Bulletin of the American Physical Society
APS March Meeting 2016
Volume 61, Number 2
Monday–Friday, March 14–18, 2016; Baltimore, Maryland
Session S22: Revealing New Physics With Petascale and Beyond Computational ResourcesFocus
|
Hide Abstracts |
Sponsoring Units: DCOMP Chair: Bogdan Mihalia, National Science Foundation Room: 321 |
Thursday, March 17, 2016 11:15AM - 11:27AM |
S22.00001: Scalable real space pseudopotential density functional codes for materials in the exascale regime Charles Lena, James Chelikowsky, Grady Schofield, Ariel Biller, Leeor Kronik, Yousef Saad, Jack Deslippe Real-space pseudopotential density functional theory has proven to be an efficient method for computing the properties of matter in many different states and geometries, including liquids, wires, slabs, and clusters with and without spin polarization. Fully self-consistent solutions using this approach have been routinely obtained for systems with thousands of atoms. Yet, there are many systems of notable larger sizes where quantum mechanical accuracy is desired, but scalability proves to be a hindrance. Such systems include large biological molecules, complex nanostructures, or mismatched interfaces. We will present an overview of our new massively parallel algorithms, which offer improved scalability in preparation for exascale supercomputing. We will illustrate these algorithms by considering the electronic structure of a Si nanocrystal exceeding 10$^4$ atoms. [Preview Abstract] |
Thursday, March 17, 2016 11:27AM - 11:39AM |
S22.00002: Large DFT: to 100K atoms and beyond. Jonathan Mullin A quantum mechanical (QM) approach to materials science provides a gold standard atomistic picture of the mechanisms responsible for a range of phenomena seen in macroscopic and experimental situations. The need to understand materials science problems from atomistic to macroscale was the impetus for ARL to initiate the Enterprise for Multiscale Material Research. This long term project attempts to redefine how materials science questions are posed, and solved. To support this goal, current state-of-the-art QM capabilities need to be extended in the number of atoms which can be treated and the length scale of the dynamics which can be simulated. This extension is referred to as large scale QM, both large spatially and temporally. This will enable fundamental advances in the understanding of materials science problems. [Preview Abstract] |
Thursday, March 17, 2016 11:39AM - 11:51AM |
S22.00003: Real Space Multigrid (RMG) Open Source Software Suite for Multi-Petaflops Electronic Structure Calculations Emil Briggs, Miroslav Hodak, Wenchang Lu, Jerry Bernholc, Yan Li RMG is a cross platform open source package for ab initio electronic structure calculations that uses real-space grids, multigrid pre-conditioning, and subspace diagonalization to solve the Kohn-Sham equations. The code has been successfully used for a wide range of problems ranging from complex bulk materials to multifunctional electronic devices and biological systems. RMG makes efficient use of GPU accelerators, if present, but does not require them. Recent work has extended GPU support to systems with multiple GPU's per computational node, as well as optimized both CPU and GPU memory usage to enable large problem sizes, which are no longer limited by the memory of the GPU board. Additional enhancements include increased portability, scalability and performance. New versions of the code are regularly released at sourceforge.net/projects/rmgdft/. The releases include binaries for Linux, Windows and MacIntosh systems, automated builds for clusters using cmake, as well as versions adapted to the major supercomputing installations and platforms. [Preview Abstract] |
Thursday, March 17, 2016 11:51AM - 12:27PM |
S22.00004: Understanding Strongly Correlated Materials thru Theory Algorithms and High Performance Computers Invited Speaker: Gabriel Kotliar A long standing challenge in condensed matter physics is the prediction of physical properties of materials starting from first principles. In the past two decades, substantial advances have taken place in this area. The combination of modern implementations of electronic structure methods in conjunction with Dynamical Mean Field Theory (DMFT), in combination with advanced impurity solvers, modern computer codes and massively parallel computers, are giving new system specific insights into the properties of strongly correlated electron systems enable the calculations of experimentally measurable correlation functions. The predictions of this "theoretical spectroscopy" can be directly compared with experimental results. In this talk I will briefly outline the state of the art of the methodology, and illustrate it with an example the origin of the solid state anomalies of elemental Plutonium. [Preview Abstract] |
Thursday, March 17, 2016 12:27PM - 12:39PM |
S22.00005: Large-scale quantum transport calculations for electronic devices with over ten thousand atoms Wenchang Lu, Yan Lu, Zhongcan Xiao, Miro Hodak, Emil Briggs, Jerry Bernholc The non-equilibrium Green’s function method (NEGF) has been implemented in our massively parallel DFT software, the real space multigrid (RMG) code suite. Our implementation employs multi-level parallelization strategies and fully utilizes both multi-core CPUs and GPU accelerators. Since the cost of the calculations increases dramatically with the number of orbitals, an optimal basis set is crucial for including a large number of atoms in the “active device” part of the simulations. In our implementation, the localized orbitals are separately optimized for each principal layer of the device region, in order to obtain an accurate and optimal basis set. As a large example, we calculated the transmission characteristics of a Si nanowire p-n junction. The nanowire is along (110) direction in order to minimize the number dangling bonds that are saturated by H atoms. Its diameter is 3 nm. The length of 24 nm is necessary because of the long-range screening length in Si. Our calculations clearly show the I-V characteristics of a diode, i.e., the current increases exponentially with forward bias and is near zero with backward bias. Other examples will also be presented, including three-terminal transistors and large sensor structures. [Preview Abstract] |
Thursday, March 17, 2016 12:39PM - 12:51PM |
S22.00006: Implementing Parquet equations using HPX Samuel Kellar, Bibek Wagle, Shuxiang Yang, Ka-Ming Tam, Hartmut Kaiser, Juana Moreno, Mark Jarrell A new C++ runtime system (HPX) enables simulations of complex systems to run more efficiently on parallel and heterogeneous systems. This increased efficiency allows for solutions to larger simulations of the parquet approximation for a system with impurities. The relevancy of the parquet equations depends upon the ability to solve systems which require long runs and large amounts of memory. These limitations, in addition to numerical complications arising from stability of the solutions, necessitate running on large distributed systems. As the computational resources trend towards the exascale and the limitations arising from computational resources vanish efficiency of large scale simulations becomes a focus. HPX facilitates efficient simulations through intelligent overlapping of computation and communication. Simulations such as the parquet equations which require the transfer of large amounts of data should benefit from HPX implementations. [Preview Abstract] |
Thursday, March 17, 2016 12:51PM - 1:03PM |
S22.00007: \textbf{Asperities, Crack Front Waves and Crack Self Healing} Pankaj Rajak, Rajiv Kalia, Aiichiro Nakano, Priya Vashishta We have performed petascale simulations to study~nanomaterial systems capable of sensing and repairing damage in high temperature/high pressure operating conditions. The system we have studied is a~ceramic nanocomposite consisting of silicon carbide/silicon dioxide core/shell nanoparticles embedded in alumina. We observe that the interaction of the crack with core/shell asperities gives rise to crack-front waves. We also study crack healing by diffusion of silica into the crack as a function of nanoparticle size and inter-particle distance. Our results are well supported by experimental observations. [Preview Abstract] |
Thursday, March 17, 2016 1:03PM - 1:39PM |
S22.00008: Biophysical Discovery through the Lens of a Computational Microscope Invited Speaker: Rommie Amaro With exascale computing power on the horizon, improvements in the underlying algorithms and available structural experimental data are enabling new paradigms for chemical discovery. My work has provided key insights for the systematic incorporation of structural information resulting from state-of-the-art biophysical simulations into protocols for inhibitor and drug discovery. We have shown that many disease targets have druggable pockets that are otherwise ``hidden'' in high resolution x-ray structures, and that this is a common theme across a wide range of targets in different disease areas. We continue to push the limits of computational biophysical modeling by expanding the time and length scales accessible to molecular simulation. My sights are set on, ultimately, the development of detailed physical models of cells, as the fundamental unit of life, and two recent achievements highlight our efforts in this arena. First is the development of a molecular and Brownian dynamics multi-scale modeling framework, which allows us to investigate drug binding kinetics in addition to thermodynamics. In parallel, we have made significant progress developing new tools to extend molecular structure to cellular environments. Collectively, these achievements are enabling the investigation of the chemical and biophysical nature of cells at unprecedented scales. [Preview Abstract] |
Thursday, March 17, 2016 1:39PM - 2:15PM |
S22.00009: Multiscale Dynamics in Soft-Matter Systems: Enzyme Catalysis, Sec-Facilitated Protein Translocation, and Ion-Conduction in Polymers. Invited Speaker: Thomas Miller Nature exhibits dynamics that span extraordinary ranges of space and time. In some cases, these dynamical hierarchies are well separated, simplifying their understanding and description. But chemistry and biology are replete with examples of dynamically coupled scales. In this talk, we will discuss the use of high-performance computing and new simulation methods that enable the inclusion of nuclear quantum effects, such as zero point energy and tunneling, in the reaction dynamics of enzymes, as well as coarse-graining strategies to enable minute-timescale simulations of protein targeting to cell membranes and ion-conduction in polymer electrolytes for lithium-ion battery applications. [Preview Abstract] |
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