BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Michigan Institute for Computational Discovery and Engineering - ECPv6.15.20//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-ORIGINAL-URL:https://micde.umich.edu
X-WR-CALDESC:Events for Michigan Institute for Computational Discovery and Engineering
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:America/Detroit
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20220313T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20221106T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20230312T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20231105T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20240310T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20241103T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=America/Detroit:20231005T110000
DTEND;TZID=America/Detroit:20231005T123000
DTSTAMP:20260605T030748
CREATED:20230918T024126Z
LAST-MODIFIED:20231018T165056Z
UID:10000649-1696503600-1696509000@micde.umich.edu
SUMMARY:SciML Webinar: Jianke Yang - Generative Adversarial Symmetry Discovery
DESCRIPTION:Speaker: Jianke Yang (UC San Diego) \n\n\nSession Chair: Bharath Ramsundar (Deep Forest Sciences) \nAbstract:Despite the success of equivariant neural networks in scientific applications\, they require knowing the symmetry group a priori. Automatic symmetry discovery methods aim to relax this constraint and learn invariance and equivariance from data.  We propose a framework\, LieGAN\, to automatically discover equivariances from a dataset using a paradigm akin to generative adversarial training. Specifically\, a generator learns a group of transformations applied to the data\, which preserves the original distribution and fools the discriminator. LieGAN represents symmetry as an interpretable Lie algebra basis and can discover various symmetries such as the rotation group and the restricted Lorentz group in trajectory prediction and top-quark tagging tasks. More generally\, LieGAN can also be extended to discover the nonlinear symmetries in high-dimensional dynamics. The learned symmetry can be readily used in several existing equivariant neural networks to improve prediction accuracy and generalization. It can also improve the symbolic equation discovery and long-term forecasting for various dynamical systems. \n\n\n 
URL:https://micde.umich.edu/event/sciml-webinar-jianke-yang-generative-adversarial-symmetry-discovery/
CATEGORIES:Sciml,SciML Webinar Series
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Detroit:20231012T110000
DTEND;TZID=America/Detroit:20231012T130000
DTSTAMP:20260605T030748
CREATED:20230915T150330Z
LAST-MODIFIED:20231018T173554Z
UID:10000644-1697108400-1697115600@micde.umich.edu
SUMMARY:SciML Webinar Justin Beroz: A closed-form mathematical framework for modeling turbulent fluids
DESCRIPTION:Speaker: Justin Beroz (ReynKo Inc.) \n\nSession Chair: Varun Shankar (Physics Inverted Mataerials) \nAbstract: Despite significant advances over the past two centuries\, a complete general mathematical framework for turbulent fluid motion has yet to be put forth\, and remains the longest standing unsolved problem in classical physics. I will present such a framework\, which is based on constructing a spectral decomposition for the fluid’s kinetic energy from first principles. The approach departs from the usual Reynolds decomposition and yields a set of closed and solvable ordinary differential equations in matrix form. Within this prescription\, the linear terms in the Navier-Stokes equations correspond to a symmetric matrix operator\, and the nonlinear convective term enters as an anti-symmetric operator that provides coupling between eigenstates of turbulent fluctuation. Specifically\, I will present a derivation for the turbulent energy spectrum\, including the Kolmogorov energy cascade; elucidate instability mechanisms for the transition to turbulence;  and detail the analytical solution for turbulence in a box. Careful attention will be given to the physical picture and scaling\, in addition to the rigorous mathematical program. The talk will conclude with a forward look into current efforts implementing the model into a numerical simulation within my company\, ReynKo Inc.
URL:https://micde.umich.edu/event/workshop-seminarsciml-webinar-justin-beroz/
CATEGORIES:Micde,Scientific Computing,Sciml,SciML Webinar Series,Webinar
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Detroit:20231019T110000
DTEND;TZID=America/Detroit:20231019T130000
DTSTAMP:20260605T030748
CREATED:20230915T150343Z
LAST-MODIFIED:20231025T194805Z
UID:10000646-1697713200-1697720400@micde.umich.edu
SUMMARY:SciML Webinar Ji Qi: DImensionality-Reduced Encoded Clusters with sTratified (DIRECT) sampling for Robust Training of Machine Learning Interatomic Potentials
DESCRIPTION:https://umich.zoom.us/j/95111677727?pwd=V1Q5MkUwT2NpOFVhd0ZRVGR1YTM3Zz09 \n\nSpeaker: Ji Qi (UC San Diego and LLNL)\nSession Chair: Daniel Schwalbe-Koda (UC Los Angeles) \nAbstract: Machine learning interatomic potentials (MLIPs) enable accurate simulations of materials at scales beyond conventional first-principles approaches\, and they have played increasingly important roles in understanding and design of materials. However\, MLIPs are only as accurate and robust as the data they are trained on. In this seminar\, I will present DImensionality-Reduced Encoded Clusters with sTratified (DIRECT) sampling as an approach to select a robust training set of structures from a large and complex configuration space. By applying DIRECT sampling on the Materials Project relaxation trajectories dataset with over one million structures and 89 elements\, we develop an improved materials 3-body graph network (M3GNet) universal potential that extrapolate more reliably to unseen structures. We further show that molecular dynamics (MD) simulations with universal potentials such as M3GNet can be used in place of expensive ab initio MD to rapidly create a large configuration space for target materials systems. For demonstration\, we combined this scheme with DIRECT sampling to develop a reliable moment tensor potential for titanium hydrides without the need for iterative augmentation of training structures. \nIn this seminar\, I will walk through two Jupiter notebooks to showcase DIRECT sampling with the two example cases demonstrated in our manuscript\, so that audience can expect to reproduce our major results with no trouble. Hopefully\, DIRECT sampling will serve as a straightforward\, efficient\, useful plug-in for the robust training of MLIPs across any compositional complexity.
URL:https://micde.umich.edu/event/workshop-seminarsciml-webinar-ji-qi-2/
CATEGORIES:Micde,Scientific Computing,Sciml,SciML Webinar Series,Webinar
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Detroit:20231026T110000
DTEND;TZID=America/Detroit:20231026T130000
DTSTAMP:20260605T030748
CREATED:20231017T170318Z
LAST-MODIFIED:20231107T231334Z
UID:10000658-1698318000-1698325200@micde.umich.edu
SUMMARY:SciML Webinar: Bowen Deng - CHGNet: pretrained universal interatomic potential to study electron coupled ionic systems.
DESCRIPTION:https://umich.zoom.us/j/95111677727?pwd=V1Q5MkUwT2NpOFVhd0ZRVGR1YTM3Zz09 \n\nSpeaker: Bowen Deng (UC Berkeley)\nSession Chair: Sakidja Ridwan (Missouri State University) \nAbstract: Large-scale simulations with complex electron interactions remain one of the greatest challenges for atomistic modeling. Although classical force fields often fail to describe the\ncoupling between electronic states and ionic rearrangements\, the more accurate ab-initio molecular dynamics suffers from computational complexity that prevents long-time and large-\nscale simulations\, which are essential to study technologically relevant phenomena. Our work presents the Crystal Hamiltonian Graph Neural Network (CHGNet)\, a graph-neural-\nnetwork-based machine-learning interatomic potential (MLIP) that models the universal potential energy surface. CHGNet is pretrained on the energies\, forces\, stresses\, and magnetic moments\nfrom the Materials Project Trajectory Dataset\, which consists of over 10 years of density functional theory calculations of ∼ 1.5 million inorganic structures. The explicit inclusion of\nmagnetic moments enables CHGNet to learn and accurately represent the orbital occupancy of electrons\, enhancing its capability to describe both atomic and electronic degrees of freedom.\nWe demonstrate several applications of CHGNet in solid-state materials and energy storage applications.
URL:https://micde.umich.edu/event/sciml-webinar-bowen-deng/
CATEGORIES:Micde,Scientific Computing,Sciml,SciML Webinar Series,Webinar
END:VEVENT
END:VCALENDAR