the difference in amplitude between the characters’ dialogue and the explosions that gives the latter their impact. As such, having a broad dynamic range is important in film soundtracks, as well as classical music, where you’ll get much quieter sections than you ever would in a dance track. For example, take a listen to the fourth movement of Dvořák’s New World Symphony below, which features very delicate, restrained sections and intense crescendos. Dynamic range is important in club music too, though, because it helps keep the percussive sounds punchy.
The importance of dynamic range becomes very apparent when compressing drum sounds. For the uninitiated, compression is a process whereby the dynamic range of a signal is reduced to get a sound with a louder average level. Once the level of the signal exceeds a certain volume, the amount it goes over that threshold is reduced by a ratio set by the user.
To compensate for the overall drop in volume that this introduces, the compressor’s output is then boosted so that it peaks at the same level as the uncompressed signal. As the average level is now higher, thanks to the reduction in dynamic range, it sounds louder – see Fig. 6.
Generally speaking – and up to a point – the louder music gets, the better it sounds. Sadly, as we’ve seen, our waveform can only get so loud: once all the bits have been set to 1 and we’ve reached the 65,536th step, there’s nowhere else for the audio information to go! So, what happens if we’re recording a signal that’s too ‘hot’ (that is, too loud) for our system?
In this situation we get what’s known as digital clipping. When a signal exceeds the maximum capacity of the system, the top of the waveform is simply cut off, or ‘clipped’. This loss of information misrepresents the amplitude of the clipped waveform and produces erroneous higher frequencies. Although clipping can have its uses – which we’ll look at later in the issue – it’s generally undesirable and is the reason why we can’t just keep boosting the level of our mixes to get them as loud as we want.
As a music connoisseur, you’ve probably already noticed that some mixes actually sound a lot louder than others. Sometimes this is because their waveforms simply peak at a higher level, but not always – you might load up two tracks in your DAW and see that they look similar in terms of amplitude, but hear that one sounds a lot louder than the other.
This is where perceived volume comes into play. The human mind’s perception of loudness is based not just on the amplitude level at which a waveform peaks, but also on how long it remains that loud. Consequently, a better way to quantify the perceived volume of a sound is by measuring not its peak volume, but its RMS level.
RMS stands for ‘root mean square’, and is a method of calculating the average amplitude of a waveform over time. Say, for example, we try to calculate the mean amplitude of a sine wave. The result is going to be zero, no matter what its peak amplitude, because a sine wave rises and falls symmetrically around its centre origin.
Clearly, then, this isn’t a useful way to measure average amplitude. RMS (the square root of the mean of the squares of the original values) works better because it always gives a positive value by negating any negatives – it’s a measure of the magnitude of a set of numbers. So, a signal that has been compressed may have the same peak level as the uncompressed version, but its RMS level will be higher, making RMS a better indicator of perceived volume.
When we talk about signal levels, we use a logarithmic unit known as the decibel, or dB for short. Decibels represent the ratio between two different values, so we use them to describe amplitude relatively. For example, in digital audio, a signal at 0dBFS (0 decibels full scale) reaches the maximum possible peak level (+/-1.0) on the vertical axis.
Putting it simply, +6dB doubles the amplitude of a signal, and -6dB halves it. If you want to get technical, it’s actually approximately +/-6.02dB (derived from 20 log 2). So, if we had a sine wave that peaked at 0dBFS, and amplified it by -6.02dB, it would peak at about +/-0.5. Each time we amplify the signal by -6.02dB, the peak level will fall by approximately half again. In terms of RMS, amplifying a signal by +/-6.02dB will give us four times the power or a quarter of the power respectively (see Fig. 7).
So, part of the reason why two tracks can have such different perceived volume levels when they peak at the same level is that they likely have a higher RMS value. That’s not the whole story, though. There are many psychoacoustic factors that affect how loud something sounds to our ears, including its frequency content, how long it lasts and, crucially, its context in a track.
This is where we actually get into the realms of arrangement and – finally! – mixing. At the most fundamental level, mixing two waveforms together is a really simple process: all your CPU has to do is add the level of each sample state together to calculate the total.
This happens in exactly the same way in all DAWs, and if someone tells you that a particular DAW’s summing engine has a certain ‘sound’ to it, they’re sadly misinformed. In fact, there’s a very simple test you can do to see for yourself how all DAWs sum signals in exactly the same way: simply load two or more audio tracks into the DAWs you want to test and bounce out a mixdown.
As long as you use the same settings (including any automatic fades, panning law and dithering) and haven’t done anything to alter the audio (such as apply timestretching), you’ll find that each DAW produces an identical file. Then load the exported files into a DAW and apply some polarity (often erroneously called phase) inversion – as long as all the files have been imported correctly and no effects have been applied by the DAW, they will cancel each other out perfectly, leaving nothing but silence.
Another thing worth bearing in mind about DAW mixers is that they use 32- or (now much more commonly) 64-bit ‘floating point’ summing to calculate what the mix should sound like before it’s either delivered to your ears via your audio interface and speakers or bounced to an audio file.
Floating point means that some of the bits are devoted to describing the exponent, which in layman’s terms means that it can be used to store a much wider array of values. Whereas fixed-point 16-bit has a range of 96dB, 32-bit floating point has a range of more than 1500dB!
This is clearly more headroom than we’d ever have any practical use for, and is the reason why pushing the level faders extremely high or low in your DAW’s mixer doesn’t really have a negative effect on the sound unless it’s passing through a plugin that doesn’t cope well with extremely high or low values, or is clipping the master channel.
In fact, you could run every channel on your mixer well into the red, but as long as the master channel isn’t clipping, no distortion of the signal would occur: 32-bit float can handle practically anything you can throw at it.
To get your mixes sounding as good as the competition it’s necessary to push your digital audio to its very limits
Running your DAW’s master channel into the red is a bad idea, because even if it’s not producing any audible distortion, you won’t really be able to trust what you’re hearing any more; this is due to the fact that different digital-to-analogue converters handle signals exceeding 0dBFS differently. Distortion can even be caused by signals that don’t clip!
This is known as inter-sample modulation, and it doesn’t occur within your DAW – it only happens once the signal has been converted from digital to analogue by your audio interface, specifically the digital-to-analogue converter’s (DAC’s) reconstruction filter. It’s what turns a stepped digital signal into a smooth analogue one, and as Fig. 8 (above left) shows, this process can cause problems when used with especially loud digital signals.
It works a little like this: imagine you want to paint a grey picture, but only have white and black paint. By painting very small squares of alternating white and black, you can create a third colour: grey. Dithering uses the same principle of rapidly alternating a bit’s state to create a seemingly wider range of values than would otherwise be available. See Fig. 9 for a visual example of dithering in action.
Dithering should be applied when lowering the bit rate. It’s therefore sensible to apply dither to your 32- or 64-bit mix when exporting it as a 16-bit WAV file to play in a club, or when supplying a 24-bit WAV file to a mastering engineer.
As such, these ‘lossy’ formats still sound great under normal consumer listening conditions, but they don’t behave exactly like their uncompressed counterparts when subjected to processing. While using an MP3 file as a source for, say, a master isn’t necessarily going to sound bad, the results will really depend on the material, the context it’s used in, and the encoding algorithm that’s been used to compress the sound. To be sure that the finished product sounds as good as possible, it pays to ensure that you use the highest-quality source material you can.
The human ear and mind are surprisingly sensitive listening devices, and, as a result, it’s easy for someone who knows nothing about production to identify a bad or weak mix – it will simply sound less ‘good’ than other tracks.
This is especially important in a club environment, where the music is usually delivered in a non-stop stream, mixed together by a DJ. When the DJ uses the crossfader to quickly cut between channels on the mixer, it’s just like A/B-ing tracks in a studio environment: any deficiencies in the mixdown of either track compared to the other will be glaringly obvious!
To get your mixes sounding as good as the competition it’s necessary to push your digital audio to its very limits, and it’s only through understanding what those limits are that you can achieve this.
Computer Music magazine is the world’s best selling publication dedicated solely to making great music with your Mac or PC computer. Each issue it brings its lucky readers the best in cutting-edge tutorials, need-to-know, expert software reviews and even all the tools you actually need to make great music today, courtesy of our legendary CM Plugin Suite.
Don’t miss the latest deals, news, reviews, features and tutorials
Thank you for signing up to Musicradar. You will receive a verification email shortly.
There was a problem. Please refresh the page and try again.
MusicRadar is part of Future plc, an international media group and leading digital publisher. Visit our corporate site (opens in new tab).
© Future Publishing Limited Quay House, The Ambury, Bath BA1 1UA. All rights reserved. England and Wales company registration number 2008885.
Phaedra and Minotaur review — an inspired double bill of music and dance in Bath – Financial Times
Then $69 per month
New customers only
Cancel anytime during your trial
OR
BEST VALUE – SAVE 20%
Then $74.75 every 3 months
Sign in
Check if your university has an FT membership to read for free.
We use for a number of reasons, such as keeping FT Sites reliable and secure, personalising content and ads, providing social media features and to analyse how our Sites are used.
International Edition
Usyk vs Joshua 2: Timing, pricing, booking details for world heavyweight title rematch on Sky Sports Box Office – Sky Sports
Book and watch repeats of Usyk vs Joshua 2 at 8am and 3pm on Sunday
Sunday 21 August 2022 10:34, UK
Please use Chrome browser for a more accessible video player
Anthony Joshua reignited his rivalry with Oleksandr Usyk in a huge rematch – and you can watch repeats of the epic show on Sunday.
Britain’s heavyweight star attempted to regain the WBA, IBF and WBO titles in his second showdown against Usyk in Jeddah.
Two repeat showings (full duration) will be shown at 8am and 3pm on Sunday August 21.
Please use Chrome browser for a more accessible video player
Booking information
Sky Q
For Sky Q – From Saturday August 13, navigate to sports and then Box Office. Use the arrow buttons to select the event. Press select, and the buy screen will be displayed. Press select to confirm your order. Enter your pin if prompted. Alternatively, from Saturday August 6 go to Channel 491 and press the green button to go to Sky Sports Box Office, following the same actions.
Sky+
Book Usyk vs Joshua 2 now!
For Sky+ – From Saturday August 13, press Box Office. Use the arrow buttons to select the event. Press select, and the buy screen will be displayed. Press select to confirm your order. Enter your pin if prompted. Alternatively, from Saturday August 6 go to Channel 491 and press the green button to go to Sky Sports Box Office, following the same actions.
You can only record the event by using this method.
Sky Glass – From Monday August 15
You can purchase the PPV event from Monday August 15 on Sky Glass:
You’ll be charged for the event when you enter your PIN (if enabled), and the order is confirmed.
If you change your mind about the event, you’ll need to call us before the event starts. There may be a delay of 45 seconds or more behind the live event.
Error messages
To book online (Sky Q and Sky+ customers only)
Book Usyk vs Joshua 2 repeats now!
There is no surcharge for online bookings.
To watch online via the Sky Sports Box Office App
Watch Usyk vs Joshua 2 online through the Sky Sports Box Office app and website.
If you don’t have a Box Office account, you’ll be able to sign up and buy the fight right now.
How to order:
Visit the Sky Sports Box Office page
If you don’t have a Box Office account, select “Buy now for £26.95 / €31.95 (ROI) “
Select “Sign up”
Select “New to Sky Box Office” and “continue to checkout”
Enter your personal and card details and pay.
This will give you access. You’ll also have seven days of unlimited access to watch back the fights from when the event begins.
Ways to watch
Now
Phoneline
To order the event, call 03442 410888 (0818 220 225 from ROI) (there may be a £2 booking fee for telephone bookings). There is a £5/€5 surcharge for bookings made via the phone from midnight Friday August 19 to midnight Saturday August 20.
Book Usyk vs Joshua 2 repeats now!
Non-Sky cable customers
Usyk vs Joshua 2 is available to non-Sky customers with Virgin Media and BT.
Sky Sports Box Office terms apply and a booking fee may also apply.
© 2022 Sky UK
Lessons from a Creative Summer – Pratt – Pratt News
The 2022 recipients of The Black Alumni of Pratt (BAP) PreCollege scholarships that support Black and Latinx high school students based in the New York City area share what’s next in their artistic lives.
Over four weeks this summer, 200 high school students from around the world took college-level courses on the Pratt Institute Brooklyn campus to develop their creative skills and prepare to turn their passions into higher education pursuits. The annual PreCollege program offers hands-on learning to young artists, designers, architects, writers, filmmakers, and more to study with world-class faculty.
The Black Alumni of Pratt (BAP) annually provides full scholarships for PreCollege for Black and Latinx students based in the New York City area. They are selected based on the strength of their portfolios with the scholarships funded by contributions from donors to The Black Alumni of Pratt Endowed Fund. Many BAP Scholars have gone on to attend Pratt or other art and design colleges. This year’s BAP PreCollege Summer Scholars were Ari Merchan Espada, Yaslin Polanco, and Av Sanchez.
While in 2020 and 2021, PreCollege was virtual, this summer PreCollege was back in-person. After an inspiring month of learning at Pratt among like-minded students, the BAP Scholars shared what they learned, how it influenced their interests, and what’s next for their art and education.
BAP Scholar Ari Merchan Espada of Port Jervis studied digital illustration, drawing, and painting at PreCollege and particularly enjoyed immersing in the study of art.
“I don’t think I could pin down a specific moment when I realized my love for art,” they said. “I remember always enjoying drawing, doodling all over the sides of my schoolwork, and being known as ‘the art kid.’ My family always shared things they enjoyed with me, and I would appreciate the artistic work that was put into it: the puppetry and creature design from my mom’s favorite old fantasy movies, the pixel art and storytelling in the games my sister played, and the color schemes and patterns of the quilts my great grandma made.”
PreCollege was an opportunity to meet new people from across the country and the world and discover what they had in common, especially relating to these artistic passions.
“Being at PreCollege definitely taught me I have a lot to work on, but it also showed me things to appreciate about my work and to look at what others see in it, in a way I wouldn’t have noticed otherwise,” Espada said. “I learned to be a little less hard on myself in regards to judging my own work. I also gained some good skills for college in getting a preview of what that’s like.”
Espada is now preparing for college applications by working on a portfolio and is looking forward to making art part of a career.
“It would be my dream to have art as my profession, and that’s really what I’m trying to work towards,” they said. “I would love to start my own animation studio one day. I know the industry can be tough, but I see so much potential in it that I feel most people aren’t taking advantage of. I want to create something beautiful that will really stand out. Creating art is just part of who I am, and I think that will follow me in whatever I end up doing.”
BAP Scholar Yaslin Polanco of Brooklyn studied sculpture, drawing, and painting at PreCollege. During the pandemic, she spent more time painting, leading her to start a small painting business to contribute to her family’s needs.
“My parents are my inspiration because for me they are among the hardest working and strongest people there are,” she said. “Thanks to them, I was able to come to the US to have the opportunities that they never had in our native country.”
For Polanco, PreCollege was a chance to try out new approaches to art. “That’s why one of the courses I chose was sculpture,” she said. “I had never done anything related to sculpture since I usually do drawings or paintings. When our teacher taught us various techniques with wood or making body parts with plaster, I realized that I want to continue making sculptures and also keep learning in the future.”
She enjoyed exploring different artistic media and learning from the other students, each with their own perspectives and styles.
“One of my PreCollege courses was drawing and painting,” Polanco said. “At first, I did not understand the painting techniques that my teacher explained and I felt that I was not going to be as good as my classmates. After my teacher gave me a critique and examples, I began to add more color and make the shapes simpler. I practiced day and night and at the end of it all, I felt like I wanted to use those techniques for my future projects. We must always learn different things or new techniques and never get stuck with a single style or a specific material.”
BAP Scholar Av Sanchez of Queens studied industrial design and illustration at PreCollege. Going into the program, Sanchez was especially interested in getting an experience with college life.
“It was incredible spending my summer on Pratt’s Brooklyn campus with other immensely talented art and design students,” they said. “I loved having the opportunity to interact with students from all over the globe. There’s great value in hearing diverse perspectives and exchanging ideas with others.”
While they arrived at PreCollege with some experience with illustration, industrial design was an opportunity to try something new, increase their skills and knowledge, and consider other creative career possibilities.
“I learned about new techniques and mediums that could strengthen the themes I strive to convey in my works,” Sanchez said. “For example, splatter paint art is a great way to express our emotions. The rigorous college courses taught me how to receive assignments and execute them in a short period. My professors also taught me that if you want something bad enough, you’ll go out and fight for it. This is a motto I’ll live by.”
Sanchez is now planning to attend the Pratt Young Scholars program. They are looking forward to continuing to be inspired by the cultural environment of New York City and discovering how they can channel that energy into their work.
“I am extremely appreciative of what The Black Alumni of Pratt has done for me,” Sanchez said. “They have provided me with the opportunity to advance my education and become a member of such a diverse community. I am honored to be a part of the BAP legacy. Without BAP’s guidance and scholarships, I don’t think I would’ve understood I wanted to pursue an art education. Most importantly, I learned about the value of helping those in need. I want to be a voice for and support students of African and Latinx origins. I intend to remain a member of the BAP community for many years to come.”
Pratt Institute © 2022
Privacy Policy
Bowling For Soup's Jaret Reddick Goes Country With a Solo Album – Dallas Observer
Dallas’ independent source of
local news and culture
account
DFW Music News
August 22, 2022 4:00AM
“A lot of people weren’t super surprised whenever I made the move to country, because [Bowling For Soup] had done several country-leaning songs throughout our career.” – Jaret Reddick
10 Best Concerts of the Week: Scarface, Echo & The Bunnymen, Rob Zombie and More
By David Fletcher
Two Dallas Lawyers Teamed Up to Get Session Musicians $45 Million in Unpaid Royalties
By Christian McPhate
The Best of the Worst Reviews of Dallas Music by Amazon Customers
By Eva Raggio
Join the Observer community and help support independent local journalism in Dallas.
Get the latest updates in news, food, music and culture, and receive special offers direct to your inbox.
Become a member and go ad-free!
The Legend of Zacian (and Zamazenta?) in GBL Master League | Pokémon GO Hub – Pokémon GO Hub
Hello, Pokéfriends! Well here we go, my sixth article over the last eight days. You’re probably sick of me by now. 😅 But this is actually the one I started on FIRST and just kept pushing it onto the back burner as the others rose in priority and had impending deadlines. So unfortunately, the subjects of today’s analysis are already out in raids as I finish this article, but the good news is there is still plenty of time to raid the heck out of ZACIAN and ZAMAZENTA, and yes, it’s well worth doing so. (Especially since, as I wrote about back in June, they’ve only been available for a mere seven days prior to this, far less than most other Legendaries.) I’m sure it’s no surprise if I tell you that Zacian in particular is very, very good in GBL Master League, but today we look at WHY.
I’ll come out and say it: I underestimated it during my initial analysis on it a year ago, when it was first released. Because what I primarily do is examine simulated battles, and those, quite frankly, sell Zacian short. Because it doesn’t play like most other Pokémon and like PvPoke and other simulators expect. And that is part of why it has taken Master League by storm.
So for this particular analysis, I’m doing things a little differently. I’ll still reference some pertinent sims but want to focus on some other analysis too. Some of it from other fellow players, not just ol’ me. Hopefully, by the end, we can get at what really makes Zacian tick, beyond even the raw numbers.
But before I go any further down that road, here’s our Bottom Line Up Front before we get into the meat of the article.
Table of Contents
Alright, now for the deeper dive….
(Assuming 15-15-15 IVs; CP 3829 at Level 40; CP 4329 at Level 50)
So the stats are… eh, alright. The closest and easiest comp I can find is Garchomp, who has slightly higher Attack and has basically a mirror image of Zacian’s Defense and HP. (Chompy has about 232 Attack, 175 Defense, and 213 HP at Master League level.) So Zacian will certainly never be considered “tanky”.
BUT, what it does have is a very favorable typing. I took a little poll on Twitter yesterday, asking what about Zacian impressed or just scared them the most. Of course many mentioned the unpredictable charge moves (we’ll get to that later), but just as many said that a huge part of Zacian’s threat profile was that it’s just so hard to counter, and that comes in very large part down to the typing. Fairy is a type commonly encountered in all major GBL Leagues, from Cottonee in Little League and Wigglytuff in Great League and all the way up to Sylveon and long-time ML stalwart Togekiss in Master League. But it’s still a relative rarity in Master League in particular, with really only Togekiss, sometimes Sylveon, and now the much different Zacian representing the type. Togekiss and Sylveon are of course Charmers, wearing things down primarily — oftentimes entirely — with heavy STAB fast move damage, and their charge moves may steal a shield here or there, but are mostly an afterthought. Zacian… Zacian operates quite differently.
I want to just throw out a few of the comments I got from the community regarding the typing and how it helps Zacian before I move on:
Think about it. How many great Steel types (that actually throw out Steel damage) are there in Master League? Basically just Dialga and Metagross, right? And Dialga, of course, has all of its normally-threatening Dragon damage resisted by Zacian, leaving it wholly dependant on getting an Iron Head past shields. And now how many Poison types can you think of? Uh… maybe an odd Poison Jab Buzzwole, or the occasional Sludge Wave Swampert, but uh… that’s it, yeah? Now of course, there’s more that Zacian has to be wary of than JUST Steel and Poison, as its more common moves can make Flyers like Ho-Oh and Landorus and Psychics like Mewtwo and Ghosts like the Giratinas and several others problematic. But as hinted by a couple of those PvPers quoted above, and as you’ll see as we progress through this analysis, Zacian has answers for nearly all of them.
And that comes from the moves.
So in my original review of Zacian a year ago, back before Zacian emerged as the menace it is now, I was actually weighing Fire Fang as potentially the way to go with Zacian to cover its famous weakness to Steel types. And while I still say that has some merit as a way to mess with the opponent’s expectations, clearly what has made Zacian so scary is maximum energy gains from Snarl. Because it is those varied and unpredictable charge moves that have made Zacian the fear-inducing monster it is today.
Speaking of which….
There are multiple viable Zacian variants out there, but they always start with Close Combat. As noted earlier, the biggest threat to Zacian in the current Master League meta is Steel damage. Not much has it, but the few things that do (primarily Metagross and Dialga) have much more reason to fear Zacian than other Fairies. Zacian forces them to burn shields, at the very least, if not beating them outright with Close Combat alone. This is particularly true of big bad Dialga, who does manage to win in a drawn-out 1v1 shield fight, but loses in 0v0 and 2v2 shielding. And realistically, many Zacian players will throw a CC or two at Dialga in 1v1, take a massive chunk of its health, and then swap out with over half its HP remaining before Dialga ever lands its KO Iron Head. Close Combat, even just on its own, also easily handles other Steels like Excadrill, Magnezone, and Melmetal, as well as other things weak to Fighting like Mamoswine and Snorlax, plus others like Machamp, Palkia, and Garchomp.
But that, of course, is just the start. There are two primary Zacian configurations beyond that, one with similarly risky but potent move Wild Charge and one with slower but still potent (and less risky) Play Rough. Put as simply as possible, Wild Charge allows Zacian to most reliably win the mirror match (and of course most efficiently beat Flyers and/or Waters like Lugia and Gyarados), while Play Rough is obviously best versus Dragons (specifically by uniquely beating Altered Giratina and Dragon Breath Dragonite).
But yet again, that’s not the whole story….
I could fill an entire LONG article with tips and tricks like this, but the basic point is this: Zacian in actual gameplay is one of few things that well overperforms how it looks in simulations. In experienced hands, Zacian is a meta-warping monster. Part of it is having a typing that leaves it exposed to very few direct threats. Part of it is having answers for even those “hard” counters, with Close Combat for Steels and Wild Charge for Fairy killers like Ho-Oh and Play Rough for most everything else. Part of it is certainly not knowing which of those moves Zacian can actually bring to bear until you’ve already burned a shield or even two feeling it out. Because that’s what Zacian does best: keep you guessing, making every “Attack incoming!” shielding decision a nerve-racking, sweaty decision.
Also as submitted by friends in the PvP community:
And it really does just come down to that last one: you have to shield it, at least until you know for sure you can survive whatever it’s throwing. Late in a battle, you may KNOW it has only Close Combat and Wild Charge and therefore your, say, Landorus can stand tall and tank a charge move. Or you may know your Gyarados is facing down only Close Combat or Play Rough and can breath a sigh of relief knowing the incoming attack is at least not a deadly Wild Charge. But until you KNOW that, for sure, you often have to burn shields even when you don’t want to. That makes Zacian very deadly as a lead, as you have to either burn shields early to stay in or give up switch advantage right away. It makes it deadly as a closer, as any shields you’ve been forced to use earlier will now be desperately missed. And it makes it even a dangerous swap, as it may not WIN all its battles, but it can produce energy faster than most everything else and outrace you to charge moves, stealing shields or big chunks of life and getting the Zacian player right back into the game in a hurry. This is why, despite having it ranking modestly in Master League at #25 (at the time of this writing), PvPoke still notes of some excellent traits with Zacian:
“Bulky, Spammy, Flexible, Dynamic, and (great at) Shield Pressure”, with “Technical” (using “complex moves that may have a high learning curve”) being the sole negative. Kind of says it all, doesn’t it?
So, there you go. An analysis where I was forced to come out of my normal simulation-heavy shell a bit, because that’s Zacian in a nutshell: it doesn’t conform to any standards but its own. The sims don’t tell the story properly of how dominant and meta-changing Zacian has been in Master League.
Ah yes, ZAMAZENTA. Poor little doggy… almost forgot about it.
‘Zenta has literally the exact same stats as Zacian, so nothing to particularly discuss there. The big difference right off the bat is the typing, as Zacian is pure Fairy, while Zamazenta is a pure Fighting type. And unlike Fairy, Fighting is a rather uncomfortable fit in Master League. It resists Bug, Dark, and Rock damage, none of which you see a lot of in Master League. Conversely, it is vulnerable to Psychic, Flying, and Fairy damage, all of which you DO see a good amount of in the Master League meta.
Then there’s the moves. Close Combat is still there, with Snarl to still power it out. And CC gets STAB now too, which is nice. There’s also Moonblast as a strictly better version of Zacian’s Play Rough, dealing20 more damage for the same cost, AND a 10% chance to debuff the opponent’s Attack! That’s all awesome, but things do drop off beyond that. No Wild Charge, replaced instead by Crunch, which is a handy move in theory for countering those problematic Psychic types, but hits like a marshmallow against Fairies and is a significant dropoff in power from something like… well, Wild Charge. Even with Psychics out there, Snarl Zamazenta is still better off running CC/Moonblast.
However, even with all that put together, this is Snarl Zamazenta’s best. Not awful, especially when one considers that sims working the way they do undersells things like Zacian and now Zamazenta, but still rather uninspiring. But the pluses are solid wins over Dialga (unlike Zacian, Iron Head isn’t so scary now), as well as wins over Dragon types Palkia, Dragonite, and Mud Shot Garchomp. But it fails to beat either of the Giratinas (or any other notable Dragons, for that matter), and while it can take out Melmetal and Snorlax and Mamoswine and Magnezone and Excadrill and even Machamp and Yveltal like Zacian, it ends there. All the Fairies are obviously a loss, and there’s no realistic path to victory over staples like Lugia, Gyarados, either Giratina, and of course Zacian itself.
BUT, there is some hope yet for ‘Zenta that I think a lot of folks have missed. Because Zamazenta trying to run with Snarl and heavy charge move pressure isn’t the way to go… Zacian already does that job better in every way. If I were to consider running Zamazenta, it would instead be with Ice Fang. That gives it a real niche as a sorta-Fighter (can still beat things like Melmetal, Mamoswine, Snorlax, and Magnezone, but now loses to Dialga — though it CAN still win that with one fast move’s worth of energy — and Excadrill) that can also take out most Dragons (Dragonite, all versions of Garchomp, and both Giratinas now) and Grounds (specifically picking up Groudon, Swampert, and both Landos) AND things like Zarude and Play Rough Zacian!
But is that worth the candy and XL grind to build? Well, if I’m being honest, probably not… especially when you consider that every Zamazenta raid means taking time and resources away from raiding Zacian instead.
Alright, that’s it for now! I hope you enjoyed this article and learned from it… not just that Zacian is good, but the why and the how behind that.
Until next time, you can always find me on Twitter with near-daily PvP analysis nuggets or Patreon, if you’re feeling extra generous.
Thanks for your faithful readership, and happy raiding! May all your raids give you Candy XL, and may your hundo Zacian materialize sooner rather than later. Do be safe out there, and catch you next time, Pokéfriends!
Pokémon GO Hub is the biggest Pokémon GO news site, publishing several informative guides, analysis, and news articles every month. Dedicated, focused and loving Pokémon GO.
© 2016 – 2022 Pokémon GO Hub | ALL RIGHTS RESERVED
Pokémon and Pokémon GO are copyright of The Pokémon Company, Niantic, Inc., and Nintendo. All trademarked images and names are property of their respective owners, and any such material is used on this site for educational purposes only. Solver d.o.o. has no affiliation with The Pokémon Company, Niantic, Inc., or Nintendo.
How Julian 'LilJuMadeDaBeat' Mason Became a Go-To Producer For Megan Thee Stallion and Beyoncé – Dallas Observer
Dallas’ independent source of
local news and culture
account
DFW Music News
August 22, 2022 4:00AM
“I love being from Dallas. I wouldn’t want to be from any other place in America.” – LilJuMadeDaBeat
10 Best Concerts of the Week: Scarface, Echo & The Bunnymen, Rob Zombie and More
By David Fletcher
Two Dallas Lawyers Teamed Up to Get Session Musicians $45 Million in Unpaid Royalties
By Christian McPhate
The Best of the Worst Reviews of Dallas Music by Amazon Customers
By Eva Raggio
Join the Observer community and help support independent local journalism in Dallas.
Get the latest updates in news, food, music and culture, and receive special offers direct to your inbox.
Become a member and go ad-free!
Fluid mixing optimization with reinforcement learning | Scientific Reports – Nature.com
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Scientific Reports volume 12, Article number: 14268 (2022)
Metrics details
Fluid mixing is crucial in various industrial processes. In this study, focusing on the characteristics that reinforcement learning (RL) is suitable for global-in-time optimization, we propose utilizing RL for fluid mixing optimization of passive scalar fields. For the two-dimensional fluid mixing problem described by the advection–diffusion equations, a trained mixer realizes an exponentially fast mixing without any prior knowledge. The stretching and folding by the trained mixer around stagnation points are essential in the optimal mixing process. Furthermore, this study introduces a physically reasonable transfer learning method of the trained mixer: reusing a mixer trained at a certain Péclet number to the mixing problem at another Péclet number. Based on the optimization results of the laminar mixing, we discuss applications of the proposed method to industrial mixing problems, including turbulent mixing.
Fluid mixing plays a fundamental role in various industrial processes. However, most mixing processes are empirically designed by using trial-and-error methods through physical experiments, rather than mathematical optimization. Although turbulence is an “effective mixer”1, in some cases (e.g., a bioreactor or a mixer in food industry processes), turbulent mixing is not appropriate because strong shear flows damage the materials to be mixed. Moreover, sustaining turbulent flows in micro-mixers is difficult due to low Reynolds numbers; this necessitates enhanced mixing by laminar flows. Therefore, mixing optimization by laminar flows is crucial. Several analytical studies have evaluated the efficiency of laminar mixing protocols2,3,4,5, e.g., proving the exponential bounds of the mixing speed; however, the research on constructive optimization methods remains limited.
This study proposes a mixing optimization based on reinforcement learning (RL) as a constructive method. To illustrate the effectiveness of the RL algorithm for fluid mixing optimization, we first summarize its mathematical framework. The RL algorithm is formulated in terms of the Markov decision process (MDP)6,7: (M= { {S}, {A}, p_{0}, P, R}), where S denotes the set of states, ({S}={s_1, cdots s_{|{S}|} }); A denotes the set of actions, ({A}={ a_{1}, cdots a_{|{A}|} }); (p_{0}) denotes the probability distribution of the initial state, (p_{0}: {S} rightarrow [0,1]); P denotes the transition probability, (P: {S} times {S}times {A} rightarrow [0,1]); and R denotes the reward function, (R:{S} times {A} rightarrow mathbb {R}). The initial state, (s_{0}), is determined by (p_{0}(cdot )), and in the next step, the state is determined by the transition probability, (P(cdot |s_{0},a_{0})), which requires the action, (a_0). The action is determined by the policy, (pi : {S} rightarrow {A}), as (a=pi (s)). The RL algorithm is implemented to determine the optimal policy, (pi ^*), for the given MDP, which maximizes the expectation of the cumulative reward, (sum _{t=0}^{infty } gamma ^{t} R_{t+1}). Here, (gamma in (0,1)) denotes the discount factor and (R_{t+1}:=R(s_{t},a_{t})).
The RL algorithm maximizes the cumulative reward (i.e., global-in-time) rather than the instantaneous reward, (R_{t}) (i.e., local-in-time). Therefore, it is suitable for global-in-time optimization problems. Designing efficient mixing protocols is one of the global-in-time optimization problems, as the final scalar field depends on the temporal order of actions in the entire mixing process, which includes stretching and folding by fluid flows and its coupling with molecular diffusion. An illustrative example was presented in History matters of Villermaux8. Despite the effectiveness of the RL algorithms in solving a diverse range of problems in fluid mechanics9,10,11, including nuclear fusion12 and turbulence modelling13, the fluid mixing problem remains unexplored.
The RL algorithm is suitable for the global-in-time optimization problems, but not for problems with a high-dimensional state space in general, which is known as the curse of dimensionality6. Particularly, the high dimensionality of the state space for fluid mixing renders the implementation of the RL algorithm challenging. This study investigates an optimization problem formulated by Mathew et al.2, in which the velocity field is given by the superposition of the prescribed fields. This reduces the dimension of the state space for the fluid motion to one2; a single parameter, denoted by (theta ) later, determines the state of the fluid motion. This optimization problem was based on a physical experiment using the electromagnetically driven flow14. The conjugate gradient descent method was introduced as a prototype of the fluid mixing optimization2. To ensure that the RL algorithm can handle the flow field with a reduced degree of freedom, we focus on the same optimization problem.
In this paper, we demonstrate for the first time that the RL algorithm is suitable for fluid mixing optimizations. This algorithm identifies an effective flow control, which results in exponentially fast mixing without prior knowledge. The mechanisms behind efficient mixing are uncovered by focusing on the flow around the fixed points from the viewpoint of dynamical system theory15,16. This study also proposes an effective transfer learning method for the trained mixer by considering the diffusion effect on mixing. Based on the optimization results of the laminar mixing, we discuss applications of the proposed method to industrial mixing problems, including turbulent mixing, in the “Conclusion and discussion” section.
We consider the following optimization problem formulated by Mathew et al.2 as the benchmark problem, in which the velocity field, (u(x,y,t) = alpha _{1}(t) u_{1}(x,y) + alpha _{2}(t) u_{2}(x,y)), is used. Here, (u_1(x,y) = (-sin (2 pi x)cos (2 pi y), cos (2 pi x) sin (2 pi y))) and (u_2(x,y) = u_1(x-0.25,y-0.25)) (see Fig. 1a). The time evolution of the passive scalar, c(x, y, t), is described by the advection–diffusion equations on the two-dimensional torus, (mathbb {T}^2) (the periodic boundary condition):
where (text {Pe} in (0,infty ]) represents the Péclet number. As a constraint on the flow control, the time integral of the kinetic energy, (frac{1}{2} int _0^{1} int _{mathbb {T}^2} u^{2} dmathbf{x} dt = int _0^{1} alpha _i(t) R_{ij} alpha _j (t) dt =:mathscr {E}), is fixed, where (R_{ij}:= frac{1}{2} int _{mathbb {T}^2} u_i cdot u_j dmathbf{x}~~(i=1,2,~j=1,2)). We set (alpha (t)=2sqrt{mathscr {E}} (cos theta (t), sin theta (t))), by which the constraint is always satisfied. We also set (mathscr {E}=1.25) as in Mathew et al.2. In this problem, the velocity field, u(x, y, t), is determined by a single parameter, (theta (t)), referred to as the flow parameter.
Optimization of fluid mixing using the reinforcement learning (RL) algorithm: (a) velocity vector fields of (u_{1}(x,y)) (left) and (u_{2}(x,y)) (right) on the two-dimensional torus, (mathbb {T}^2); (b) snapshots of the time evolution of scalar field, c(x, t), advected by the optimized flow; (c) mix-variance, (Phi _{n}(t)~(t in [0,1])), for the nth episode ((n=1), 800, 1600, 2400, 3200, and 4000); (d) mix-variance, (Phi _{n}(t=1)), at the end of each episode.
The variance of the scalar field is often used to measure the degree of mixedness. However, as it is a conserved quantity in the absence of diffusion (i.e., (frac{d}{dt} int _{mathbb {T}^2} c^{p}dx equiv 0~~(forall p in mathbb {N}))), it is unsuitable as a measure of the mixing process. Instead, we employ the mix-variance defined by (Phi (c)=Vert c Vert ^2_{H^{-1/2}} := sum _{k} frac{1}{sqrt{1+ (2 pi Vert k Vert )^2}} | c_k |^2), where (c_k) denotes the Fourier coefficient of the scalar field17. The mix-variance is equivalent to the Mix-Norm that was originally introduced to characterize the multiscale property of the mixed scalar field17. Furthermore, Mathew et al.17 showed the equivalence between the decay of (Phi (c)), the weak convergence in (L^{2}), and the mixing of ergodic dynamical systems in Theorem 3.2 (see also Lin et al.3 for the extension of the theorem). To summarize the optimization problem, we use the RL algorithm to determine the function, (theta : [0,1] rightarrow mathbb {R}), that minimizes the mix-variance at the end of the mixing process, (Phi ( c (cdot , t=1))).
We conduct a numerical simulation of the advection–diffusion equations (Eq. 1) using the fourth-order Runge–Kutta scheme for the temporal integration with (Delta t=0.001) and the Fourier spectral method for spatial discretization with a grid of (250 times 250), which is the same as the one used in Mathew et al.2.
Here, we consider the optimization of the action-value function (Q function) (Q^{pi }(s,a):= mathbb {E}[ sum _{t=0}^{infty } gamma ^{t} R_{t+1}|S_{0}=s, A_{0}=a]) instead of the policy (pi ), and obtain the optimal Q function, (Q^*: {S} times {A} rightarrow mathbb {R}). The Banach fixed-point theorem mathematically ensures that such an optimal Q function exists as a fixed point of the Bellman operator6,7. We obtain the optimal policy as (pi ^{*}(s) := mathrm{argmax}_{a in {A}} Q^{*}(s,a)).
As a standard implementation of the RL algorithm, we employ the deep Q network18, which approximates the Q function by using the deep neural network denoted by (Q^w: mathbb {R}^{N_s} times {A} rightarrow mathbb {R}). Here, (N_s) and w denote the dimension of the state space and the connection weights in the neural network, respectively. The inputs to the network are the scalar field, c(x, y, t), and the velocity field, u(x, y, t). The values of these fields on (mathbb {T}^2) are observed on the square grid (83times 83), and the state, s, of the MDP is defined as the observed values of the velocity field, (u(x_{i}, y_{i}, t)~~(i =1,ldots , N_O)), where (N_O=83times 83), and those of the scalar fields over the last five steps; that is, (s_{t} := (u (x_{i},y_{i}, t), { c(x_{i}, y_{i}, tau ) }_{tau = t,~t- Delta t_{O},ldots ,~t – 4 Delta t_{O}}) in mathbb {R}^{N_s}), and (Delta t_O) denotes the time interval of the successive observations. Therefore, the dimension of the state space is (N_s=7 times N_O). The network consists of four hidden layers, and each activation function is ReLU as Mnih et al.18. The discount factor is (gamma =0.99). The more details of the deep Q network structure and its implementations are described in the “Supplementary Information”. The initial distribution, (p_{0}), is given by the delta function such that (theta (0)=0) and (c(x,y,0)=sin (2 pi y)).
The time interval of the successive observations is (Delta t_O=0.004), which is the same value used in the benchmark problem2, and (Delta t_Q=5 Delta t_O), where (Delta t_Q) denotes the time interval of the successive updates of the Q function. Hence, for each period of (Delta t_Q), the RL algorithm observes the scalar fields determined by the advection–diffusion equations (Eq. 1) with the fixed velocity field. Then, the Q function, i.e., the weights in the neural network, is updated. A single unit of episode corresponds to a single mixing process, i.e., solving the initial value problem of the advection–diffusion equations (Eq. 1) for (0 le t le 1). The total number, (N_{e}), of episodes for the training is (N_{e}=4000). The results with the larger number of episodes, (N_{e} = 5000), are qualitatively identical to those with (N_{e} = 4000).
As the action, A, of the MDP, the RL algorithm can change the value of the flow parameter, (theta (t)~~(0le tle 1)). The velocity field, u(x, y, t), is determined by the single-parameter (theta (t)), and the flow control is realized by changing (theta (t)). The discretization of the temporal change of the flow parameter is (theta (t + Delta t_{Q}) = theta (t) + omega Delta t_{Q}) with (omega in { 0, omega _{+}, omega _{-}}(={A})), where (omega _{+}=pi /(4Delta t_{Q})) and (omega _{-}=-pi /(4Delta t_{Q})). The action, (omega ), is selected following the (varepsilon )-greedy method6,7,18, which changes the value of (varepsilon ) linearly from 1 to 0.001.
The reward function, R, is defined by using the mix-variance, (Phi ), which is set to be a monotonically decreasing function of (Phi ) to ensure that the smaller value of (Phi ) represents a better mixed scalar field:
where (tilde{Phi }), (Phi _{0}), and (Phi _{T}) denote a threshold, an initial value, and a target value of the mix-variance, respectively. By definition, (R=-1) initially, and (R = +1) if the mix-variance, (Phi ), attains the target value. The values of (tilde{Phi }) and (Phi _{T}) are set based on the Péclet number: ((tilde{Phi },~Phi _{T})=(1 times 10^{-2},~4 times 10^{-3})) for (text {Pe} =infty ) and ((tilde{Phi },~Phi _{T})=(5 times 10^{-3},~1 times 10^{-4})) for (text {Pe}=100).
The optimization results are presented in the absence of diffusion ((text {Pe}=infty )). The optimal policy, (pi ^*: mathbb {R}^{N_s} rightarrow {A}), approximated by the deep Q network, is obtained from the RL-based optimization. Thereafter, the state vector, (s_t in mathbb {R}^{N_s}), determines the optimal action through (omega _t = pi ^*(s_t)). This determines the velocity field during the next interval, (Delta t_Q), which advects the scalar field, and the process carries on to the next observation. This flow controller based on the optimal policy, (pi ^*), is referred to as a trained mixer. Figure 1b shows from the left to right panels that the trained mixer makes the scalar field, c(x, t), evolve in time. Here, the black and white colors correspond to the high and low values of the scalar field, respectively. The trained mixer produces a complex layered structure of the scalar field. The following subsection presents a detailed description of the successive stretching and folding of the interface between the two colors.
The mix-variance, (Phi _{n}(t)~~(n=1,ldots , 4000)), is shown in Fig. 1c. During the initial stage of the training, (i.e., in the first half of the total episodes such as (n=1, 800,) and 1600), the RL algorithm with the (epsilon )-greedy method chooses actions randomly. Although this “random mixer” can decrease the mix-variance, such mixing is inefficient, as explained below.
Mathew et al.2 reported that the proposed conjugate gradient descent method resulted in (Phi (t=1) simeq 6 times 10^{-3}); this value of the mix-variance is used for the comparison as a reference. In the first half of the total episodes, the mix-variance at the end of the mixing process, (Phi _{n}(t=1)), is larger than the reference value; that is, the insufficient training of the mixer results in inefficient mixing. Conversely, (Phi _{n}(t=1)) is reduced in the latter half of the total episodes, (n=2400), 3200, and 4000. Particularly, (3 times 10^{-3}< Phi _n(t=1) < 4 times 10^{-3}) for (n=4000), which are almost identical to (slightly smaller than) the reference value. Interestingly, the mix-variance decreases exponentially fast for (0.3 le t le 1) for the latter episodes such as (n=3200) and (n=4000). While here we focus on the quantitative comparison using the mix-variance, there are some qualitative differences between the method by Mathew et al.2 and our RL-based method. In the “Conclusion and discussion” section, we illustrate significant advantages of the RL-based method.
Figure 1d presents the mix-variance at the end of each mixing process, (Phi _n(t=1)), which fluctuates due to the (varepsilon)-greedy methods and the fact that the policy, (Q^{w}), is not converged. However, the fluctuation decreases as the episode progresses; see also Figs. S1 and S2 in the “Supplementary Information”. The RL algorithm significantly decreases the mix-variance, (Phi _n(t=1)); that is, the RL-based optimization effectively enhances the mixing.
The flow parameter in episode n is denoted by (theta _{n}(t)). In the first half of the training, (n<2000), the flow parameter, (theta _{n}(t)), evolves randomly in time due to the (varepsilon )-greedy methods and the fact that the policy is not converged. However, as the episode progresses, (theta _{n}(t)) converges to a single function, (theta ^*(t)), except for the final stage of the process, as shown in Fig. 2a. The time series of (theta _{n}(t)) consists of square waves, as the velocity field (i.e., (theta _{n}(t))) is fixed in each interval, (Delta t_Q). The optimal mixing process by the trained mixer corresponding to (theta ^{*}(t)) is divided into the following three stages:
Initial stage ((0< t le 0.3)): the flow parameter is a constant; (theta ^{*}(t) =pi /4), indicating the steady flow, (u(x,y)= – c sin 2 pi (x+y),~ v(x,y)= c sin 2 pi (x+y)~~(c:text {const.})). The velocity vector is parallel to the diagonal line; for example, the flow along the line, (x+y=1/4), traverses the domain, (mathbb {T}^2), with the velocity vector ((u,v)=(-c,c)).
Middle stage ((0.3 < t le 0.7)): the flow parameter changes linearly; (theta ^{*}(t) = omega ^{*} t~~(omega ^{*} simeq 16)), indicating the temporally periodic flow with a constant angular frequency.
Final stage ((0.7 < t le 1)): there are no common characteristics of the time evolution of the flow parameter.
Characteristics of the trained mixer. (a) Time series of the flow parameter, (theta _{n}(t)), for the n-th episode: (n=2000), 3000, and 4000. The vertical axis is (theta _{n}(t)/pi ) and the horizontal dot lines represent (k/4~(k in mathbb {Z})). Inset: the mix-variance, (Phi (t=1)), in the case of the flow parameter with the constant angular frequency, (theta (t)= omega t). The horizontal dashed-dotted line indicates the value of the mix-variance by the trained mixer. (b) Time evolution of the velocity vector field by the trained mixer. The blue line and the red point represent the material line and one of the fixed points, respectively: (t=0,~0.1,~0.2,~0.3,~0.36,~0.46,~0.54), and 0.68. (c) Probability density functions of the mix-variance, (Phi (t=1)), by the completely randomized mixer and the partially randomized mixers I and II from top to bottom panels. (d) Scalar fields, (c(x,t=1)), at the end of the mixing process by the trained mixer (left) and the completely randomized mixer (right).
The different time evolutions of the flow parameter, (theta (t)), in the final stage result in the almost identical value of the mix-variances, (Phi (t=1)), at the end of the process. Therefore, the essential process in the mixing is not the final stage but the initial and middle stages. Numerical experiments supporting this point are presented in the next subsection.
Figure 2b presents the time evolution of the velocity fields corresponding to the flow parameter obtained in the final episode, i.e., (theta _{n}(t)~(n=4000)). The blue line in each panel indicates the material line advected by the flow, which is initially placed along with the line (y=0.5). The upper panels in Fig. 2b depict the flow in the initial stage, where the material line is elongated to have the diagonal length of the domain. Subsequently, the flow is temporally periodic in the middle stages, which are shown in the lower panels in Fig. 2b. Each velocity field has eight fixed (stagnation) points, (u_1) and (u_2). Half of them are elliptic; that is, the Jacobian matrix has purely imaginary eigenvalues. The other half are saddle points; that is, the Jacobian matrix has real eigenvalues15,16. We focus one of them at ((x,y)=(0.5,0.5)), which is depicted by the red point in each panel of Fig. 2b as a reference. The material line around the fixed point is stretched along the unstable eigendirections when the fixed point is a saddle, whereas it is folded (approximately (pi /2) rotation) when the fixed point is elliptic. The local stretching and folding around the eight fixed points occur simultaneously, resulting in efficient mixing. The use of the specific protocol by the trained mixer with the constant angular frequency, (theta (t)=omega ^{*} t), is explained in the section of “Conclusion and discussion” section.
Remarkably, the period of the flow in the middle stage, (2pi /omega ^*), which determines the period of the successive switching of the saddle and elliptic types of the fixed points, is optimal in the following sense. Apart from the RL algorithm, we conduct numerical simulations of the scalar field advected by the flow determined by (theta (t)= omega t) with a constant angular frequency, (omega ), throughout the mixing process, (0le t le 1). The inset of Fig. 2a shows (Phi (t=1)) evaluated for (omega in [0,30]). The minimum of (Phi (t=1)) in this setting is obtained at (omega simeq omega ^{*}). This implies that the RL algorithm determines the optimal angular frequency, (omega ^{*}), without any prior knowledge, and the trained mixer uses the temporally periodic flow with the optimal period in the middle stage of the process.
To characterize the flow by the trained mixer in the initial and middle stages, we introduce three different mixing processes, called randomized mixers:
Completely randomized mixer: It uses the random controller that takes one of the three actions, (omega in { 0, omega _{+}, omega _{-}}), independently, with the same probabilities for all the stages ((0 le t le 1)).
Partially randomized mixer I: It uses the trained mixer for the initial stage ((0 le t < 0.3)), and then switches to use the random controller for (0.3 le t le 1).
Partially randomized mixer II: It uses the trained mixer for the initial and middle stages ((0 le t < 0.7)), and then switches to use the random controller for (0.7 le t le 1).
Numerical simulations are conducted 200 times independently for each control. Fig. 2c presents the probability density functions (PDFs) of the mix-variance, (Phi (t=1)), at the end of the mixing process. The gray solid line indicates the value of the mix-variance by the trained mixer, (Phi _{n}(t=1)~(n=4000)) (see Fig. S1 in the “Supplementary Information” for the related PDF of the trained mixer).
The top panel of Fig. 2c depicts the PDF in the case of the completely randomized mixer, where the mix-variances are larger than the reference value of the trained mixer. The left and right panels of Fig. 2d represent the final state of the scalar field, (c(x,t=1)), produced by the trained mixer and one completely randomized mixer that presents the mix-variance, (Phi (t=1)), close to the median value of the PDF. Videos 1 and 2 in the “Supplementary Information” correspond to the scalar fields mixed by the trained mixer and the completely randomized mixer, respectively. Large unmixed blobs remain in the scalar field produced by the completely randomized mixer. That is, the training mixer with the RL algorithm is effective. The second panel of Fig. 2c depicts the PDF in the case of the partially randomized mixer I, which is more effective than the completely randomized mixer. However, a substantial gap exists between the results of the partially randomized mixer I and that of the trained mixer. This indicates that the mixing process during the middle stage is also crucial. Finally, the third panel of Fig. 2c depicts the PDF produced by the partially randomized mixer II. The results are almost identical to those obtained using the trained mixer. Therefore, the effectiveness of the partially randomized mixer II is the same as that of the trained mixer. These observations demonstrate that the mixing process during the initial and middle stages is essential for the mixing efficiency, whereas the mixing process during the final stage is not.
This subsection considers the diffusion effect on the RL-optimization of the mixing described by the advection–diffusion equations (Eq. 1) with finite Péclet numbers. The details of the problem settings are identical to those in the previous sections, except for the values of the Péclet numbers. The RL-based optimization is applied to the mixing problem for the case of (text {Pe}=10^2, 10^3), and (10^4), which are as effective as for the case of (text {Pe}=infty ), regardless of the Péclet numbers. For instance, at (text {Pe}=100), the mix-variance, (Phi _{n}(t)), decreases faster for the later episodes, as shown in the inset of Fig. 3b, where (n=1,600,1200,1800,2400), and 3000 and lighter (thicker) curves correspond to larger n. We note that the curves of (Phi _{n}(t)) for (nge 1200) are almost the same, implying that the RL algorithm converges to find the optimal policy at (n=1200). Interestingly, this convergence is faster than the case of (text {Pe}=infty ) (Fig. 1c). The number of episodes required for convergence is (n simeq 3000) at (text {Pe}=infty ); however, (n simeq 1200) seems to be sufficient for convergence around (text {Pe}=100).
Mix-variance (Phi (t)) for (0 le t le 1) at (a) (text {Pe} =infty ) and at (b) (text {Pe} =100). The thick blue and thin red lines represent the results of the mixer trained at (text {Pe}_T =infty ) and (text {Pe}_T =100), respectively. The solid, dashed, and dashed-dotted lines correspond to the results with different random numbers used for the learning. Inset: the mix-variance (Phi _{n}(t)) at (text {Pe} =100) and (text {Pe}_{T} = 100) for the n-th episode: (n=1, 600, 1200, 1800, 2400), and 3000, where lighter (thicker) curves correspond to larger n.
The diffusion effect appears in the flow controls at the later stages. If the mixer successfully generates fine layered structures in an early stage, the flow control becomes less important in the later stages of mixing due to the diffusion effect. In other words, at a low Péclet number, once the RL algorithm finds the optimal mixing control in an early stage of mixing, nothing is to be learned as the diffusion reduces the mix-variance rapidly, irrespective of the control by the mixer. This may result in the faster convergence observed above. The implications of the fast convergence at the low Péclet numbers to the training mixer are given in the “Conclusion and discussion” section.
This diffusion effect implies the asymmetric transferability of a trained mixer; that is, a mixer trained at a high Péclet number can be used for mixing at a lower Péclet number, whereas the converse does not hold true. Let (text {Pe}_T) be the Péclet number where the mixer is trained, and the asymmetric transferability is then rephrased as follows: the trained mixer can be reused for the same mixing process for the range of ((0,text {Pe}_T]). Figure 3a presents the mix-variance, (Phi (t)), for (0 le t le 1) at (text {Pe} =infty ). The thick blue lines indicate the results for the case of (text {Pe}_T =infty ), and the thin red lines indicate the results for the case of (text {Pe}_T =100). In Fig. 3a,b, the solid, dashed, and dashed-dotted lines indicate the results with different random numbers for learning. The mixers that trained at (text {Pe}_T =infty ) realize the exponentially fast mixing for the entire process when we use it for (text {Pe}=infty ). On the other hand, the mixers that trained at (text {Pe}_T =100) realize the exponentially fast mixing only for the first half of the process, but fail to mix for the latter half.
Fig. 3b presents the mix-variance, (Phi (t)), for (0 le t le 1) at (text {Pe} =100). Similar to Fig. 3a, the thick blue lines represent the results for the case of (text {Pe}_T =infty ), and the thin red lines represent the results for the case of (text {Pe}_T =100). Unlike the case of (text {Pe}=infty ), no significant difference exists between the results for the cases of (text {Pe}_T =100) and (text {Pe}_T =infty ), and both cases realize the exponentially fast mixing. In summary, the mixers of (text {Pe}_T=infty ) can be used for the mixing at (text {Pe}=100), whereas the converse does not hold true. Therefore, a mixer trained at a higher Péclet number can be used for the mixing process for a broader range of (text {Pe}).
By illustrating why the RL algorithm is suitable for fluid mixing optimization, we demonstrated, as a proof-of-concept, that the mixer trained by using the RL algorithm is effective for the two-dimensional fluid mixing problem (Fig. 1), which paves the way for the development of RL-based training of mixers. The proposed method was quantitatively evaluated by focusing on the benchmark problem of the mixing optimization studied in the pioneer work2. In addition to the comparison of mix-variance values, we note that our RL-based method solves the optimization problem in more restrictive conditions compared to the method proposed by Mathew et al.2. For example, in our setting, the number of the states of the velocity field is restricted to eight, (theta = 0,pi /4, ldots , 7pi /4). Furthermore, the proposed method is more flexible; that is, it uses only the scalar and velocity field as the input to the neural network. Provided that these fields can be observed, physical implementations are possible in principle, even if the evolution equations of these fields are unknown. For instance, mixing problems of granular or viscoelastic fluids are essential; however, the evolution equation of such a complex material has not necessarily been established, and therefore, the conjugate gradient descent method2 cannot be applied to these industrially fundamental problems. On the other hand, the RL based method is equation-free, hence it is applicable if the sensory data of the mixture states are available as the input to the neural network.
The optimized mixing process was divided into three distinct stages. It is particularly interesting to note that, in the middle stage, the optimized flow is temporally periodic with the constant angular frequency. Here, we discuss why the RL algorithm makes the angular frequency constant. The fixed points in both velocity fields, (u_{1}) and (u_{2}), are located at the same position and are homogeneously placed in the domain, (mathbb {T}^{2}). If the angular frequency is not constant, the switching period between the saddle and elliptic types of the fixed point may differ at each location. This spatial difference makes the scalar field inhomogeneous. The inhomogeneity increases the amplitude of the Fourier coefficient of the small wavenumber, thereby increasing the mix-variance. Consequently, the time variation of the angular frequency results in the larger value of the mix-variance. The RL algorithm employs the constant angular frequency to avoid this undesired effect. The detailed justification of the aforementioned interpretation is one of the future works.
Another related future work is to understand the optimal mixing in more detail. For instance, we claim that the random variation of the flow parameter in the final stage ((t>0.7)) is not essential for optimal mixing, in the sense that the results of the partially randomized mixer II (Fig. 2c) and the trained mixer (Fig. S1 in the “Supplementary Information”) are almost identical. However, there is a small difference between these PDFs, which suggests that the randomization of the actions in the final stage may eliminate some actions, which the RL algorithm regards to be essential, in the optimized mixing process.
For practical application, reduction of learning costs is crucial. Despite the effectiveness of transfer learning in reducing learning cost, its application to the problems in fluid mechanics remains limited19. In this regard, this study has introduced the physically-reasonable notion of the asymmetric transferability of the trained mixer. The demonstration in this study (Fig. 3) indicates that, in terms of transfer learning, the Péclet number of the source domain (text {Pe}_{T}) should be as high as possible, if the trained mixer is required to reuse for the broader range. If the mixer is trained at a high Péclet number, it can learn how to mix the scalar field to create the fine striped structures. If the trained mixer is transferred to a lower Péclet number, it makes the fine structures at the beginning of the mixing process. Then, smoothing such structures by diffusion reduces the mix-variance, irrespective of the actions of the trained mixer in the later stage. Therefore, the transfer of the trained mixer from a high Péclet number to a lower one is effective.
Concerning another aspect of learning costs, we have found that the learning of mixing at a lower Péclet number converges faster (Inset of Fig. 3b). Therefore, if fast learning at a Péclet number is required, the Péclet number of the source domain (text {Pe}_{T}) should be as low as possible. Considering together with the discussion in the previous paragraph, the above discussions suggest a trade-off between broad transferability and fast learning; in other words, there is an optimal Péclet number of the source domain that balances these two advantages in each application. Although this study is restricted to the transfer of the trained mixer over the different Péclet numbers, future developments of transfer learning methods of trained mixers may be significant.
Large gaps exist between the mathematical toy problem discussed in this study and the existing mixing problems in the industrial processes. However, the results of this study indicate some directions overcoming these gaps. First, we discuss the implications of this study to turbulent mixing. Turbulence comprises multi-scale counter-rotating pairs of coherent vortices20, and strong turbulent mixing stems from the effective mixing around such pairs of vortices at each scale1. As observed in the transfer learning method, scalar mixing occurs from larger to smaller scales. Since the time scale of turbulent mixing is shorter for smaller scales, the total mixing efficiency is determined by the mixing at the largest scale. Thus, measuring the velocity and scalar field at the largest scale may be sufficient for the proposed training method. Despite the significant gap between laminar and turbulent mixing, the insights from the present study will be useful for training mixers with turbulent flows.
In addition, in industry, the multiphase and/or thermal flows with chemical reactions may have to be considered, which increases the complexity of the flow dynamics. In such cases, embedding prior knowledge, such as the evolution equations or some physical constraints into the RL-based optimization may be effective, as discussed in Brunton11. As another future task for the RL-based optimization in industrial mixing problems, it will be important to study the robustness of the mixing control with the obtained policy with respect to changes in the initial scalar field. Furthermore, whereas the deep Q network is employed as the first step in this study, a more specific and state-of-the-art implementation of the RL algorithm would be necessary for such complex flows. Extending the proposed method to incorporate knowledge on fluid mechanics and suitable RL implementation techniques can further enhance mixing even in industrial processes with laminar and turbulent flows.
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Goto, S. & Kida, S. Reynolds-number dependence of line and surface stretching in turbulence: Folding effects. J. Fluid Mech. 586, 59–81 (2007).
ADS MathSciNet Article Google Scholar
Mathew, G., Mezic, I., Grivopoulos, S., Vaidya, U. & Petzold, L. Optimal control of mixing in stokes fluid flows. J. Fluid Mech. 580, 261–281 (2007).
ADS MathSciNet CAS Article Google Scholar
Lin, Z., Thiffeault, J.-L. & Doering, C. R. Optimal stirring strategies for passive scalar mixing. J. Fluid Mech. 675, 465–476 (2011).
ADS MathSciNet CAS Article Google Scholar
Seis, C. Maximal mixing by incompressible fluid flows. Nonlinearity 26, 3279–3289 (2013).
ADS MathSciNet Article Google Scholar
Alberti, G., Crippa, G. & Mazzucato, A. L. Exponential self-similar mixing by incompressible flows. J. Am. Math. Soc. 32, 445–490 (2019).
MathSciNet Article Google Scholar
Szepesvàri, C. Algorithms for Reinforcement Learning. (Synthesis Lectures on Artificial Intelligence and Machine Learning Morgan and Claypool Publishers, 2010)
Sutton, R. S. & Barto, A. G. Reinforcement Learning: An Introduction (MIT Press, 2018).
MATH Google Scholar
Villermaux, E. Mixing versus stirring. Ann. Rev. Fluid Mech. 51, 245–273 (2019).
ADS MathSciNet Article Google Scholar
Garnier, P. et al. A review on deep reinforcement learning for fluid mechanics. Comput. Fluids 225, 104973–104996 (2021).
MathSciNet Article Google Scholar
Brunton, S. L., Noack, B. R. & Koumoutsakos, P. Machine learning for fluid mechanics. Ann. Rev. Fluid Mech. 52, 477–508 (2020).
ADS Article Google Scholar
Brunton, S. L. Applying machine learning to study fluid mechanics. Acta Mechanica Sinica. 37, 1718–1726. https://doi.org/10.1007/s10409-021-01143-6 (2021).
ADS Article Google Scholar
Degrave, J. et al. Magnetic control of tokamak plasmas through deep reinforcement learning. Nature 602, 414–419 (2022).
ADS CAS Article Google Scholar
Novati, G., de Laroussilhe, H. L. & Koumoutsakos, P. Automating turbulence modelling by multi-agent reinforcement learning. Nat. Mach. Intell. 3, 87–96 (2021).
Article Google Scholar
Rothstein, D., Henry, E. & Gollub, J. Persistent patterns in transient chaotic fluid mixing. Nature 401, 770–772 (1999).
ADS CAS Article Google Scholar
Wiggins, S. & Ottino, J. M. Foundations of chaotic mixing. Philos. Trans. R. Soc. Lond. Series A Math. Phys. Eng. Sci. 362, 937–970 (2004).
ADS MathSciNet Article Google Scholar
Aref, H. et al. Frontiers of chaotic advection. Rev. Modern Phys. 89, 025007 (2017).
ADS MathSciNet Article Google Scholar
Mathew, G., Mezic, I. & Petzold, L. A multiscale measure for mixing. Physica D Nonlinear Phenomena 211, 23–46 (2005).
ADS MathSciNet CAS Article Google Scholar
Mnih, V. et al. Human-level control through deep reinforcement learning. Nature 518, 529–533 (2015).
ADS CAS Article Google Scholar
Inubushi, M. & Goto, S. Transfer learning for nonlinear dynamics and its application to fluid turbulence. Phys. Rev. E. 102, 043301(8) (2020).
ADS Article Google Scholar
Goto, S., Saito, Y. & Kawahara, G. Hierarchy of antiparallel vortex tubes in spatially periodic turbulence at high Reynolds numbers. Phys. Rev. Fluids 2, 064603 (2017).
ADS Article Google Scholar
Download references
This work was partially supported by JSPS Grant-in-Aid for Early-Career Scientists No. 19K14591 and JSPS Grants-in-Aid for Scientific Research Nos. 19KK0067, 20H02068, 20K20973, and 22K03420.
These authors contributed equally: Mikito Konishi and Susumu Goto.
Graduate School of Engineering Science, Osaka University, Osaka, 560-8531, Japan
Mikito Konishi, Masanobu Inubushi & Susumu Goto
Department of Applied Mathematics, Tokyo University of Science, Tokyo, 162-8601, Japan
Masanobu Inubushi
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
M.K. and M.I. conceived and conducted the numerical experiments; M.K., M.I, and S.G. analyzed the results. All authors reviewed the manuscript.
Correspondence to Masanobu Inubushi.
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Reprints and Permissions
Konishi, M., Inubushi, M. & Goto, S. Fluid mixing optimization with reinforcement learning. Sci Rep 12, 14268 (2022). https://doi.org/10.1038/s41598-022-18037-7
Download citation
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-022-18037-7
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
Advertisement
Advanced search
© 2022 Springer Nature Limited
Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.
Music Production Software Market: USD 369.87 million Growth from 2021 to 2026, High Competition Among Vendors – Technavio – PR Newswire
Searching for your content…
In-Language News
Contact Us
888-776-0942
from 8 AM – 10 PM ET
News provided by
Aug 22, 2022, 03:15 ET
Share this article
NEW YORK , Aug. 22, 2022 /PRNewswire/ — Music production software is also known as digital audio workstation (DAW). This software is used by musicians, artists, disc jockeys (DJs), and music production companies to perform multiple functions such as composing, recording, editing, mixing, and mastering to produce music.
Technavio estimates that the music production software market size is expected to grow by USD 369.87 million from 2021 to 2026. The growth momentum of the market will accelerate at a CAGR of 7.31% during the forecast period.
Get a comprehensive report summary describing the market size and forecast along with research methodology. The FREE sample report is available in PDF format
The market is characterized by low capital requirements and moderate product differentiation. Therefore, it is easy for new players to enter the market. However, the strong brand image of the existing vendors and their wide geographic presence makes it difficult for new vendors to gain a market share. Hence, the threat of new entrants was moderate in 2021, and it is expected to remain the same during the forecast period.
The Report offers information on all Porter’s Five Forces. Buy Now to gain insights that will help you improve your business.
The music production software market is fragmented, and the vendors are deploying organic and inorganic growth strategies to compete in the market. The market is in its growth stage and is expected to grow significantly during the forecast period. Owing to significant growth opportunities in this market, the competition will intensify with new product launches, technological innovations, and partnerships. Established vendors have increased their focus on the development and introduction of music production software as well as updating the old versions. Moreover, the growing number of musicians, artists, DJs, and music production companies is accelerating the demand for music production software. Some of the key vendors operating in the market include Ableton AG, Acon AS, Acoustica Inc., Adobe Inc., and Apple Inc.
The report provides a full list of key vendors, their strategies, and the latest developments. Request a FREE PDF Sample Now
The music production software market has been segmented by type (editing, mixing, and recording), end-user (professionals and non-professionals), and geography (North America, Europe, APAC, South America, and the Middle East and Africa)
By end-user, the professional segment will account for the highest market growth. These advertisements have a stronger audio-visual effect, which gives viewers a positive product impression. The demand for music production software is rising, owing to the growing audio and video content.
By geography, the market has been segmented into North America, Europe, APAC, South America, and the Middle East and Africa. North America will have the highest market growth during the forecast period. The growth of this region is attributed to factors such as a significant increase in the adoption of music production software by professionals such as musicians and music production companies. Moreover, market growth in this region will be faster than the growth of the market in other regions. The US is a crucial country for the music production software market in North America.
Learn about the contribution of each segment summarized in concise infographics and thorough descriptions. View a FREE PDF Sample Report
The increase in the adoption of DJ software to produce music will drive the growth of the music production software market during the forecast period. The rising interest in music is providing several opportunities for DJ artists to give live performances. This will drive the adoption of music production software to compose, edit, record, and produce new songs and update old songs by adding new lyrics or music.
The growing demand for digital audio content is a trend in the market. Digital audio content, including music and podcasts, is gaining popularity worldwide. The rising adoption of electronic devices such as smartphones, smartwatches, laptops, and tablets has helped people consume more digital audio content. Apple Music, Amazon Music, YouTube Music, and Spotify are some of the major music streaming platforms.
Apart from the growing demand for digital audio content, many other trends are also shaping the future of the market. View FREE PDF Sample Report to learn about these trends
Recommendation Engine Market by End-user and Geography – Forecast and Analysis 2022-2026
Conversational Computing Platform Market by Type and Geography – Forecast and Analysis 2022-2026
Music Production Software Market Scope
Report Coverage
Details
Page number
120
Base year
2021
Forecast period
2022-2026
Growth momentum & CAGR
Accelerate at a CAGR of 7.31%
Market growth 2022-2026
USD 369.87 million
Market structure
Fragmented
YoY growth (%)
6.52
Regional analysis
North America, Europe, APAC, South America, and Middle East and Africa
Performing market contribution
North America at 42%
Key consumer countries
US, China, UK, Germany, and France
Competitive landscape
Leading companies, Competitive strategies, Consumer engagement scope
Key companies profiled
Ableton AG, Acon AS, Acoustica Inc., Adobe Inc., Apple Inc., Avid Technology Inc., Bitwig GmbH, Cockos Inc., GoldWave Inc., Image Line Software NV, iZotope Inc., MAGIX Software GmbH, MakeMusic Inc., MOTU Inc., MuTools, NCH SOFTWARE Pty Ltd., PreSonus Audio Electronics Inc., Reason Studios AB, Serato Ltd., and Yamaha Corp.
Market dynamics
Parent market analysis, market growth inducers and obstacles, fast-growing and slow-growing segment analysis, COVID-19 impact and recovery analysis and future consumer dynamics, and Market condition analysis for the forecast period.
Customization purview
If our report has not included the data that you are looking for, you can reach out to our analysts and get segments customized.
Browse Information Technology Market Reports
1 Executive Summary
2 Market Landscape
3 Market Sizing
4 Five Forces Analysis
5 Market Segmentation by End-user
6 Market Segmentation by Type
7 Customer Landscape
8 Geographic Landscape
9 Drivers, Challenges, and Trends
10 Vendor Landscape
11 Vendor Analysis
12 Appendix
Technavio is a leading global technology research and advisory company. Their research and analysis focus on emerging market trends and provide actionable insights to help businesses identify market opportunities and develop effective strategies to optimize their market positions. With over 500 specialized analysts, Technavio’s report library consists of more than 17,000 reports and counting, covering 800 technologies, spanning across 50 countries. Their client base consists of enterprises of all sizes, including more than 100 Fortune 500 companies. This growing client base relies on Technavio’s comprehensive coverage, extensive research, and actionable market insights to identify opportunities in existing and potential markets and assess their competitive positions within changing market scenarios.
Technavio Research
Jesse Maida
Media & Marketing Executive
US: +1 844 364 1100
UK: +44 203 893 3200
Email: [email protected]
Website: www.technavio.com/
SOURCE Technavio
Cision Distribution 888-776-0942
from 8 AM – 9 PM ET
Ultra-strong spin–orbit coupling and topological moiré engineering in twisted ZrS2 bilayers – Nature.com
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Nature Communications volume 13, Article number: 4915 (2022)
Metrics details
We predict that twisted bilayers of 1T-ZrS2 realize a novel and tunable platform to engineer two-dimensional topological quantum phases dominated by strong spin-orbit interactions. At small twist angles, ZrS2 heterostructures give rise to an emergent and twist-controlled moiré Kagome lattice, combining geometric frustration and strong spin-orbit coupling to give rise to a moiré quantum spin Hall insulator with highly controllable and nearly-dispersionless bands. We devise a generic pseudo-spin theory for group-IV transition metal dichalcogenides that relies on the two-component character of the valence band maximum of the 1T structure at Γ, and study the emergence of a robust quantum anomalous Hall phase as well as possible fractional Chern insulating states from strong Coulomb repulsion at fractional fillings of the topological moiré Kagome bands. Our results establish group-IV transition metal dichalcogenide bilayers as a novel moiré platform to realize strongly-correlated topological phases in a twist-tunable setting.
Twisted van der Waals heterostructures have recently emerged as an intriguing and highly tunable platform to realize unconventional electronic phases in two dimensions1,2,3,4. Spurred by the discovery of Mott insulation and superconductivity in twisted bilayer graphene5,6, remarkable progress in fabrication and twist-angle control has led to observations of correlated insulating states or superconductivity in a variety of materials, including trilayer and double-bilayer graphene, homo- and hetero-bilayers of twisted transition metal dichalcogenides (TMDs)7,8,9,10,11,12,13,14,15,16,17,18, and heterostructures at a twist on hexagonal boron nitride substrates19,20. At its heart, this rich phenomenology stems from electronic interference effects due to the moiré superlattice, which can selectively quench kinetic energy scales to realize almost dispersionless bands, permitting a twist angle controlled realization of regimes dominated by strong electronic interactions. At the same time, the drastic reduction of kinetic energy of the low-energy moiré bands implies straightforward gate-tunable access to a wide range of filling fractions, permitting wide-ranging experimental access to the phase diagrams of paradigmatic models of strongly-correlated electrons2. Consequently, the putative realization of strongly-correlated electron physics in a tunable setting has garnered significant attention, resulting in growing experimental evidence for novel correlated phases, including unconventional superconductivity4,21,22.
Notably, and despite negligible intrinsic spin–orbit coupling in graphene, these were found to include topological states of matter. Here, the realization of the interaction-induced quantum anomalous Hall effect without external magnetic fields in twisted bilayer23,24,25 and trilayer26 graphene has spurred numerous proposals for more exotic fractionalized topological states of matter27,28,29,30, which however rely on a delicate interplay of spontaneous ferromagnetic order, valley polarization, and substrate engineering effects to induce the requisite nontrivial band topology. Generalizations to twisted transition-metal dichalcogenides have focused on telluride-based group VI compounds with 2H structure in the monolayer which exhibit an intrinsic quantum spin Hall effect31, with the quantum anomalous Hall effect recently observed32 and similarly expected to emerge from spontaneous valley polarization33,34.
Central to the present work, we demonstrate for twisted bilayer ZrS2 with 1T structure that the paradigm of twist-controlled suppression of the bare kinetic energy scales can be straightforwardly extended to instead promote spin–orbit coupling to constitute the dominant energy scale at low energies, opening up a new and exotic regime for experimental and theoretical investigation. Remarkably, we find that the two-component character of the valence band maximum in such two-dimensional group IV transition metal dichalcogenides enters in an essential manner, leading to the emergence of a clean moiré Kagome lattice with almost dispersionless quantum spin Hall bands at small twist angles. We demonstrate that this tunable realization of a ZrS2 moiré heterostructure with strong spin-orbit coupling and strong interactions can therefore provide a robust and novel platform to probe the profound interplay of non-trivial band topology and electronic correlations, and shed light on elusive quantum phases beyond the purview of conventional condensed matter systems.
ZrS2 is a group IV transition metal dichalcogenide with an exfoliable layered structure. Different from group VI TMDs such as Mo2 and WS2 that normally adopt a 2H layered structure, ZrS2 has a stable 1T structure35 in its ground state as shown in Fig. 1a, without distorting into the 1T’ structure36,37. Bulk ZrS2 is a semiconductor with a band gap of 1.80 eV38,39 and it remains semiconducting when thinned down to the monolayer35,40. In contrast to group-VI transition metal dichalcogenides such as MoS2 with 2H structure, the valence band maximum in ZrS2 and other group-IV transition metal dichalcogenides is located at Γ already in the monolayer and is composed of twofold degenerate chalcogen px, py orbitals. Spin–orbit coupling lifts their degeneracy and introduces a ~100 meV gap [Fig. 1b]. This property readily carries over to aligned bilayers with symmetric AA and AB stacking configurations [Fig. 1c, d]; here, the valence band maximum at Γ follows from antibonding combinations of the out-of-plane chalcogen px, py orbitals. These are energetically separated from bonding combinations by ~80−100 meV [Fig. 1c, d], with a secondary local valence band maximum of pz orbitals furthermore located close to Γ and similarly detuned by ~50 meV for AA stacking.
a Top and side view of the 1T structure of ZrS2. b The band structure of ZrS2 monolayer hosts a valence band maximum at Γ composed of the px, py chalcogen orbital J = ± 3/2 states, with a spin-orbit splitting of ~ 100 meV. c, d In aligned bilayers with AA or AB stacking, the valence band maximum at Γ (composed of of antibonding states of px, py symmetry) is energetically separated by ~ 50meV from bonding px, py states and pz orbitals. Hence, antibonding px, py states compose the top-most moiré valence bands at small twist angles. e In small-angle twisted heterostructures of ZrS2, the interlayer alignment interpolates from AA to AB/BA stacking as a function of position in the moiré unit cell. The px, py antibonding states remain degenerate for both AA and AB/BA stacking in the absence of spin-orbit coupling. Their overall energetic shift is encoded in f as an effective scalar moiré superlattice potential, with minima that form a moiré honeycomb lattice. g Crucially, the loss of rotational symmetry away from local AA, AB stacking further splits the px, py states. This orbital splitting contribution to the moiré potential is maximal in three “domain wall” regions “X” in the moiré unit cell [(e)]; upon exceeding the scalar potential, electrons in the top-most moiré valence band form an emergent moiré Kagome lattice [dashed lines in e; guide to the eye].
In twisted bilayers, the atomic interlayer registry interpolates continuously between local AA, AB, and BA alignment as a function of position and a moiré pattern with three-fold rotation symmetry forms [Fig. 1e]. At sufficiently small twist angles, the energetic considerations for aligned bilayers discussed above immediately suggest that the top-most (highest energy) moiré valence bands should be similarly composed of antibonding px, py chalcogen orbitals. If spin-orbit coupling is neglected, these are degenerate at Γ in both AA and AB regions [Fig. 1e] by virtue of rotation symmetry. However, the valence band edge differs between the two stackings, with the smooth interpolation between local alignments in the moiré unit cell encoded in an effective periodic scalar moiré potential V(r) [Fig. 1f]. Minima of V(r) is located at the AB and BA regions and form an effective honeycomb lattice. Notably, for the purposes of capturing the highest-energy moiré valence bands, the potential retains to an excellent approximation of the full sixfold rotation and mirror symmetries of the monolayer, even though the macroscopic crystal is chiral. This situation is in principle analogous to twisted bilayer MoS241,42, which hosts a series of almost dispersionless bands of Mo ({d}_{{z}^{2}}) orbital character on an emergent moiré honeycomb lattice.
Crucially, the loss of rotational symmetry away from local AA, AB stacking lifts the orbital degeneracy between px, py antibonding orbitals (in the absence of spin-orbit coupling), introducing a second energy scale into the problem. In stark contrast to 2H TMD bilayers, the two-component character of the Γ valley states enters in an essential manner. From symmetry considerations, their orbital splitting is expected to be maximal in three “domain wall” regions “X” per moiré unit cell in Fig. 1e, in which the transition-metal atoms of both layers form “stripes”, with the rotational symmetry of the local stacking order reduced to C2. Contrary to the scalar moiré potential, maxima of orbital px, py splitting hence form a Kagome pattern [Fig. 1g]. Remarkably, if the resulting energetic gain exceeds the scalar potential V(r), it becomes favorable for charge to migrate from the honeycomb AB/BA regions to “X” regions, realizing an emergent Kagome lattice of s-like moiré orbitals in a highly-tunable setting [Fig. 1e].
A minimal continuum model of this scenario readily follows from the above symmetry considerations as
where ({hat{H}}_{0}) describes the two-fold degenerate antibonding px, py chalcogen states
with the orbital degree of freedom represented via Pauli matrices (hat{{{{{{{{boldsymbol{tau }}}}}}}}}). Here, m⋆ denotes the effective average band mass, and (eta =frac{{m}_{+}-{m}_{-}}{{m}_{+}+{m}_{-}}) parametrizes the ratio of light (m−) and heavy (m+) hole p bands at Γ. Atomic spin–orbit interactions
lift the orbital degeneracy, opening up a gap at Γ as discussed in detail below. Here, ({hat{{{{{{{{boldsymbol{sigma }}}}}}}}}}_{z}) acts on spin. Central to the emergence of the Kagome lattice, the moiré potential acts nontrivially on the orbital pseudospin, and can generically be written as a Fourier expansion
Here, n indexes the n-th moiré Brillouin zone. Vn parameterizes the Fourier modes of the scalar potential in direct analogy to twisted WS242, with ({f}_{n}^{(0)}({{{{{{{bf{r}}}}}}}})=cos ({{{{{{{{bf{b}}}}}}}}}_{n,1}{{{{{{{bf{r}}}}}}}})+cos ({{{{{{{{bf{b}}}}}}}}}_{n,2}{{{{{{{bf{r}}}}}}}})+cos ({{{{{{{{bf{b}}}}}}}}}_{n,3}{{{{{{{bf{r}}}}}}}})) chosen to retain the full sixfold rotation symmetry and bn,i describing the three reciprocal lattice vectors i = 1, 2, 3 (related via C3 rotations) to the n-th Brillouin zone.
The pseudospin ({hat{{{{{{{{boldsymbol{tau }}}}}}}}}}_{x}), ({hat{{{{{{{{boldsymbol{tau }}}}}}}}}}_{z}) contributions to the potential are related in the presence of (approximate) mirror symmetry, with ({f}_{n}^{(x)}({{{{{{{bf{r}}}}}}}})=-frac{sqrt{3}}{2}cos ({{{{{{{{bf{b}}}}}}}}}_{n,1}{{{{{{{bf{r}}}}}}}})+frac{sqrt{3}}{2}cos ({{{{{{{{bf{b}}}}}}}}}_{n,3}{{{{{{{bf{r}}}}}}}})) and ({f}_{n}^{(z)}({{{{{{{bf{r}}}}}}}})=frac{1}{2}cos ({{{{{{{{bf{b}}}}}}}}}_{n,1}{{{{{{{bf{r}}}}}}}})-cos ({{{{{{{{bf{b}}}}}}}}}_{n,2}{{{{{{{bf{r}}}}}}}});+frac{1}{2}cos ({{{{{{{{bf{b}}}}}}}}}_{n,3}{{{{{{{bf{r}}}}}}}})). The salient physics is encoded already in the lowest harmonic—with ({{{{{{{{bf{b}}}}}}}}}_{1,1}=[2pi ,-2pi /sqrt{3}]/{a}_{0}), ({{{{{{{{bf{b}}}}}}}}}_{1,2}=[0,4pi /sqrt{3}]/{a}_{0}), ({{{{{{{{bf{b}}}}}}}}}_{1,3}=[2pi ,2pi /sqrt{3}]/{a}_{0}) and a0 the moiré lattice length, the scalar potential hosts two minima in the AB and BA regions at ({{{{{{{bf{r}}}}}}}}=[1/2,pm 1/(2sqrt{3})]). Conversely, the pseudospin potential that determines the splitting of px, py orbitals has three maxima in the Kagome X regions at ({{{{{{{bf{r}}}}}}}}=[1/2,0]{a}_{0},,[1/4,sqrt{3}/4]{a}_{0},,[3/4,sqrt{3}/4]{a}_{0}). Up to an overall energy scale, a minimal continuum model that includes only the first harmonic will therefore depend on just three dimensionless parameters η, (2V{m}^{star }{a}_{0}^{2}/{hslash }^{2}), (2{V}^{prime}{m}^{star }{a}_{0}^{2}/{hslash }^{2}), where a0 ∝ θ−1 scales with the twist angle. Local lattice relaxation effects are encoded in the higher harmonics of the potential; as the local stacking of AB and BA regions arise energetic favorable, lattice relaxation results in large domains with almost uniform AB or BA stacking [see domains highlighted with red and purple dashed lines in Fig. 1e] with the salient stacking variation in the X regions. These parameters can be obtained by fitting the band structure obtained from DFT calculations as described below.
Figure 2a depicts the structure of the resulting moiré bands without spin-orbit coupling, as a function of scalar V ≡ V1 and pseudospin ({V}^{prime}equiv {V}_{1}^{prime}) potentials. For ({V}^{prime}=0), the scalar potential V localizes the hole charge density on a honeycomb lattice of AB/BA regions [Fig. 2a, left column; Fig. 2b (I)], and an energetically well-separated set of honeycomb bands with Dirac points at K, ({{{{{{{bf{K}}}}}}}}^{prime}) emerges at the top of the valence band. These retain a twofold orbital px, py character, with the degeneracy of the bands weakly broken due to orbital anisotropy η ≠ 0. This directly mirrors the low-energy band structure of twisted bilayer graphene, however with the two-orbital structure resulting from the px, py degeneracy of the constituent states at Γ as opposed to a valley degeneracy.
a Anatomy of the moiré band structure in the two-orbital pseudospin continuum theory, as a function of the scalar (V) and pseudospin (({V}^{prime})) potential, with η = 0.15 and energies normalized to (hslash /2{m}^{star }{a}_{0}^{2}equiv 1). In the absence of ({V}^{prime}), a well-separated honeycomb lattice with two p orbitals per sublattice emerges with increasing V [bottom left], directly analogous to twisted bilayer graphene. b The calculated charge density of bands (I) is localized in a honeycomb pattern of AB/BA regions; however, the next lower-energy band (II) already exhibits a Kagome charge density pattern. Conversely, the pseudospin moiré potential ({V}^{prime}) favors charge patterns localized in (III) the Kagome “X” regions as well as on rings (IV) around the Kagome hexagons. As a0 ~ θ−1, an energetically well-separated Kagome moiré band structure emerges at sufficiently small twist angles [a, right column, marked in blue]. c Spin–orbit interactions lift the quadratic touching of Kagome bands at Γ. The resulting band structure is gapped and realizes a novel Kagome moiré flat-band quantum spin Hall insulator with spin Chern numbers ± 1 of the top and bottom Kagome bands. d depicts the ab initio band structure of twisted bilayers for three representative twist angles with [right column] and without [left column] spin-orbit coupling, with the top-most valence bands arising from the emergent Kagome lattice. Colored lines indicate the continuum model band structure, fitted to the top valence bands [see main text]. Additional deeper pz orbital valence bands appear > 50 meV below the band edge and are not accounted for in the continuum theory, however are progressively separated energetically from the Kagome bands as the twist angle is reduced. Importantly, spin-orbit interactions split off a topological band with spin Chern number ({{{{{{{{mathcal{C}}}}}}}}}_{s}=1) [thick blue line], as discussed in the main text. e Matching charge density distributions from the pseudospin continuum theory and ab initio calculations confirm the emergent Kagome band structure.
However, already the next lower in energy (fifth) moiré valence band reveals upon closer inspection a charge density distribution with a Kagome pattern [Fig. 2b, pattern (II)], localized in the “X” regions of the moiré unit cell [Fig. 1e]. These states gain energy from a finite pseudospin potential ({V}^{prime}), which lifts them to higher energies: Beyond a critical ({V}^{prime}), the fifth “Kagome” band and the bottom px, py honeycomb bands invert their energetic ordering at Γ. Consequently, the charge density distribution of the top px, py bands shifts from AB/BA honeycomb regions to “X” Kagome sites [Fig. 2b, pattern (III)]. If the moiré potentials are sufficiently weak, the three resulting bands that constitute the emergent moiré Kagome lattice couple to a fourth moiré orbital centered on the hexagons of the lattice, with a charge density distribution that forms a ring around the AA regions of the moiré unit cell [Fig. 2b, pattern (IV)]. As the twist angle is further reduced, an energetically well-separated set of three Kagome lattice bands emerges as the top-most set of moiré valence states [Fig. 2b, patterns (V), (VI)].
The above behavior closely matches the results from large-scale ab initio calculations of the twisted moiré supercell, depicted in Fig. 2d, left column, for three representative twist angles [see Supplementary Note 1]. As the angle is reduced, a set of bands with a Kagome charge distribution at Γ splits off progressively from deeper valence bands. For the larger twist angles ≥2.28∘ that are still within computational reach for density functional calculations, this energetic separation is not yet sufficient to completely separate the Kagome bands of chalcogen antibonding px, py character from states with pz or bonding px, py character (<50 meV below the band edge), not included in the continuum theory. Nevertheless, the top-most Kagome bands of interest are already well-captured via the continuum model for the smallest twist angle [Fig. 2d, bottom-left] upon accounting only for the lowest harmonic of the moiré potential.
Crucially, the inclusion of spin-orbit coupling [Eq. (3)] now opens up a gap at the Kagome Dirac points and lifts the quadratic band touching degeneracy at Γ [Fig. 2c], reflected in ab initio simulations with spin-orbit interactions [Fig. 2d, right column]. As the top-most valence states originate from px, py orbitals at small twist angles, spin-flip spin-orbit interactions are negligible and spin-z remains a good quantum number. Remarkably, this results in three almost dispersionless moiré bands that realize a novel Kagome topological quantum spin Hall insulator with spin Chern numbers ({{{{{{{{mathcal{C}}}}}}}}}_{s}=pm !1) for the first and third flat band [Fig. 2c, d]. In marked contrast to conventional topological materials, however, while superlattice interference quenches the kinetic energy scales, spin-orbit coupling λsoc enters as a bare atomic scale and hence becomes the dominant energy scale that governs the low-energy physics of the moiré valence bands in ZrS2. This highly-tunable materials realization of an “ultra-strong” spin-orbit interaction regime in a moiré heterostructure constitutes a central result of this paper.
To model the emergent top-most flat topological moiré band in twisted ZrS2, we proceed with a fit of the pseudospin continuum theory [Eq. (1)] to the spin-orbit-coupled ab initio band structure for θ = 2.64∘ [Fig. 2d, middle-right panel]. As the minimal model of Eq. (1) does not account for bonding px, py, or pz states, the third-highest ab initio valence band (−50 meV below the valence band edge) is composed primarily of bonding px, py, and pz orbitals and is excluded from the fit. We note that this band separates energetically from the three Kagome moiré bands at lower twist angles. We obtain excellent agreement for the top two bands of px, py antibonding character using η = 0.33, m⋆ = 0.27m0, λsoc = 57 meV, V1 = 5.5 meV, ({V}_{1}^{prime}=-9.3;{{{{{{{rm{meV}}}}}}}}), V2 = 11.5 meV, ({V}_{2}^{prime}=-5.1;{{{{{{{rm{meV}}}}}}}}). Scaling with twist angle similarly matches the ab initio band structure at 2.45∘ [Fig. 2d, bottom-right panel]. As expected, the top-most band is topologically nontrivial with spin Chern number ({{{{{{{{mathcal{C}}}}}}}}}_{s}=pm !1). Figure 2e compares the corresponding charge density distributions at Γ for ab initio and continuum model calculations; both exhibit comparable Kagome patterns as well as a competing band at lower energies with a ring-shaped charge pattern around the AA region, which similarly becomes energetically separated from Kagome bands at lower twist angles [Fig. 2a].
A key advantage of the continuum theory is the possibility to study the behavior at small twist angles in a computationally feasible manner. Figure 3a depicts the bandwidth of the top-most topological moiré Kagome band, as well as the single-particle gap to the next deeper valence band, as a function of twist angle a0 ~ θ−1. The bandwidth of the top-most topological band decreases exponentially with twist angle, whereas the ratio between bandwidth and band gap saturates below ≈ 2∘ and approaches one. Below this twist angle, the three Kagome bands become fully isolated in energy from deeper valence states [Fig. 3g]. This immediately suggests a fruitful tight-binding parameterization at ultra-small angles, presuming that local lattice relaxation effects remain manageable. Results are shown in Fig. 3b for a tight-binding model depicted schematically in (c), but including up to 8th-neighbor hopping to ensure a good fit over all angles [see Supplementary Note 2]. For small angles ≪ 2∘, the top three bands become well-captured by a nearest-neighbor Kagome tight-binding model with imaginary hoppings. Third-neighbor hopping tδ through the hexagons are leading corrections to this model and follow from the elliptical shapes of the charge density distribution at the Kagome “X” sites.
a Band width and energetic separation of the top-most moiré valence band, extrapolated from the continuum theory as a function of twist angle. b Kagome tight-binding parameterization of the top three moiré bands, with real and imaginary hoppings depicted schematically in c. Shaded regions denote larger twist angles for which the third Kagome band does not remain well-isolated from lower-lying states; tight-binding parameterization in this region fits only the top two bands. d Berry curvature at 2.64∘ for the first moiré Kagome valence band, as well as e for the third band at 1.06∘. f Berry curvature fluctuations ΔΩ are suppressed as the twist angle is reduced, approaching a moiré realization of a Landau level. g Band structure of the continuum model at θ ≈ 1.06∘, 1.98∘.
The sizable imaginary nearest-neighbor hopping [Fig. 3b] is a direct consequence of the strong spin–orbit coupling limit and can be interpreted as a finite effective staggered magnetic flux through the elementary triangles of the Kagome lattice. It lifts the quadratic touching of flat and dispersive Kagome bands and opens up a gap at the Dirac points, realizing a time-reversal-invariant version of a parent model for fractional Chern insulators43,44.
The tunable realization of isolated time-reversal symmetric topological flat bands in twisted ZrS2 is an ideal starting point for the stabilization of a host of correlated topological states of matter, ranging from interaction-induced quantum anomalous Hall effects45 to elusive fractional Chern and topological insulators46,47,48,49. To investigate the role of electronic interactions and propensity for correlated topological phases in twisted ZrS2, we now study the top-most moiré Kagome band at fractional fillings and augment the effective three-band Kagome tight-binding description [Fig. 3]—derived from the continuum theory, and a continuous function of the twist angle—via a screened Coulomb repulsion. The interaction is constrained for simplicity to a local Hubbard ((U{sum }_{i}{hat{n}}_{iuparrow }{hat{n}}_{idownarrow })) and nearest-neighbor density ((({U}^{prime}/2){sum }_{ < ij > sigma {sigma }^{prime}}hat{n}_{isigma} hat{n}_{j{sigma }^{prime}})) interaction, expected to be a good approximation for screening due to metallic gates50,51. Suppose first that the top-most topological moiré Kagome band is tuned to half filling via electrostatic gating. A nontrivial spin Chern number precludes a straightforward Wannier tight-binding representation of this individual band. Instead, as deeper fully-filled valence bands are energetically separated, the low-energy behavior can be captured starting Kagome tight-binding model via projecting Coulomb interactions (U,{U}^{prime}) onto the Bloch states of a single fractionally-filled flat topological band, in direct analogy to lowest Landau level projections for the fractional quantum Hall effect. The resulting interacting problem is governed by an effective Hamiltonian
where ({hat{c}}_{{{{{{{{bf{k}}}}}}}}sigma }^{{{{dagger}}} },hat{c}_{{{{{{{{bf{k}}}}}}}}sigma}) create/annihilate electrons in the flat band with Bloch momenta k, ϵk denotes the residual band dispersion, L is the system size, and
is the Coulomb repulsion projected to the Bloch states ({u}_{{{{{{{{bf{k}}}}}}}}}^{(alpha sigma )}) of the top-most band, derived from the tight-binding model, with
Here, momenta ({{{{{{{bf{k}}}}}}}},{{{{{{{{bf{k}}}}}}}}}^{prime},{{{{{{{bf{q}}}}}}}}) are defined in the moiré Brillouin zone, ai denote the moiré lattice vectors, and α, σ denote the sublattice and spin degrees of freedom. Since a sufficiently short-ranged interaction (U ; > ;{U}^{prime}) mainly imparts a local energetic penalty for electron pairs of opposite spin occupying the same Kagome “X” sites, a flat-band ferromagnetic instability generically ensues52 at half filling of the top-most quantum spin Hall band, in direct analogy to quantum Hall ferromagnetism53. The resulting spontaneous spin-polarized state is gapped and aligned in the z direction—it exhibits a quantum anomalous Hall effect by virtue of filling a quantum spin Hall band for one spin component only; this fully spin-polarized state is an exact gapped zero-energy ground state in the absence of dispersion45 and entails a quantized Hall conductivity. To study its robustness to the finite residual dispersion of the moiré band, we evaluate the phase diagram via exact diagonalization of Eq. (5) on a 4 × 4 unit cell cluster. Figure 4a depicts the phase diagram as a function of twist angle (which parameterizes the low-energy electronic band structure and Bloch states) and interaction strength U vs bandwidth W of the top-most band. A robust quantum anomalous Hall state emerges for interactions on the order of four times the moiré bandwidth and remains robust over a wide range of twist angles. Notably, the underlying mechanism is distinct from the observed quantum anomalous Hall effect in twisted bilayer graphene, relying instead on the intrinsic topologically non-trivial moiré band structure due to strong spin-orbit coupling and obviating the necessity for concurrent valley polarization and substrate effects.
a At half filling, the first moiré band displays a ferromagnetic instability for a local Hubbard repulsion U on the order of ~4W the electronic bandwidth, depicted via the energy density gain from spontaneous spin polarization. Spin-polarized moiré quantum spin Hall bands realize a correlated quantum anomalous Hall phase. b At 1/6 hole doping, ferromagnetism (FM) persists for finite U [right axis] over a wide range of twist angles. Finite longer-ranged Coulomb repulsion ({U}^{prime}) [left axis] between nearest-neighbor Kagome sites drives this spin-polarized metal into a ν = 1/3 Laughlin phase, realizing a robust fractional Chern insulator, depicted in terms of the many-body excitation gap. c Fingerprint of the fractional Chern insulator in the low-energy spectrum for L = 24 (L = 30) unit cells with periodic boundary conditions and N = 8 (N = 10) fermions, depicted as a function of linearized total momentum (K1, K2). Inset depicts the characteristic spectral flow under insertion of a magnetic flux through the torus.
Having established a readily-accessible interaction-induced quantum anomalous Hall state at half filling, we now turn to the possibility of more exotic topologically-ordered phases at fractional fillings. Here, previous theoretical works established numerous examples of topological tight-binding models46,54,55,56,57 that a fractionally filled flat band with non-zero Chern number can in principle behave analogous to a fractionally filled Landau level and realize Abelian and non-Abelian fractional quantum Hall states in the absence of an external magnetic field. Uniformity of the Berry curvature is a key figure of merit58 and determines, jointly with the ideal droplet condition for the Fubini-Study metric59,60,61, the propensity for flat Chern bands to host fractional quantum Hall phases. Figure 3f quantifies Berry curvature fluctuations ({({{Delta }}Omega )}^{2}={int}_{{{{{{{{rm{BZ}}}}}}}}}{[{{Omega }}({{{{{{{bf{k}}}}}}}})-sqrt{3}{{{{{{{mathcal{C}}}}}}}}/8{pi }^{2}]}^{2}d{{{{{{{bf{k}}}}}}}}) for the Kagome moiré bands, where ({{{{{{{mathcal{C}}}}}}}}=pm !1) is the spin Chern number. One finds that the Berry curvature flattens monotonically as the twist angle is reduced, with fluctuations substantially suppressed for the third Kagome valence band at small angles.
To demonstrate twisted ZrS2 as a candidate to observe FQH physics with external magnetic fields, we focus on the conceptually simplest Laughlin ν = 1/3 state at 1/6 hole doping, and study the interacting problem at small twist angles in exact diagonalization. Analogous to the half-filled case, electrons in the almost-flat band can avoid local Coulomb repulsion U via spontaneous spin polarization, yielding a robust ferromagnetic instability as a function of U [Fig. 4b, right axis, dashed line] over all investigated twist angles. However, spontaneous spin polarization due to U now leaves a single Chern band at 1/3 hole doping, with the resulting electronic phase governed by longer-ranged Coulomb interactions ({U}^{prime}). To study the propensity to realize a Laughlin state, we numerically investigate the resulting phase diagram as function of bandwidth (W/{U}^{prime}) [Fig. 4b, left axis]. For W = 0, corresponding to the Landau level limit of a perfectly-flat Chern band, our exact diagonalization calculations for 6 × 5 unit cells reveal a three-fold ground state degeneracy for periodic boundary conditions [Fig. 4c] with a gap to well-separated many-body excitations which persists as a function of system size. These ground states lie in three total momentum sectors that match the generalized Pauli principle for FCIs56, flow into each other upon adiabatic insertion of a magnetic flux through handles of the torus (periodic boundary conditions) and remain energetically separated from excitations, confirming the ν = 1/3 FCI44,56. Combined, these results indicate the robust stabilization of a fractional Chern insulator. The conclusions remain largely independent of the twist angle, and the FCI persists upon inclusion of finite band dispersion W until the many-body excitation gap closes for (W/{U}^{prime} sim 0.3) [Fig. 4b, false color].
Having established a robust correlated quantum anomalous Hall phase at half filling and evidence for a ν = 1/3 fractional Chern insulator at one-sixth hole doping, an interesting follow-up question concerns the role of proximal deeper moiré valence bands, beyond the single-band approximation. For interactions that exceed the single-particle gap to other bands but remain smaller than the overall bandwidth of the three Kagome bands, the robustness of fractional Chern insulator phases has been well-documented62, in direct analogy to Landau level mixing in the conventional quantum Hall effect. A more substantial challenge however stems from details of possible longer-ranged electron interactions and exchange processes, which could serve to either enhance or suppress the stability of the fractionalized phases at different filling fractions. These processes sensitively depend on the screening environment and gating50, and microscopic calculations present a substantial methodological obstacle for twisted materials63,64. Conversely, analyzing the potential stability of more exotic yet more fragile non-Abelian quantum Hall states remains an interesting topic for future investigation. Furthermore, for sufficiently small twist angles, if the Coulomb repulsion exceeds the overall bandwidth of the three Kagome bands, sufficient screening could serve to form a local moment at overall half filling ν = 3/2. Such a Kagome Mott insulator would constitute a Moiré realization of a paradigmatic frustrated magnetic model, which has been under intense scrutiny for the potential to host an elusive quantum spin liquid phase.
Beyond the (fractional) quantum anomalous Hall effect, the realization of flat-band quantum spin Hall insulators further opens up the possibility to realize a myriad of unconventional ordered states of matter with non-trivial topology, including time-reversal invariant fractionalized phases, or topological superconductors. Consequently, twisted ZrS2 bilayers constitute a promising and tunable materials platform for such investigations, granting access to a novel and exotic regime of ultra-strong spin–orbit coupling that is not readily realizable in conventional crystalline solid-state systems. More broadly, a natural question concerns the extension of similar ideas of pseudospin potential engineering and strong spin-orbit coupling to other transition-metal dichalcogenide heterostructures such as TiS2 and HfS2 with a multi-component character of the valence band edge. At the same time, the emergence of a moiré Kagome lattice from the fortuitous but robust interplay of geometry and interlayer coupling at small twist angles opens up a new pathway towards a moiré realization of magnetic phases in a paradigmatic frustrated system.
Ab initio calculations are performed with the Vienna Ab initio Simulation Package (VASP)65 based on density functional theory (DFT). Plane-wave basis sets are employed with an energy cutoff of 450 eV. The pseudopotentials are constructed with the projector augmented wave method66 and the exchange-correlation functionals are treated within the generalized gradient approximation (GGA)67. Only the Γ point is considered in the calculations due to the large size of the moiré supercells. A vacuum region larger than 15 Angstrom along the z-axis is applied to eliminate artificial interactions between periodic slab images. All atoms are relaxed until the forces on each atom are less than 0.01 eV/Angstrom. Van der Waals corrections are applied with the Tkatchenko-Scheffler method68 during the relaxation. The figures for the atomic structures and the charge density distributions are generated with the VESTA code69.
In this work, DFT calculations are only used to provide a reliable single-particle description of the top moiré valence bands of twisted bilayer ZrS2. The role of strong electronic interactions within the flat band is subsequently investigated in detail starting from the continuum theory and the effective tight-binding models described in the main text (with parameters extracted from the DFT calculations), using large-scale many-body exact diagonalization calculations (see below). Although the band gaps in the systems to the conduction band are underestimated by the DFT calculations at the GGA level, the conduction band lies at high energies and remains empty, hence does not affect the emergent many-body state at small hole doping of the flat moiré valence bands. Only the band dispersion and the shape of the moiré bands are relevant in this work and these are well captured by GGA. As shown in refs. 38,70, many-body corrections to GGA for 2D transitional metal dichalcogenides mainly appear as a rigid shift of the bands such that band gap is enlarged.
Fractional Chern insulating phases and the spontaneous quantum anomalous Hall effect are studied using exact diagonalization calculations of the many-body ground state of the projected interaction Hamiltonian [Eq. (5)]. Calculations are performed for L1 × L2 unit cell clusters with periodic boundary conditions and discrete momenta k = n1b1/L1 + n2b2/L2, with ni = 0, …, Li − 1 and reciprocal lattice vectors ({{{{{{{{bf{b}}}}}}}}}_{1,2}={[2pi ,pm 2pi /sqrt{3}]}^{top }). Electron spin is explicitly included. Simulations of the interaction-induced quantum anomalous Hall effect at half filling are performed for 4 × 4 clusters. Results for ν = 1/3 fractional Chern insulators at 1/6 hole doping are obtained for 24-site (6 × 4) and 30-site (6 × 5) clusters.
The raw data sets used for the presented analysis within the current study are available from the corresponding authors on reasonable request.
Custom codes used in this work can be provided by the corresponding author on reasonable request. Ab initio calculations were performed with VASP (version 5.4.4).
Balents, L., Dean, C. R., Efetov, D. K. & Young, A. F. Superconductivity and strong correlations in moiré flat bands. Nat. Phys. 16, 725–733 (2020).
CAS Google Scholar
Kennes, D. M. et al. Moiré heterostructures as a condensed-matter quantum simulator. Nat. Phys. 17, 155–163 (2021).
CAS Google Scholar
Andrei, E. Y. & MacDonald, A. H. Graphene bilayers with a twist. Nat. Mat. 19, 1265–1275 (2020).
CAS Google Scholar
Andrei, E. Y. et al. The marvels of moiré materials. Nat. Rev. Mater. 6, 201–206 (2021).
ADS CAS Google Scholar
Cao, Y. et al. Correlated insulator behaviour at half-filling in magic-angle graphene superlattices. Nature 556, 80–84 (2018).
ADS CAS PubMed Google Scholar
Cao, Y. et al. Magic-angle graphene superlattices: a new platform for unconventional superconductivity. Nature 556, 43–50 (2018).
ADS CAS PubMed Google Scholar
Lu, X. et al. Superconductors, orbital magnets and correlated states in magic-angle bilayer graphene. Nature 574, 653–657 (2019).
ADS CAS PubMed Google Scholar
Stepanov, P. et al. Untying the insulating and superconducting orders in magic-angle graphene. Nature 583, 375–378 (2020).
ADS CAS PubMed Google Scholar
Rozhkov, A., Sboychakov, A., Rakhmanov, A. & Nori, F. Electronic properties of graphene-based bilayer systems. Phys. Rep. 648, 1–104 (2016).
ADS MathSciNet CAS Google Scholar
Shen, C. et al. Correlated states in twisted double bilayer graphene. Nat. Phys. 16, 520–525 (2020).
CAS Google Scholar
Kerelsky, A. et al. Moiréless correlations in abca graphene. Proc. Natl Acad. Sci. 118, https://www.pnas.org/content/118/4/e2017366118 (2021).
Rubio-Verdú, C. et al. Moiré nematic phase in twisted double bilayer graphene. Nat. Phys. 18, 196–202 (2022).
Arora, H. S. et al. Superconductivity in metallic twisted bilayer graphene stabilized by WSe2. Nature 583, 379–384 (2020).
ADS CAS PubMed Google Scholar
Chen, G. et al. Tunable correlated Chern insulator and ferromagnetism in a moiré superlattice. Nature 579, 56–61 (2020).
ADS CAS PubMed Google Scholar
Liu, X. et al. Tunable spin-polarized correlated states in twisted double bilayer graphene. Nature 583, 221–225 (2020).
ADS CAS PubMed Google Scholar
Cao, Y. et al. Tunable correlated states and spin-polarized phases in twisted bilayer–bilayer graphene. Nature 583, 215–220 (2020).
ADS CAS PubMed Google Scholar
He, M. et al. Symmetry breaking in twisted double bilayer graphene. Nat. Phys. 17, 26–30 (2021).
Wang, L. et al. Correlated electronic phases in twisted bilayer transition metal dichalcogenides. Nat. Mater. 19, 861–866 (2020).
ADS CAS PubMed Google Scholar
Chen, G. et al. Gate-tunable Mott insulator in trilayer graphene-boron nitride moiré superlattice. Nat. Phys. 15, 237–241 (2019).
CAS Google Scholar
Chen, G. et al. Signatures of Gate-tunable superconductivity in trilayer graphene/boron nitride moiré superlattice. Nature 572, 215–219 (2019).
CAS PubMed Google Scholar
Lu, X. et al. Superconductors, orbital magnets, and correlated states in magic angle bilayer graphene. Nature 574, 653–657 (2019).
ADS CAS PubMed Google Scholar
Cao, Y. et al. Strange metal in magic-angle graphene with near Planckian dissipation. Phys. Rev. Lett. 124, 076801 (2020).
ADS CAS PubMed Google Scholar
Sharpe, A. L. et al. Emergent ferromagnetism near three-quarters filling in twisted bilayer graphene. Science 365, 605 (2019).
ADS CAS PubMed Google Scholar
Serlin, M. et al. Intrinsic quantized anomalous hall effect in a moiré heterostructure. Science 367, 900–903 (2020).
ADS CAS PubMed Google Scholar
Wu, S., Zhang, Z., Watanabe, K., Taniguchi, T. & Andrei, E. Y. Chern insulators and topological flat-bands in magic-angle twisted bilayer graphene. Nat. Mat. 20, 488–494 (2021).
CAS Google Scholar
Chen, G. et al. Tunable correlated chern insulator and ferromagnetism in trilayer graphene/boron nitride moiré superlattice. Nature 579, 56–61 (2020).
ADS CAS PubMed Google Scholar
Liu, Z., Abouelkomsan, A. & Bergholtz, E. J. Gate-tunable fractional Chern insulators in twisted double bilayer graphene. Phys. Rev. Lett. 126, 026801 (2021).
Abouelkomsan, A., Liu, Z. & Bergholtz, E. J. Particle-hole duality, emergent Fermi liquids and fractional Chern insulators in moiré flatbands. Phys. Rev. Lett. 124, 106803 (2019).
ADS Google Scholar
Repellin, C. & Senthil, T. Chern bands of twisted bilayer graphene: fractional chern insulators and spin phase transition. Phys. Rev. Research 2, 023238 (2019).
Ledwith, P. J., Tarnopolsky, G., Khalaf, E. & Vishwanath, A. Fractional Chern insulator states in twisted bilayer graphene: an analytical approach. arXiv:1912.09634 (2019).
Wu, F., Lovorn, T., Tutuc, E., Martin, I. & MacDonald, A. H. Topological insulators in twisted transition metal dichalcogenide homobilayers. Phys. Rev. Lett. 122, 086402 (2018).
ADS Google Scholar
Li, T. et al. Quantum anomalous Hall effect from intertwined moiré bands. Nature 600, 641–646 (2021).
Zhang, Y., Devakul, T. & Fu, L. Spin-textured Chern bands in AB-stacked transition metal dichalcogenide bilayers. Proc. Nat. Acad. Sci. 118, e2112673118 (2021).
Xie, Y. M., Zhang, C. P., Hu, J. X., Mak, K. F. & Law, K. T. Valley-polarized quantum anomalous Hall state in moiré MoTe2/WSe2 heterobilayers. Phys. Rev. Lett. 128, 026402 (2022).
Zhang, M. et al. Controlled synthesis of zrs2 monolayer and few layers on hexagonal boron nitride. J. Am. Chem. Soc. 137, 7051–7054 (2015).
CAS PubMed Google Scholar
Qian, X., Liu, J., Fu, L. & Li, J. Quantum spin hall effect in two-dimensional transition metal dichalcogenides. Science 346, 1344–1347 (2014).
ADS CAS PubMed Google Scholar
Yu, Y. et al. High phase-purity 1t-´mos2-and 1t-´mose2-layered crystals. Nat. Chem. 10, 638–643 (2018).
CAS PubMed Google Scholar
Jiang, H. Structural and electronic properties of zrx2 and hfx2 (x = S and Se) from first principles calculations. J. Chem. Phys. 134, 204705 (2011).
ADS PubMed Google Scholar
Martino, E. et al. Structural phase transition and bandgap control through mechanical deformation in layered semiconductors 1t–zrx2 (x= s, se). ACS Mater. Lett. 2, 1115–1120 (2020).
CAS Google Scholar
Rasmussen, F. A. & Thygesen, K. S. Computational 2d materials database: electronic structure of transition-metal dichalcogenides and oxides. J. Phys. Chem. C. 119, 13169–13183 (2015).
CAS Google Scholar
Xian, L. et al. Realization of nearly dispersionless bands with strong orbital anisotropy from destructive interference in twisted bilayer MoS2. Nat. Comm. 12, 5644 (2021).
Angeli, M. & MacDonald, A. H. γ valley transition metal dichalcogenide moirè bands. Proc. Nat. Acad. Sci. 118, e2021826118 (2021).
CAS PubMed PubMed Central Google Scholar
Tang, E., Mei, J. W. & Wen, X. G. High temperature fractional quantum hall states. Phys. Rev. Lett. 106, 236802 (2010).
ADS Google Scholar
Wu, Y. L., Bernevig, B. A. & Regnault, N. Zoology of fractional chern insulators. Phys. Rev. B 85, 075116 (2011).
ADS Google Scholar
Neupert, T., Santos, L., Ryu, S., Chamon, C. & Mudry, C. Topological Hubbard model and its high-temperature quantum hall effect. Phys. Rev. Lett. 108, 046806 (2011).
ADS Google Scholar
Bergholtz, E. J. & Liu, Z. Topological flat band models and fractional chern insulators. Int. J. Mod. Phys. B 27, 1330017 (2013).
ADS MathSciNet MATH Google Scholar
Parameswaran, S. A., Roy, R. & Sondhi, S. L. Fractional quantum hall physics in topological flat bands. Comptes Rendus Phys. 14, 816 (2013).
ADS CAS Google Scholar
Neupert, T., Chamon, C., Iadecola, T., Santos, L. H. & Mudry, C. Fractional (chern and topological) insulators. Phys. Scr. T164, 014005 (2015).
ADS Google Scholar
Stern, A. Fractional topological insulators: a pedagogical review. Annu. Rev. Cond. Mat. Phys. 7, 349–368 (2016).
ADS CAS Google Scholar
Throckmorton, R. E. & Vafek, O. Fermions on bilayer graphene: symmetry breaking for b=0 and nu=0. Phys. Rev. B 86, 115447 (2011).
ADS Google Scholar
Bernevig, B. A., Song, Z. D., Regnault, N. & Lian, B. Tbg iii: Interacting hamiltonian and exact symmetries of twisted bilayer graphene. Phys. Rev. B 103, 205413 (2020).
ADS Google Scholar
Mielke, A. Ferromagnetic ground states for the Hubbard model on line graphs. J. Phys. A: Math. Gen. 24, L73 (1991).
ADS MathSciNet Google Scholar
Sondhi, S. L., Karlhede, A., Kivelson, S. A. & Rezayi, E. H. Skyrmions and the crossover from the integer to fractional quantum hall effect at small Zeeman energies. Phys. Rev. B 47, 16419–16426 (1993).
ADS CAS Google Scholar
Neupert, T., Santos, L., Chamon, C. & Mudry, C. Fractional quantum hall states at zero magnetic field. Phys. Rev. Lett. 106, 236804 (2011).
ADS PubMed Google Scholar
Sheng, D. N., Gu, Z. C., Sun, K. & Sheng, L. Fractional quantum hall effect in the absence of Landau levels. Nat. Commun. 2, 389 (2011).
ADS CAS PubMed Google Scholar
Regnault, N. & Bernevig, B. A. Fractional chern insulator. Phys. Rev. X 1, 021014 (2011).
Google Scholar
Wu, Y. L., Bernevig, B. A. & Regnault, N. Zoology of fractional chern insulators. Phys. Rev. B 85, 075116 (2012).
ADS Google Scholar
Parameswaran, S. A., Roy, R. & Sondhi, S. L. Fractional chern insulators and the w-infinity algebra. Phys. Rev. B 85, 241308 (2011).
ADS Google Scholar
Claassen, M., Lee, C. H., Thomale, R., Qi, X. L. & Devereaux, T. P. Position-momentum duality and fractional quantum hall effect in chern insulators. Phys. Rev. Lett. 114, 236802 (2015).
ADS PubMed Google Scholar
Jackson, T. S., Möller, G. & Roy, R. Geometric stability of topological lattice phases. Nat. Comm. 6, 8629 (2015).
ADS CAS Google Scholar
Lee, C. H., Claassen, M. & Thomale, R. Band structure engineering of ideal fractional chern insulators. Phys. Rev. B 96, 165150 (2017).
ADS Google Scholar
Kourtis, S., Neupert, T., Chamon, C. & Mudry, C. Fractional chern insulators with strong interactions that far exceed band gaps. Phys. Rev. Lett. 112, 126806 (2014).
ADS PubMed Google Scholar
Pizarro, J. M., Rösner, M., Thomale, R., Valentí, R. & Wehling, T. O. Internal screening and dielectric engineering in magic-angle twisted bilayer graphene. Phys. Rev. B 100, 161102 (2019).
ADS CAS Google Scholar
Vanhala, T. I. & Pollet, L. Constrained random phase approximation of the effective Coulomb interaction in lattice models of twisted bilayer graphene. Phys. Rev. B 102, 035154 (2020).
ADS CAS Google Scholar
Kresse, G. & Hafner, J. Ab initio molecular dynamics for liquid metals. Phys. Rev. B 47, 558 (1993).
ADS CAS Google Scholar
Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953 (1994).
ADS Google Scholar
Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865 (1996).
ADS CAS PubMed Google Scholar
Tkatchenko, A. & Scheffler, M. Accurate molecular van der Waals interactions from ground-state electron density and free-atom reference data. Phys. Rev. Lett. 102, 073005 (2009).
ADS PubMed Google Scholar
Momma, K. & Izumi, F. Vesta 3 for three-dimensional visualization of crystal, volumetric and morphology data. J. Appl. Crystallogr. 44, 1272–1276 (2011).
CAS Google Scholar
Qiu, D. Y., Felipe, H. & Louie, S. G. Optical spectrum of mos 2: many-body effects and diversity of exciton states. Phys. Rev. Lett. 111, 216805 (2013).
ADS PubMed Google Scholar
Download references
This work is supported by the European Research Council (ERC-2015-AdG-694097), Grupos Consolidados (IT1249-19), and SFB925. M.C. is supported by a startup grant from the University of Pennsylvania. A.R. is supported by the Flatiron Institute, a division of the Simons Foundation. We acknowledge funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under RTG 1995, within the Priority Program SPP 2244 “2DMP”, under Germany’s Excellence Strategy – Cluster of Excellence and Advanced Imaging of Matter (AIM) EXC 2056 – 390715994 and RTG 2247. L.X. acknowledges the support from Distinguished Junior Fellowship program by the South Bay Interdisciplinary Science Center in the Songshan Lake Materials Laboratory and the Key-Area Research and Development Program of Guangdong Province of China (Grants No. 2020B0101340001). We acknowledge computational resources provided by the Simons Foundation Flatiron Institute, the Max Planck Computing and Data Facility, and the Platform for Data-Driven Computational Materials Discovery of the Songshan Lake laboratory. This work was supported by the Max Planck-New York City Center for Nonequilibrium Quantum Phenomena.
Open Access funding enabled and organized by Projekt DEAL.
These authors contributed equally: Martin Claassen, Lede Xian.
Department of Physics and Astronomy, University of Pennsylvania, Philadelphia, PA, 19104, USA
Martin Claassen
Songshan Lake Materials Laboratory, 523808, Dongguan, Guangdong, China
Lede Xian
Max Planck Institute for the Structure and Dynamics of Matter, Center for Free Electron Laser Science, 22761, Hamburg, Germany
Lede Xian, Dante M. Kennes & Angel Rubio
Institut für Theorie der Statistischen Physik, RWTH Aachen University and JARA-Fundamentals of Future Information Technology, 52056, Aachen, Germany
Dante M. Kennes
Center for Computational Quantum Physics, Simons Foundation Flatiron Institute, New York, NY, 10010, USA
Angel Rubio
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
M.C. and L.X. conceived the project and coordinated the research. M.C. devised the theoretical models and performed the many-body calculations. L.X. performed the ab initio calculations and analysis. M.C., L.X., D.M.K., and A.R. discussed and analyzed the results and contributed to writing the manuscript.
Correspondence to Martin Claassen, Dante M. Kennes or Angel Rubio.
The authors declare no competing interests.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
Reprints and Permissions
Claassen, M., Xian, L., Kennes, D.M. et al. Ultra-strong spin–orbit coupling and topological moiré engineering in twisted ZrS2 bilayers. Nat Commun 13, 4915 (2022). https://doi.org/10.1038/s41467-022-31604-w
Download citation
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-022-31604-w
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
Advertisement
Advanced search
© 2022 Springer Nature Limited
Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.
RECAP: Bobby Nightingale Murder Trial Day 4 – WAGM
CARIBOU, Maine (WAGM) -The state rested its case in the bobby nightingale murder trial. Newssource 8′s Corey Bouchard is at the courthouse and has details about the case.
The State wrapped up presenting their case in the Bobby Nightingale double murder trial.
Prior to resting their case, the state finished up questioning of Detective Gregory Roy who played an audio recording of Bobby Nightingale where nightingale said “its whatever at this point because they caught me with a gun so I don’t give a F”.
The next witness was Detective Christopher Cookson of the Maine State Police Computer Crimes Unit. He showed a powerpoint presentation depicting the location of Bobby Nightingale’s cell phone from the days prior to the murders, until August 15th 2019 when a signal was no longer transmitted .The powerpoint showed Nightingale in the location at the times of the murders.
A Number of other witnesses testified from the Maine Crime Lab about DNA analysis and Fingerprint Analysis, where on cross examination, Defense Lawyer Verne Paradis had asked a witness if it was possible that no fingerprints from nightingale found at the scene proves that he wasn’t there. Attorney Meg Elam rebutted by asking the expert “is it possible there really is a loch ness- monster” But the Testimony that caught the most attention from the Jury was from Kimberly James a Senior Firearm Analyst at the Maine State Police crime lab. She showed analysis of the Bullets and Casings found at the scene and how they compared to the .380 Pistol Found on Nightingale and the .45 Cartridge found at Nightingales residence. The expert testified that the.380 cases at scene match the gun found on Nightingale and that “the cartridge case found In the windowsill matches the 7 cartridge cases found at the scene”
With that, the state rested it’s case and the judge dismissed the court for the day , stating there will be a meeting to talk about some housekeeping issues related to the wrapping up of the case
And the Defense will begin their case August 19th where they are expected to call a number of different witnesses as well as to try to instill reasonable doubt into the jury that bobby nightingale did not commit the crimes the state has charged.
we will be continuing to monitor this case as it goes into it’s 5th day, Friday. Unless the defense finishes their case within a day and the jury is fast at deliberations, the case is likely to extend into Monday.
Copyright 2022 WAGM. All rights reserved.
CBRE Arranges $50.6M Construction Loan for Multifamily Redevelopment Project in Chapel Hill, North Carolina – REBusinessOnline

Upon completion, Link Apartments Calyx will offer studio, one- and two-bedroom apartments, as well as a fitness center, coworking space, a cycle center, pool, rooftop resident lounge and a parking garage.
CHAPEL HILL, N.C. — CBRE has arranged $50.6 million in construction financing for Link Apartments Calyx, a 304-unit multifamily redevelopment project in Chapel Hill. Nate Sittema, Kristen Reilley, Elliott Voreis and Ben Hardee of CBRE Capital Markets’ Debt and Structured Finance team in the Carolinas originated the 10-year, 60 percent loan-to-stabilized value construction loan through an unnamed life insurance company on behalf of the borrower, Grubb Properties. Link Apartments Calyx is located at the intersection of Lanark and Maxwell roads within Phase II of the redevelopment of Glen Lennox, an approximately 70-acre historic master-planned neighborhood that was purchased by Grubb Properties in 1985.
Link Apartments Calyx will offer studio, one- and two-bedroom apartments, as well as a fitness center, coworking space, a cycle center, pool, rooftop resident lounge and a parking garage. Construction is expected to be completed in 2023. Overall, the second phase of the Glen Lennox redevelopment is set to include 650 residential units, 90,000 square feet of retail space, a 150-room hotel and 275,000 square feet of office space.
FAQs & Guide on How To Book & Sign up for In-Person Classes at Peloton Studios in New York or London (PSNY & PSL) – Peloton Buddy – Pelo Buddy
After a more than two year hiatus, the Peloton studios are once again open to the public. Following the soft reopening phase that began in June, Peloton Studios New York (PSNY) and Peloton Studios London (PSL) officially celebrated their grand reopening on August 19, 2022. When Peloton first shared the news, they also relaunched the studio booking site, which had been dormant since the initial studio closure. If you have questions about how to book an in-person class, you’re not alone! This article provides an in-depth overview of how to book and take an in-person class at PSNY and PSL. First, we’ll cover a lot of frequently asked questions about how to book classes; then we’ll take you through a step-by-step tutorial of the actual booking process.
There is a special website to book in-person classes, which you can find at studio.onepeloton.com. There you will find the list of classes. See the bottom of this article for an in-depth walkthrough, guide, and screenshots of the booking process and what to expect.
Classes at PSNY cost $35 and classes at PSL cost £25. You technically buy a credit, which you will then use to reserve your spot in the class. You can buy credits at https://studio.onepeloton.com/shop. One interesting note is that you can essentially gift class credits to friends and family by purchasing the credits and then transferring them to another member. Similarly, if you purchased credits that you will no longer be able to use, you can transfer them to another member. You can visit your account page to view your credits and any classes you have scheduled, or previously taken at the studio.
Yes, each class is the same cost no matter the class length or modality.
The Peloton Studios are currently open to members on Fridays through Sundays. Mondays through Thursdays are reserved for coach-to-camera instruction.
Peloton is currently releasing classes for bookings 6 weeks in advance at both Peloton Studios New York & London.
Peloton is releasing new in-person classes for booking on Mondays around 12:00pm ET (this is of course subject to change, but has been stated as the policy). Each Monday, they’ll release the next weekend’s worth of classes – which will be six weeks away. Thus, the schedule will be released on a rolling six week basis.
Members who visited Peloton’s original 23rd Street Studio will recall that the schedule was released two weeks out, and you could book exactly one week in advance. However, now that Peloton is wanting their studios to be a destination experience, they are opening up the schedule much further in advance so that members can make travel arrangements.
With the studio only recently having reopened, demand is high to take in-person classes. This means that classes are selling out within minutes or hours of them becoming available (which again, is typically around 12:00pm ET on Mondays). We expect over the coming months it to become easier to book a spot once more people have had a chance to visit the studios.
No, you cannot pick your own Bike/Tread/mat location when booking. The assignment is random – your confirmation email may note a spot auto-assigning to you, but this does not really matter. When at the studio, it is essentially first-come first-served as you enter the room. However, Peloton may move you at their discretion (potentially based on height or other factors). This applies in particular to the yoga and strength studios, which are smaller and present more of an opportunity for taller people potentially blocking the cameras.
At the original 23rd Street Studio, one of the steps was using a map to select a specific Bike or Tread. Peloton has done away with this since their reopening.
Although there isn’t a published map of the Peloton Studios cycling studio (or other studios), there are still limits for each class. The capacities for Peloton Studios New York (PSNY) are below:
The capacities for Peloton Studios London (PSL) are below:
(At time of publishing PSL does not have yoga classes, and strength classes are recorded without members).
No. You’ll know the modality, length, and instructor – but no other additional information will be available upon booking. You will be able to check the upcoming schedule approximately two weeks in advance to learn additional information.
If you cancel more than 12 hours in advance, your credit will be returned to your account (but you will not receive a refund). That credit must be used in 60 days.
We have heard that in mid-August Peloton implemented a limit of two classes (per person, per day) in order to allow more members the opportunity to check out the new studios. So for example, if you’re visiting NYC for the weekend you can take a maximum of six classes throughout the weekend: two on Friday, two on Saturday, and two on Sunday. We imagine this limit might be reexamined in the future.
In theory, yes. However, it is important to note that the doors to the class open 30 minutes prior to the class start time, and “close” 15 minutes prior to the start time. So if you aren’t there 15 minutes before, Peloton might give your spot to someone on the waitlist. You should avoid booking back-to-back classes too close together, as you want to allow time following the class to get your picture taken with the instructor.
According to the FAQ page members may book up to five spots in a class. You can do this by booking once, then going back and booking the same class again as many times as the number of spots you’re looking to book. However, we’ve received reports that Peloton plans to change this and only allow one spot in a class per member – so this means that if you want to go with friends they would need to book their own spot. We’ll keep an eye on this.
For group bookings – perhaps such as an unofficial home-rider invasion from a particular leaderboard tag – you can reach out to [email protected] (PSNY) or [email protected] (PSL). You can expect to hear back from Peloton within 1-2 business days regarding pricing and availability. You can also contact the concierge for visitor information at [email protected] (PSNY) or [email protected] (PSL).
Sort of. To actually book a class, you do need a Peloton account, which you can sign up for for free as part of the booking process. However, you don’t need to be an active member to do so.
Bookings at the studio are open to all members regardless of whether they are All-Access (those who own a Peloton device such as the Bike or Tread) or App-only. Note that you do still need a Peloton account, which you can sign up for for free as a part of the booking process. The cost of an in-person class is the same regardless of your membership type as well.
At the 23rd Street studio members could buy packs of class credits at a slight discount. As of right now, they are not allowing you to do this at Peloton Studios New York (PSNY) and Peloton Studios London (PSL).
You should see a “find a class” button in the top right corner of the studio booking site – clicking this will allow you to toggle between the NYC and London studios. There is also a “location” field in the left-hand side bar of the page that allows you to select either NYC or London.
The 23rd Street studio offered a monthly pass (that cost approximately $400) for unlimited studio rides. Since Peloton wants a visit to Peloton Studios New York (PSNY) or Peloton Studios London (PSL) to be a member experience with the opportunity to meet instructors and other members, they aren’t treating the studios like a local gym – so they are currently no longer offering an unlimited pass option for locals.
If you don’t have a credit, the booking process will let you buy one credit as you check out. Just keep in mind to have your credit card handy, as you can’t save your credit card info to your profile; you have to enter it as you buy that credit. Note that credits are only good for 60 days, so if you’re not exactly sure when you’ll visit, it is probably safer to wait until you’re ready to book before purchasing the credit.
Peloton will periodically add the “Waitlist” button next to classes that had been previously full, and even more rarely, but the “Book” button back so you can directly book and skip the waitlist. This might occur if there were cancellations, or they had been holding spots for something and no longer need them. There is no set time or date these get released back – you just have to get lucky and check the studio website at the right time to see that.
A waitlist spot means that you don’t have a guaranteed spot in the class, but there is a chance Peloton will email you if someone cancels. There is no guarantee you will get into the class.
If you’re not moved from the waitlist into the class you will not receive your money back, but your credit will be returned to you. You can then book another class before your credit expires in 60 days.
There is no solid answer – it simply depends on when people who were able to book a spot in the class cancel. It could be a few days or weeks in advance of the class; it could be the morning of; or it could be 15 minutes before the class begins (in the instance of a no-show, for example).
This entirely depends on your personal situation and whether you’re willing to gamble on a potentially “wasted” trip if you go to the studios but ultimately don’t get off the waitlist. If you’re local or already in town and don’t have anything else to do, it is potentially worth a shot. However, we would not advise planning an entire trip around a visit if you’re only on the waitlist, because there is no guarantee you’ll get into the class.
If you’re interested in taking a class in-person at the Peloton studios, head to https://studio.onepeloton.com/. You’ll be brought to a welcome screen where you have two options: you can click the “book your spot” button to jump straight to the schedule, or you can hit the down carrot button at the bottom of the screen to learn more about the studio space and explore instructors.
Upon clicking the “book your spot” button, you will be brought to the main schedule page. The first thing to look for is whether you’re viewing the New York studio or the London studio, as you can only view the schedule for one at a time.
Below the location box there is a list of fitness disciplines that you can filter by: cardio, cycling, running, running, strength, walking, and yoga. You can select certain boxes so that you only view classes of that modality. For example, if you’re dead-set on taking a cycling class at Peloton studios, you should select the “cycling” box so that you only see upcoming rides.
You can also filter by instructors. Note that there is only basic information listed for each of the classes: the modality, length of the class, and the instructor. If the date is within the next two weeks you can head to the Peloton class schedule to get additional information, such as whether the class is music-focused vs. fitness-focused, or centers around a particular theme. Peloton has been adding more details to the studio booking site once the class is within two weeks; but the classes have been completely booked by the time those updates are made.
When viewing the list of upcoming classes you will be able to see on the right-hand side whether the class is available to book, waitlisted, or full. Completely full classes will not have a clickable button. You can only sign up for a class if you see a “book” button, or if there is a waitlist option available.
After clicking the “book” button for the class you’d like to take, you’ll be brought to a screen that contains only that class information: the modality, date, time, instructor, and duration. There is also a disclaimer provided by Peloton: “As we’re a LIVE production studio, spots are assigned by us. You may be placed on spot and then be asked to move for production purposes.”
This is a major change from how Peloton previously operated their studio booking, as members used to be able to select their specific spot each class. It appears that moving forward Peloton will assign that spot for you.
When the studio booking site first relaunched members booking a cycling class were asked to input their shoe size or indicate if they were bringing their own, but that section has since been removed from the site.
If you don’t have any credits, you will need to purchase those in order to book the class. Classes at PSNY cost $35 and classes at PSL cost £25. You can buy them as a step of the booking process, or buy them ahead of time. Note if you buy a credit ahead of time, they are only good for 60 days, so be sure you will be able to find a class at the time you can visit before doing so.
Upon booking the class you’ll be brought to a confirmation screen that says: “You’re booked!” From there you can return to the schedule, view your classes, and review the studio rules. You’ll also receive any email containing the booking confirmation.
If you need to cancel the class you’ve booked, you can do so by navigating to the my classes page and clicking the “cancel” button next to that particular class. You may also reach out by phone, email, or chat. You must cancel at least 12 hours before the class starts in order for your class credit to be returned to your account. Otherwise it will be considered a “late cancellation” and you will lose the credit.
If you’re on the waitlist for a class but can no longer attend, you should also remove yourself from the waitlist at least 12 hours ahead of the class start time.
You can visit your account page to view your credits and any classes you have scheduled, or previously taken at the studio.
The day has finally arrived and you’re ready to head to the studio! You’ll want to make sure you’re completely prepared for the experience. In the coming days we’ll have a separate guide covering everything to know for the actual visit.
For additional information & questions not covered about booking Peloton classes, in either New York or London, you can always visit the Peloton Studios FAQ site.
Want to be sure to never miss any Peloton news? Sign up for our newsletter and get all the latest Peloton updates & Peloton rumors sent directly to your inbox.
Get our newsletter and stay up-to-date with all the Peloton news you need to know!
Your email address will not be published.
Pelo Buddy is in no way affiliated with Peloton Cycle or Peloton Interactive. Peloton is a registered trademark of Peloton Interactive, Inc. Some links on the site are affiliate links. As an Amazon Associate I earn from qualifying purchases.
























