Bulletin of the American Physical Society
APS March Meeting 2018
Volume 63, Number 1
Monday–Friday, March 5–9, 2018; Los Angeles, California
Session C34: Petascale Science and Beyond: Applications and Opportunities for Materials, Chemical, and Bio Physics IIIFocus
|
Hide Abstracts |
Sponsoring Units: DCOMP DBIO DCP DCMP Chair: Jack Deslippe, Lawrence Berkeley National Laboratory Room: LACC 409A |
Monday, March 5, 2018 2:30PM - 3:06PM |
C34.00001: Large-scale first principles calculations with leadership class HPC using many-body perturbation theory Invited Speaker: Marco Govoni Predictive modeling of excited state properties typically employs Green's function based many-body perturbation theory (MBPT) methods. Pre-exascale machines offer the opportunity to expand the science domain of such methods to systems of unprecedented size, accounting for the realistic complexity of e.g. nanostructured, interfaced, disordered, and defective materials. In this talk we discuss how methodological advances coded in WEST [www.west-code.org] provide an efficient formulation of electron-electron, electron-phonon and electron-hole interactions, that is applied to systems with thousands of electrons. We will discuss the advantages of the algorithms used in WEST over standard techniques, its parallel performance on leadership class HPC machines, and provide results for the calculation of spectroscopic features of liquids and defective solids. We will discuss the inclusion of electron-phonon coupling effects to simulate photoelectron spectra and carrier lifetimes of carbon-based nanoparticles. Finally, we will present the new functionalities enabled by the concurrent use of WEST and the Qbox code [qboxcode.org], with focus on the interoperability paradigm that the coupling advocates. Work in collaboration with: A. Gaiduk, G. Galli, M. Gerosa, F. Gygi, I. Hamada, C. Knight, H. Ma, R. McAvoy, N.L. Nguyen, H. Yang, H. Seo, H. Zheng. |
Monday, March 5, 2018 3:06PM - 3:18PM |
C34.00002: Massively Parallel Real-Time TDDFT Simulations of Electronic Excitations in DNA by Energetic Protons Dillon Yost, Yosuke Kanai In electronic stopping processes, electronic excitations are produced by fast-moving ions. This phenomenon is at the heart of DNA damage via ion irradiation, which is central to ion beam cancer therapy. A molecular-level understanding of this process is necessary to advance medical technology. Unfortunately, experimental studies remain scarce due to their requirement for extensive use of synchrotron facilities. Quantum-mechanical simulations on supercomputers offer an alternative route for investigating this problem. We will discuss results acquired from our highly scalable implementation of real-time time-dependent density functional theory (RT-TDDFT) simulations (1,2) on current generation HPC systems, as well as next-generation many-integrated-core architectures. This cutting-edge application of RT-TDDFT simulations has unveiled details of energy transfer, electronic excitations, and charge fluctuations in DNA under ion irradiation that would be unattainable without petascale simulations. |
Monday, March 5, 2018 3:18PM - 3:30PM |
C34.00003: Petascale simulation of electron currents in nanoscale interconnects Lin-Wang Wang The on-die interconnect has become the dominate heat generating source in a modern chip. As the transistor size gets even smaller, or the device architecture changes to 3D stacking, the interconnect problem will only become worse. Petascale high performance computing is used to study the current flow in such a nanoscale interconnect. Ab initio simulation is used to construct the Schrodinger’s equation for a system with tens of thousands of copper atoms based on linear scaling density functional theory. The electron scattering states are calculated using a linear combination of perturbation induced “system” states. The direct viewing of the intensity of such scattering states (current) will show some insights for where the heat is generated. Future calculation of electron-phonon coupling will allow the calculation of heat generation. The challenges of the simulations will be discussed, as well as the computational issues in such large scale simulation. |
Monday, March 5, 2018 3:30PM - 3:42PM |
C34.00004: Improved Molecular Dynamics Calculations using High-Order Forces in Real Space Density Functional Theory Joshua Neitzel, Charles Lena, James Chelikowsky We present an improved method for calculation of nonlocal contributions to interatomic forces resulting from the real space density functional theory Hamiltonian. Our method utilizes a high order Gaussian integration scheme. We demonstrate the efficacy of our method through molecular dynamics simulations of polyatomic molecules. We show that our method reduces energy drift for each molecule. Additionally, our method improves convergence of dynamic variables including center of mass drift and vibrational frequency. |
Monday, March 5, 2018 3:42PM - 3:54PM |
C34.00005: Scaling the Force Calculations of the Real Space Pseudopotential DFT solver PARSEC on Haswell and KNL systems Kevin Gott, Charles Lena, Kai-Hsin Liou, James Chelikowsky, Jack Deslippe The ability to compute atomic forces through quantum contributions rather than through simple pairwise potentials is one of the most compelling reasons materials scientists use Kohn-Sham pseudopotential density functional theory (DFT). PARSEC is an actively developed real space pseudopotential DFT solver that uses Fortran MPI+OpenMP parallelization. PARSEC provides atomic forces by self-consistently solving for the electronic structure and then summing local and nonlocal contributions. Through experimentation with PARSEC, we present why increasingly bulk synchronous processing and vectorization of the contributions is not enough to fully utilize current HPC hardware. We address this limitation through a demonstration of multithreaded communication approaches for local and nonlocal force computations on Intel Knights Landing supercomputers that yield feasible calculation times for systems of over 20,000 atoms. |
Monday, March 5, 2018 3:54PM - 4:06PM |
C34.00006: A parallel eigensolver: Using spectrum slicing method to solve the KohnSham problem for large systems Kai-Hsin Liou, Charles Lena, James Chelikowsky, Yousef Saad Solving the Kohn–Sham equation within density functional theory (DFT) finds extensive use in solving for the electronic structure of a variety of condensed materials, including complex biomolecules, nanostructures and interfacial systems. PARSEC, a real-space pseudopotential DFT code developed by our group, routinely solves the electronic structure problem for localized and extended systems of thousands of atoms using Chebyshev polynomial subspace filtering; however, many systems of interest contain tens of thousands of atoms, where the computational demands of orthonormalization can overwhelm current machines. In an effort to address this numerical constraint, we add an additional level of parallelism to our eigensolver by focusing on spectrum slicing. Spectrum slicing assigns energy windows to eigenpairs and treats each window as an independent task, which may be assigned to groups of processing units. We will demonstrate the advantages of this approach for a large (over 20,000 atoms) silicon nanocrystal with and without defects. |
Monday, March 5, 2018 4:06PM - 4:18PM |
C34.00007: Towards Abinit on ExaScale supercomputers : the challenge for electronic structure physicists. Jordan Bieder, Marc Torrent, Yohan Chatelain In Abinit, physical properties are computed from the state-of-the art first principles methods. |
Monday, March 5, 2018 4:18PM - 4:30PM |
C34.00008: Unified Access To Kohn-Sham DFT Solvers for Different Scales and HPC: The ELSI Project William Huhn, Alberto Garcia, Luigi Genovese, Ville Havu, Mathias Jacquelin, Weile Jia, Murat Keceli, Raul Laasner, Yingzhou Li, Lin Lin, Jianfeng Lu, Stephan Mohr, Nkwe Monama, Happy Sithole, Alvaro Vazquez-Mayagoitia, Chao Yang, Haizhao Yang, Victor Yu, Volker Blum One major bottleneck in Kohn-Sham DFT is the solution of the Kohn-Sham eigenvalue equation to obtain the density matrix, which limits achievable system size. In this talk, we present updates to the ELectronic Structure Infrastructure (ELSI), an open-source effort that facilitates access to efficient Kohn-Sham solvers via a unified, code-independent interface for different system size scales, system types, and HPC architectures. Features that will be presented include support for several solver libraries (ELPA, libOMM, PEXSI, SIPs, CheSS) and a flexible, massively-parallel density-matrix-based restart method. Comparative benchmarks up to ten thousands of atoms on leading HPC resources will be shown. Special attention will be paid to the performance and scaling benchmark set created for this work, which is designed to be applicable to other aspects of electronic structure theory. These benchmarks will be the foundation of a decision layer to automatically select the optimal solver for a given problem based on the performance profile of the underlying HPC architecture. |
Monday, March 5, 2018 4:30PM - 4:42PM |
C34.00009: RMG – An Open Source Electronic Structure Code for Multi-Petaflops Calculations Emil Briggs, Wenchang Lu, Miroslav Hodak, Jerry Bernholc Density functional theory provides an accurate but computationally demanding framework for studying the properties of a broad range of problems. RMG is an open source package that solves the Kohn-Sham equations that arise from the DFT formalism using real-space grids, multigrid pre-conditioning, and subspace diagonalization. RMG is well suited for petascale level calculations and makes efficient use of GPU accelerators on platforms that support them. It is cross platform and runs on Linux, Windows and MacIntosh systems as well as clusters and supercomputers. Applications range from semiconductors to biological systems. RMG provides support for a variety of exchange correlation functionals including Van der Waals interactions via vdW-df. The base code uses delocalized orbitals but a localized orbital variant is also available. It is used for quantum transport calculations via the non-equilibrium Green’s function formalism, by which self-consistent calculations for over 10,000 atoms have been performed. |
Monday, March 5, 2018 4:42PM - 4:54PM |
C34.00010: Full-potential fully relativistic LSMS Method Xianglin Liu, Markus Eisenbach, Yang Wang, G. Malcolm Stocks The locally self-consistent multiple scattering (LSMS) method is an electronic structure calculation method based on the multiple scattering theory (MST) and density functional theory (DFT). As an O(N) approach, the LSMS method makes use of the "nearsightedness" principle, and consider the multiple-scattering process only within the "local interaction zone" (LIZ), therefore is highly parallelizable and very efficient to treat large systems. Traditionally, the LSMS method adopts the muffin-tin approximation, and has been successfully used to treat alloy systems with tens of thousands of atoms. For a more general application of the LSMS method, such as treating directional covalent bonding, 2D structures, or evaluating the interatomic forces, it is desirable to implement the full potential scheme instead of the muffin-tin approximation. Moreover, due the crucial role of spin-orbit coupling in many technological applications, it is important to extend the LSMS method to a relativistic scheme. In this talk, we will demonstrate our implementation of the LSMS method that solves the full-potential Dirac equation directly, show example calculations, and discuss the results. |
Monday, March 5, 2018 4:54PM - 5:06PM |
C34.00011: Opportunities for Simulating Large Systems using Accurate and Efficient Density Functional Theory Calculations Laura Ratcliff, Stephan Mohr, Michel Masella, Luigi Genovese Thanks to a combination of the availability of petascale supercomputers and algorithmic developments such as linear scaling density functional theory, in recent years it has become possible to perform quantum mechanical (QM) simulations for systems containing several thousand atoms. This not only opens up new possibilities for treating large and complex systems using QM, but also enables the direct comparison with non-QM methods such as force fields or coarse graining methods, at the (large) length scales at which they are typically applied. To this end, it is desirable to have a means of reducing the effective complexity of a QM simulation, i.e. to reduce the number of degrees of freedom without a significant loss of accuracy or the introduction of any bias. For such a complexity reduction, it is necessary to identify other analysis schemes for understanding and interpreting the behaviour of a QM system, for example by partitioning a large system into objects, or fragments, of a smaller size. In this talk we will discuss the opportunities for large scale QM simulations, and introduce a quantitative method for assessing the partitioning of a system into fragments, presenting examples of how such a partitioning might be exploited. |
Monday, March 5, 2018 5:06PM - 5:18PM |
C34.00012: Proper alignment of Fermi Level in a Tri-anionic Molecule Der-you Kao, Shawn Domagal-Goldman Descriptions of anionic molecules, within approximations to density-functional theory, has been problematic due to self-interaction error which pushes the eigenvalues to higher energy and creates non-systematic errors in Fermi levels. A new approach for improving the Perdew-Zunger self-interaction[A], referred to as the Fermi-Löwdin orbital-based self-interaction correction (FLO-SIC), is introduced to treat a tri-anionic system. The Fermi–Löwdin orbitals are localized but due to the fact they are constructed directly from the spin-density matrices, they lead to a set of orbitals that have the property that each and every orbital is explicitly invariant to unitary transformations which leads to a unitarily invariant and size-extensive expression for the energy. The improvement of alignment of Fermi level in the system is demonstrated and related isotope fractionation is reported. Zero-point splittings will also be presented and the impact of self-interaction correction will be discussed. |
Monday, March 5, 2018 5:18PM - 5:30PM |
C34.00013: Rigorous definition of built-in potentials at interfaces Duk-Hyun Choe, Damien West, Shengbai Zhang The built-in potential is the interfacial potential difference due to electric dipole at the interface of two dissimilar materials. It is of central importance to the understanding of many phenomena in interface/surface science because it determines the band alignment between the materials. Despite more than 50 years of investigation, however, its exact sign and magnitude have generally been generally recognized as an ill-defined quantity. In this work, we provide a universal definition of the built-in potential. We find that the built-in potential is explicitly determined by the bulk (i.e., innate) properties of the constituent materials when the system is in electronic equilibrium, while it is subject to the properties of the interface only in the absence of equilibrium. Our theory enables a unified description of a variety of important properties in all types of interfaces, ranging from work functions and Schottky barriers in electronic devices to electrode potentials and redox potentials in electrochemical cells. |
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