Bulletin of the American Physical Society
APS March Meeting 2018
Volume 63, Number 1
Monday–Friday, March 5–9, 2018; Los Angeles, California
Session A34: Petascale Science and Beyond: Applications and Opportunities for Materials, Chemical, and Bio Physics IFocus
|
Hide Abstracts |
Sponsoring Units: DCOMP DCMP DMP DCP Chair: Nichols Romero, Leadership Computing Facility, Argonne National Laboratory Room: LACC 409A |
Monday, March 5, 2018 8:00AM - 8:36AM |
A34.00001: Accurate, scalable computations in many-electron systems Invited Speaker: Shiwei Zhang One of the grand challenges in materials physics and chemistry is the accurate treatment of interacting many-electron systems. Computational methods need to reach beyond the incredible success afforded by the Kohn-Sham density functional theory (KS-DFT), and its independent-electron and perturbative extensions. This is difficult because of the combinatorial growth of the dimension of the Hilbert space involved, along with the high degree of entanglement produced by the combination of Fermi statistics and electron-electron interactions. Progress in addressing this challenge will be fundamental to the realization of “materials genome” or materials by design initiatives. Recently, significant advances have been achieved by the combination of methodological and algorithmic developments with petascale computing, which has lead to the solution of a number of challenging problems. I will give a brief overview of these advances, and then discuss one of the computational frameworks that have played an important role in them, the auxiliary-field quantum Monte Carlo method. The framework can be viewed as a superposition of KS-DFT systems evolving in fluctuating auxiliary fields, which are treated by stochastic sampling. The approach has demonstrated excellent accuracy across a wide range of quantum many-electron systems; it has computational complexity that scales as a low polynomial of the number of electrons; and it is ideally suited for high-performance computing platforms. A few illustrative examples will be discussed from recent applications in physics and chemistry. |
Monday, March 5, 2018 8:36AM - 8:48AM |
A34.00002: Stacking Interactions of B-DNA and Their Nonadditivities: A QMC Study Ken Qin, Ryo Maezono, Kenta Hongo Noncovalent interactions play significant roles in structures and functionality of biomolecules, but their accurate evaluation remains challenge in first-principles simulations. Fixed-node diffusion Monte Carlo (FNDMC) is one of the most promising approaches to tackle this problem in terms of computational costs and accuracy: It is known to be more accurate than MP2 (the second-order Møller-Plesset) and as accurate as "gold standard" quantum chemistry, CCSD(T)/CBS (coupled cluster including single, double, and noniterative triple excitations with complete basis set). Furthermore, FNDMC has a more desirable computational scaling (N3-4) with respect to the number of electrons (N), compared to CCSD(T)/CBS (N7). With the help of modern supercomputers, we have applied FNDMC to ten unique B-DNA base-pair steps, for the first time, in order to evaluate their stacking energies. In addition, we evaluated their nonadditive (many-body effects) interaction energies (defined as the total stacking energy minus the sum of pair interaction energies) beyond the MP2 level (identical to SCF), which has not been investigated so far at the CCSD(T)/CBS level. Intriguingly, it is found that our “correlated” nonadditivities drastically change from the SCF/MP2 ones. |
Monday, March 5, 2018 8:48AM - 9:00AM |
A34.00003: Calculating the force constant matrix via quantum Monte Carlo Yu Yang Liu, Bartholomew Andrews, Gareth Conduit Quantum Monte Carlo methods have become a leading contender for high accuracy calculations for the electronic structure. Calculating energy derivatives such as atomic forces and the matrix of force constants is important in relaxing structures, calculating vibrational properties, and performing molecular dynamics simulations. We develop a quantum mechanical expectation value to evaluate the matrix of force constants directly in Quantum Monte Carlo. The approach allows the full modeling of the Van der Waals force, opening new applications to molecules and solids in condensed matter. |
Monday, March 5, 2018 9:00AM - 9:12AM |
A34.00004: Towards the solution of the many-electron problem in real materials:
equation of state of the hydrogen chain with state-of-the-art many-body methods Mario Motta, David Ceperley, Garnet Chan, John Gomez, Emanuel Gull, Sheng Guo, Carlos Jiménez-Hoyos, Tran Lan, Jia Li, Fengjie Ma, Andrew Millis, Nikolai Prokof'ev, Ushnish Ray, Gustavo Scuseria, Sandro Sorella, Edwin Stoudenmire, Qiming Sun, Igor Tupitsyn, Steven White, Dominika Zgid, Shiwei Zhang We present numerical results for the equation of state of an infinite chain of hydrogen atoms. A |
Monday, March 5, 2018 9:12AM - 9:24AM |
A34.00005: Diffusion Monte Carlo Re-evaluation of Bulk Aluminium with QMCPACK Adie Hanindriyo, Hyeondeok Shin, Anouar Benali, Ryo Maezono In the past decades, diffusion Monte Carlo (DMC) has proven to be one of the most accurate ab initio methods for molecules and solids. Its parallel algorithm enables efficient use of High Performance Computing (HPC) facilities around the world. In 2012, a DMC study on bulk aluminium with the CASINO code developed by the Theory of Condensed Matter group in University of Cambridge[1] was one of the first calculations on a solid using modern algorithms, corrections to finite size effects, and time step extrapolation. Good agreement with experimental data was achieved, exceeding by far, those obtained by Density Functional Theory (LDA and GGA exchange correlation functionals). |
Monday, March 5, 2018 9:24AM - 9:36AM |
A34.00006: Accelerating Quantum Monte Carlo via Graphics Processing Units Benjamin Himberg, Adrian Del Maestro The Calogero-Sutherland model describes a system of particles on a ring with weakly decaying inverse-distance-square-law interactions. The interaction potential must be treated as long range and thus exact simulation studies of its finite temperature properties have been limited to small systems consisting of few particles. We report on a new path Integral quantum Monte Carlo algorithm that exploits the extreme parallelism of Graphics Processing Units to dramatically reduce the computational time of simulations. Speedup factors of up to 500 times allow us to explore finite temperature observables for a range of interaction strengths. While the model itself is of great interest due to its exactly solvable nature, the algorithm is model agnostic and may be particularly useful when applied to experimentally relevant models of high density superfluids. |
Monday, March 5, 2018 9:36AM - 9:48AM |
A34.00007: Large-Scale GW Calculations on Pre-Exascale HPC Systems Mauro Del Ben, Felipe H. da Jornada, Andrew Canning, Nathan Wichmann, Karthik Raman, Ruchira Sasanka, Chao Yang, Steven Louie, Jack Deslippe Large-scale GW calculations are required to accurately describe excited state phenomena in materials, which is critical for the design of novel devices in many fields. However, application of the GW method to complex systems is often limited due to the high computational cost. Reduced time to solution can be achieved through novel methods, algorithms and optimal implementations on modern HPC systems. We demonstrate these capabilities utilizing a highly-optimized version of the BerkeleyGW software package for HPC many-core architectures. The developed code, tested on Cori@NERSC (a Cray XC40, Xeon-Phi powered system), is capable of scaling to the full-machine, using a high fraction of peak performance and achieving excellent time to solution for systems of thousands of atoms. A high fraction of peak and good parallel scaling comes from an improved data layout where the computationally intensive work at the node level is cast as large ZGEMM operations, combined with a ring-based communication scheme, which avoids collective operations and allows overlapping with computation. |
Monday, March 5, 2018 9:48AM - 10:00AM |
A34.00008: Tuning the Polaronic Properties of Hybrid Organic-Inorganic Perovskites CH3NH3PbI3 by Mixing Cation Liujiang Zhou, Amanda J. Neukirch, Jacky Even, Claudine Katan, Sergei Tretiak Formation of polaron quasiparticles has been proved to be very important for describing the charge carrier behavior in hybrid organic-inorganic perovskites. Hybrid organic-inorganic perovskite materials with mixed cations have made tremendous achievement in the enhancement of power conversion efficiency and improved perovskite stability. Inspired by this point, here we report the results from computationally intensive studies of formation energies, electronic structure, charge density and tunable polaronic properties via mixing cations by Cs and FA ions in MAPbI3 (MA = CH3NH3) based on periodic boundary conditions and isolated structures. Our results show that the introduction of an individual cesium (Cs) substitution enhance the ambient stability, in contrast to the decreased stability by doping one formamidinium (FA = HC(NH2)2) cation. Importantly, the cations-alloying significantly reduces the polaron binding energies both for holes and electrons, about 2-times decrease compared to pure MAPbI3. Such reduced reorganization energies suggest the appearance of enlarged polarons in HOP materials with delocalized spin density distribution and finally attenuate electron-phonon coupling guaranteeing efficient charge separation. |
Monday, March 5, 2018 10:00AM - 10:12AM |
A34.00009: HPC in Polymeric Materials: Unraveling the Morphologies of Miktoarm Star Terpolymers Monojoy Goswami, Matthias Arras, Gregory Smith Recent developments in high performance computing are quickly shaping the precise understanding of polymer self-assembly at the molecular level. In this talk I will present coarse-grained molecular dynamics simulations of a set of miktoarm star polymer using LAMMPS MD package. The use of HPC and LAMMPS helped us simulate systems comparable to experiments. Our simulations were able to reproduce results obtained from experiments using transmission electron microscopy (TEM) and small-angle neutron (SANS) and X-ray (SAXS) scattering. While coarse-grained simulations often provide approximate results, we were able to reproduce SANS and SAXS data by simulating large systems and systematically developing calculation strategies for scattering experiments. |
Monday, March 5, 2018 10:12AM - 10:24AM |
A34.00010: Creation of defects in nanograin tungsten Yang Zhang, Jason Trelewicz, Predrag Krstic Grain boundaries play an important role in nanocrystalline material properties. The present molecular dynamics study is focused to the role of the grain boundaries in radiation damage. Computer simulations of cumulative self-atom irradiation were conducted for 4 different types of grain boundary configurations (Σ3<110>{112}, Σ3<110>{111}, Σ5<100>{130}, Σ5<100>{120}), at impact energy of 1 keV, and for two temepratures, 300 K ad 1000 K. Five different interatomic potentials for tungsten were applied to determine the most suitable one for our simulation conditions. The size of the target tungsten structure was 160,000 atoms. The calcuations were performed at supercomputing facilities of NCCS (Titan, EOS) and of XSEDE (Stampede 2 and Comet). The analysis of the surface morphology, number and structure of deffects, sputtering and implantation revealed the mechanism of the direct and indirect mutual interactions of the collision cascades and grain boundaries. |
Monday, March 5, 2018 10:24AM - 10:36AM |
A34.00011: Three-Orbital Spin-Fermion Model for CuO2 Planes Mostafa Hussein, Christopher Bishop, Elbio Dagotto, Adriana Moreo High Tc superconducting cuprates have been studied using single orbital Hubbard or t-J models because numerical studies of more realistic multiorbital Hamiltonians could not be done in sufficiently large systems. The use of these simplified models was justified by the experimental observation of a single band Fermi surface. However, the single band models are Mott insulators in the undoped state while it is well-known that the parent compound of the cuprates are charge-transfer insulators. The discovery of the iron-based superconductors brought to the forefront the need to develop models and numerical approaches for multiorbital systems. Thus, effective multiorbital spin-fermion models were developed [1] that allowed the study of many properties of these materials. A spin-fermion model for the CuO2 planes of the cuprates will be presented.[2] Using techniques developed for the pnictides results will be presented for magnetic and charge ordering, pairing tendencies, spectral functions and density of states as a function of temperature and doping. Comparisons with experimental data for the real materials and numerical results for single orbital models will be presented. |
Monday, March 5, 2018 10:36AM - 10:48AM |
A34.00012: Angle-Adjustable Phase Field Crystal Method for Modeling Crystalline Microstructures Zile Wang, Zhi-Rong Liu, Zhi Feng Huang The phase field crystal (PFC) method has emerged as an attractive continuum modeling approach which resolves system microstructures and dynamics on atomic length and diffusive time scales and effectively addresses the elastoplastic properties of the system. Most PFC models are constructed for systems of isotropic interactions, with lattice symmetry controlled by microscopic length scales, and thus are lack of angle dependency which is important for a broad range of material systems with directional interaction or bonding. Here we develop a new PFC model incorporating both the anisotropic property of angular dependence and the rotational invariance of the whole system. Our analysis is based on the property of isotropic Cartesian tensor and the complete Fourier expansion of any n-point direct correlation function that satisfies the rotational invariance condition. Applications of this new PFC model include some examples of 3D structure modeling (such as simple cubic and diamond cubic phases) via a single length scale and angle-dependent effects, and importantly, the achieving of continuous angle control in crystalline structures (such as the rhombic phase), which demonstrates the advantage of this angle-adjustable density field approach. |
Monday, March 5, 2018 10:48AM - 11:00AM |
A34.00013: Ground state critical exponents in disordered XYZ spin chain Brenden Roberts, Olexei Motrunich We perform a study of the XYZ spin chain model with random independently distributed couplings. We focus on the critical line between x and y Ising antiferromagnetic phases, given by identical x and y distributions with a z-coupling distribution of varying bandwidth. In this way one can tune between the free-fermion fixed point and the multicritical point with identically distributed couplings in all directions. Using a new unbiased numerically exact method called rigorous renormalization group (RRG) we target the ground state physics, focusing on evaluating critical exponents, which have been proposed to vary continuously along the critical line between the XY and statistically isotropic XYZ points based on quantities computed using strong disorder RG schemes. Similarly, we investigate the critical line between the U(1) symmetric XX point and the anisotropic XY point. |
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