Bulletin of the American Physical Society
APS March Meeting 2021
Volume 66, Number 1
Monday–Friday, March 15–19, 2021; Virtual; Time Zone: Central Daylight Time, USA
Session V19: Building the Bridge to Exascale: Applications and Opportunities for Materials, Chemistry, and Biology IIFocus Live
|
Hide Abstracts |
Sponsoring Units: DCOMP DCMP DAMOP DCP Chair: Jack Deslippe, Lawrence Berkeley National Laboratory |
Thursday, March 18, 2021 3:00PM - 3:36PM Live |
V19.00001: Preparing for exascale: additive manufacturing process modeling at the fidelity of the microstructure Invited Speaker: James Belak In FY17, the USDOE Exascale Computing Project (ECP) initiated projects to design and develop simulation codes to use exascale computing. This application development is organized around computational motifs. Here, we present an overview of the motifs of computational materials science, from the “particles” used by molecular dynamics to the “grids” used by phase-field models and the various solution algorithms. Examples will be taken from the co-design centers ExMatEx and CoPA, as well as the Additive Manufacturing (AM) application development project ExaAM. This project includes an integration of all the computational components of the metal AM process into a coupled exascale modeling environment, where each simulation component itself is an exascale simulation. What has emerged is that exascale computing will enable AM process modeling at the fidelity of the microstructure. Here we discuss what this means, in particular, tight coupling of Process-Structure-Property calculations, and the feedback into AM-specific constitutive models. Macroscopic continuum codes (Truchas and OpenFOAM) are used to simulate melt-refreeze, within which mesoscopic codes (Phase-field and Cellular Automata) are used to simulate the development of material microstructure. This microstructure is then used by polycrystal plasticity codes (ExaConstit) to calculate local material properties. The project is driven by a series of demonstration problems (NIST AMBench) that are amenable to experimental observation and validation. We present our coupled exascale simulation environment for additive manufacturing and its initial application to AM builds. |
Thursday, March 18, 2021 3:36PM - 4:12PM Live |
V19.00002: A Path to the Exascale for Atomistic Simulations with Improved Accuracy, Length and Time Scales Invited Speaker: Mitchell Wood With exascale super computers arriving in the near future, it is timely to ask whether our simulation software is capable of matching this unprecedented computing capability. While many research challenges in material physics, chemistry and biology lie just out of reach on peta-scale machines due to length and time restrictions inherent to Molecular Dynamics(MD), questions of the accuracy of our simulations will continue to linger. Simply running the same peta-scale simulations with more atoms on a larger computer (weak scaling) does not advance the accessible timescales, nor does it avoid the pitfalls of empirically developed constitutive models. This talk will overview the U.S. Department of Energy* EXAALT (EXascale Atomistics for Accuracy, Length and Time) project and our efforts to provide software tools for MD that not only scale efficiently to huge atom counts, but also enable efficient MD simulations for smaller systems too. New parallel time-acceleration methods such as sublattice-ParSplice and local hyperdynamics have been developed along with quantum accurate machine learned interatomic potentials to study damage accumulation in plasma facing materials. |
Thursday, March 18, 2021 4:12PM - 4:24PM Live |
V19.00003: Accelerating Quantum Molecular Dynamics simulations with GPUs Jean-Luc Fattebert Graphics Processing Units (GPU) accelerators on the most powerful supercomputers give us an opportunity to speed up time-to-solution for large-scale Quantum Molecular Dynamics simulations that would otherwise be too slow for practical purposes. But using this type of hardware in an efficient manner presents serious challenges. Large amounts of code may need to be rewritten and algorithmic changes may be required for optimal efficiency. Even then, using GPUs at full capacity is not straightforward, especially if there is not enough computational work for a GPU. In this talk I will present some software library solutions in development to facilitate porting electronic structure codes to new architectures, as well as some parallel strategies and algorithms that can help speed up time-to-solution in real applications. |
Thursday, March 18, 2021 4:24PM - 4:36PM Live |
V19.00004: Computational cost - accuracy comparison for machine learned interatomic models across hardware Sam Reeve, Kashyap Ganesan, Saaketh Desai, James Belak Predicting atomic level behavior and mechanisms in materials is increasingly using complex, machine learned (ML), and data-driven interatomic models. This has and will continue to enable scientific discovery with more accurate and flexible atomic descriptions than their traditional, empirical counterparts; however, these models are also generally much more computationally expensive. In light of the continued conversion towards GPUs (and hybrid CPU/GPU) in scientific computing and particularly exascale computing, we demonstrate performance portable ML interatomic models, including Behler-style neural network (NNP) and spectral neighbor analysis (SNAP) potentials, employing the Kokkos programming model and the Co-design center for Particle Applications (CoPA) Cabana particle library. We discuss the strategies for improving performance across architectures and hardware vendors. In addition, we discuss plans for additional performance portable ML interatomic models and the potential pathways for others. This includes interfacing with existing codes, emphasizing kernel-based code, and increasing exposed parallelism at multiple levels. |
Thursday, March 18, 2021 4:36PM - 4:48PM Live |
V19.00005: QTensor: Parallel Quantum Circuit Simulator Yuri Alexeev, Danylo Lykov, Cameron Ibrahim, Alexey Galda We present a parallel quantum circuit simulator* designed to run on large supercomputers with the eventual goal to run at scale on exa-scale supercomputers Aurora and Frontier. The simulator is based on the tensor network representation of quantum circuits. We proposed a novel parallelization strategy that is based on the splitting of the partially contracted tensor expression. The resulting slices of the tensor expression had significantly smaller memory footprints which allowed us to contract partial tensor networks on individual MPI ranks. We will discuss the performance considerations of working with tensor networks at scale and demonstrate the efficiency of QAOA simulation in a hybrid GPU-CPU environment using both Xeon Phi and Nvidia GPUs, on circuits with over a large number of qubits. |
Thursday, March 18, 2021 4:48PM - 5:00PM Live |
V19.00006: Memory-function approach for electronic transport in disordered solids Maria Troppenz, Brett Green, Santiago Rigamonti, Claudia Draxl, Jorge Osvaldo Sofo We present an ab initio method for the evaluation of transport coefficients from electronic-structure calculations which goes beyond the relaxation-time approximation. We focus on the electrical conductivity that within the Kubo linear response theory is related to the current-current correlation function. We evaluate this correlation with a memory function approach including scattering processes to infinite order. We have implemented this method in the exciting code [1]. It only requires the non-interacting space- and time-dependent charge-charge correlation function and a description of the scattering mechanism. As a proof of principle, we calculate the conductivity of sodium limited by static disorder. The self-consistent approach describes the localization of the carriers in this simple metal for strong disorder. In comparison with the relaxation-time approximation, our method captures the Anderson localization of the system and corrects the overestimation of the conductivity in the metallic phase. |
Thursday, March 18, 2021 5:00PM - 5:12PM Live |
V19.00007: Scalable Frameworks for Reinforcement Learning for Control of Self-Assembling Materials and for Chemistry Design Christine Sweeney, Paul Welch, Malachi Schram, Logan Ward The ExaLearn Exascale Computing Project has developed scalable frameworks for reinforcement learning (RL) to create policies to control scientific processes such as the self-assembly of block copolymers and chemical design. These policies could drastically reduce the time required to navigate large parameter spaces, optimizing experimental protocols. This accelerated search methodology may thus guide materials annealing experiments, exploration of candidate structures for battery materials, or evaluation of the configurational space for low-energy water clusters. The frameworks use various RL algorithms, environments and fast-running scientific simulations for the training process. RL training can be thought of as creating a sequence of moves in a game; at each move the player (agent) may decide to exploit previous knowledge (a policy) or explore new parameters (run a simulation). Scalability is achieved by running many RL training episodes on different nodes and aggregating models. Challenges include developing simulations, fully utilizing CPU and GPU resources on each node, and aggregating policies so as not to impede learning. Initial results for single node performance show high resource utilization, fast learning per episode, and promise for the target applications. |
Thursday, March 18, 2021 5:12PM - 5:24PM Live |
V19.00008: Gauge-covariant derivatives of the Berry curvature and orbital moment by Wannier interpolation Xiaoxiong Liu, Miguel Ángel J. Herrera, Stepan Tsirkin, Ivo Souza The momentum-space derivatives of the Berry curvature Ω and intrinsic orbital magnetic moment m of the Bloch states arise in multiple problems, such as the nonlinear anomalous Hall effect [1] and magneto-transport within the Boltzmann-equation formalism [2]. To study these properties from first principles, we developed a Wannier interpolation scheme for evaluating "generalized derivatives" of the non-Abelian Ω and m matrices for a group of bands of interest confined by some energy region. |
Thursday, March 18, 2021 5:24PM - 5:36PM Live |
V19.00009: First-Principle Calculation of Charge Carrier Mobility in 3D and 2D Hybrid Perovskites xiaoliang zhang, Andrew V Brooks, Xiaoguang Zhang We compute charge carrier mobilities of hybrid perovskite from complex band structure using the Quantum Espresso suite. An effective medium obtained from an appropriate configurational average acquires an imaginary ‘self-energy’ to that related to the mean scattering lifetime. We calculate the self-energy by matching the imaginary part of complex wave vector of the effective medium to that of the real configuration, where the interference effect between neighboring supercells is removed by an absorbing complex potential. The scattering lifetime yields the carrier mobility through Boltzmann transport equation. To allow computation over a sufficiently large supercell, the organic molecules in the hybrid perovskites are represented by an effective Molecular Pseudopotential. We calculate the 3D and 2D hybrid perovskite carrier mobility limited by the orientational disorder of the methylammonium ion and its dependence on temperature and, in the case 2D perovskite, thickness. |
Thursday, March 18, 2021 5:36PM - 5:48PM Live |
V19.00010: Energy of many-particle quantum states Purnima Ghale, Harley T Johnson Here, we derive a functional form for the energy of quantum many-particle systems from first principles. The trick is to understand the Hamiltonian as a generalized harmonic oscillator and use uncertainty relations to constrain allowed zero-point fluctuations. In particular, we write the expectation value of the many-particle Hamiltonian as the sum of classical and quantum contributions. Then, for a given charge density distribution, fluctuations of the Coulomb field and momentum can be related via the respective commutation of operators. When combined with the Lieb-Thirring bound on kinetic energy, we obtain the energy of interacting many-particle quantum states for non-uniform densities. The approach is applicable to bosonic as well as fermionic systems, and does not require exchange-correlation separation. In the case of uniform density, the functional form agrees with benchmark Quantum Monte Carlo data, including phase transitions. In addition to accuracy, our goal is to develop a deeper appreciation of the first principles at play in many-body quantum systems. (based on arXiv:2010.01656) |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700