Bulletin of the American Physical Society
APS March Meeting 2014
Volume 59, Number 1
Monday–Friday, March 3–7, 2014; Denver, Colorado
Session Z26: Novel Technologies and Algorithms |
Hide Abstracts |
Sponsoring Units: DCOMP Chair: Jack Wells, Oak Ridge National Laboratory Room: 502 |
Friday, March 7, 2014 11:15AM - 11:27AM |
Z26.00001: A neural network representation of the potential energy surface in Si- and Si-Li systems Brad Malone, Ekin Cubuk, Efthimios Kaxiras A recently developed technique for extending calculations to longer length and time scales is based upon the training of the biologically-inspired neural network machine learning architecture to reproduce the potential energy surface. Trained with ab-initio density functional theory information, such neural networks can reproduce DFT-level accuracy in the study of processes traditionally limited to empirical potentials. We describe progress in constructing accurate neural network potentials for both elemental Si systems as well as Si systems which incorporate Li, the latter being of great current interest because of the promise of Si as an anode material in Li-ion batteries. These potentials allow for the study of interesting phase transformation behavior that occurs in these systems inaccessible by traditional approaches. [Preview Abstract] |
Friday, March 7, 2014 11:27AM - 11:39AM |
Z26.00002: Machine Learning for Dynamical Mean Field Theory Louis-Francois Arsenault, Alejandro Lopez-Bezanilla, O. Anatole von Lilienfeld, P.B. Littlewood, Andy Millis Machine Learning (ML), an approach that infers new results from accumulated knowledge, is in use for a variety of tasks ranging from face and voice recognition to internet searching and has recently been gaining increasing importance in chemistry and physics [1]. In this talk, we investigate the possibility of using ML to solve the equations of dynamical mean field theory which otherwise requires the (numerically very expensive) solution of a quantum impurity model. Our ML scheme requires the relation between two functions: the hybridization function describing the bare (local) electronic structure of a material and the self-energy describing the many body physics. We discuss the parameterization of the two functions for the exact diagonalization solver and present examples, beginning with the Anderson Impurity model with a fixed bath density of states, demonstrating the advantages and the pitfalls of the method.\\[4pt] [1] J. Chem. Theory Comput., 9 3404 (2013) [Preview Abstract] |
Friday, March 7, 2014 11:39AM - 11:51AM |
Z26.00003: Modeling quantum physics with machine learning Alejandro Lopez-Bezanilla, Louis-Francois Arsenault, Andrew Millis, Peter Littlewood, Anatole von Lilienfeld Machine Learning (ML) is a systematic way of inferring new results from sparse information. It directly allows for the resolution of computationally expensive sets of equations by making sense of accumulated knowledge and it is therefore an attractive method for providing computationally inexpensive 'solvers' for some of the important systems of condensed matter physics. In this talk a non-linear regression statistical model is introduced to demonstrate the utility of ML methods in solving quantum physics related problem, and is applied to the calculation of electronic transport in 1D channels. [Preview Abstract] |
Friday, March 7, 2014 11:51AM - 12:03PM |
Z26.00004: Finite Temperature Quantum Effects on Confined Charges Jeffrey Wrighton, Sandipan Dutta, James Dufty The equilibrium density profile for charges confined in a harmonic trap is described for a wide range of temperatures and densities, including the strong coupling classical limit of dusty ion plasmas and low temperature limit of electrons in warm, dense matter. The theoretical description is based on a classical density functional theory (liquid state HNC approximation [1]) using effective quantum charge-charge and confining potentials [2]. Attention is focused on the role of quantum effects in shell formation. These quantum effects range from quantitative modifications of structure due to classical Coulomb correlations to qualitatively different quantum origins of shell structure due to exchange at temperatures below the Fermi temperature. \\[4pt] [1] J. Wrighton, J. W. Dufty, H. K\"{a}hlert, and M. Bonitz, Phys. Rev. E 80, 066405 (2009).\\[0pt] [2] Sandipan Dutta and James Dufty, Phys. Rev. E 87, 032102 (2013); EPL 102 67005 (2013). [Preview Abstract] |
Friday, March 7, 2014 12:03PM - 12:15PM |
Z26.00005: Plasmon Excitations in a Triad of Coulomb-coupled spherical shells Bo Gao, Godfrey Gumbs, Antonios Balassis, Andrii Iurov, Danhong Huang Plasmon modes for a bundle of three spherical of two-dimensional electron gases (S2DEG's) have been obtained within the random-phase approximation (RPA). The inter-sphere Coulomb interaction matrix elements and their symmetry properties were also investigated in detail. The case of a bundle gives an adequate picture of the way in which the Coulomb interaction depends on orbital angular momentum quantum number $L$ and its projection $M$. We concluded that the interaction between the S2DEG's aligned at an angle of $45^0$ with the axis of quantization is negligible compared to the interaction along and perpendicular to the quantization axis. Consequently, the plasmon excitation frequencies reveal an interesting orientational anisotropic coupling to an external EM field probing the charge density oscillations. [Preview Abstract] |
Friday, March 7, 2014 12:15PM - 12:27PM |
Z26.00006: Automated, ab initio calculations of X-ray spectra including many-body excitations and vibrational damping Kevin Jorissen, Shauna Story, John Rehr Accurate calculations of x-ray absorption spectra (XAS) often require linking several materials science codes [1]. To reduce the complexity and support the hardware requirements of such calculations, we have virtualized XAS modeling workflows using a Cloud-based approach, with interfacing and configuration of codes handled by developers, and virtual HPC resources allocated on demand [2]. When coupled to user-friendly GUIs this puts powerful multi-code simulations in the hands of general users. For instance, FEFF users can improve XAS interpretation and analysis using accurate ab initio Debye-Waller factors and self energy from the ABINIT DFT code, rather than semi-empirical models. Additionally, such workflows allow robust automation of large-scale calculation sets such as the Materials Project [3] where our approach could enable a theoretical spectroscopy database of many thousands of structures for systematic study of materials.\\[4pt] [1] Rehr et al., C.R. Phys. 10, 548 (2009).\\[0pt] [2] Jorissen et al., Comp. Phys. Comm. 183, 1922 (2012).\\[0pt] [3] www.materialsproject.org [Preview Abstract] |
Friday, March 7, 2014 12:27PM - 12:39PM |
Z26.00007: Quantitative simulation and density reconstruction in high-energy X-ray radiograph Li Tang, Haibo Xu Numerical radiograph using Monte Carlo method is used to study fidelity of density reconstruction in high-energy X-ray radiography. A density reconstruction method for a polyenergetic X-ray source and an object composed of different materials is proposed. The method includes energy spectrum, angular spectrum and spot size of photon source. And it includes mass absorption coefficients explicitly in density reconstruction as well. A constrained conjugate gradient algorithm and variation regularization are applied to determine material edges and density reconstruction of a French test object. It shows that the method is valid for density reconstruction and energy spectrum of imaging photons is important in obtaining accurate material densities in high-energy X-ray radiograph. [Preview Abstract] |
Friday, March 7, 2014 12:39PM - 12:51PM |
Z26.00008: Compressed modes for variational problems in mathematical physics and compactly supported multiresolution basis for the Laplace operator Vidvuds Ozolins, Rongjie Lai, Russel Caflisch, Stanley Osher We will describe a general formalism for obtaining spatially localized (``sparse'') solutions to a class of problems in mathematical physics, which can be recast as variational optimization problems, such as the important case of Schr\"odinger's equation in quantum mechanics. Sparsity is achieved by adding an $L_1$ regularization term to the variational principle, which is shown to yield solutions with compact support (``compressed modes''). Linear combinations of these modes approximate the eigenvalue spectrum and eigenfunctions in a systematically improvable manner, and the localization properties of compressed modes make them an attractive choice for use with efficient numerical algorithms that scale linearly with the problem size. In addition, we introduce an $L_1$ regularized variational framework for developing a spatially localized basis, compressed plane waves (CPWs), that spans the eigenspace of a differential operator, for instance, the Laplace operator. Our approach generalizes the concept of plane waves to an orthogonal real-space basis with multiresolution capabilities. [Preview Abstract] |
Friday, March 7, 2014 12:51PM - 1:03PM |
Z26.00009: Phase behavior of the 38-atom Lennard-Jones cluster Ray Sehgal, David Ford, Dimitrios Maroudas We have developed a coarse-grained description of the phase behavior of the isolated 38-atom Lennard-Jones cluster (LJ$_{\mathrm{38}})$. The model captures both the solid-solid polymorphic transitions that the cluster undergoes at low temperatures and the complex cluster breakup and melting transitions at higher temperatures. For this coarse model development, we employ the manifold learning technique of diffusion mapping. The outcome of the diffusion mapping analysis over a broad temperature range indicates that two order parameters are sufficient to describe the cluster's phase behavior; we have chosen two such appropriate order parameters that are metrics of condensation and overall crystallinity. In this well-justified coarse-variable space, we calculate the cluster's free energy landscape (FEL) as a function of temperature, employing Monte Carlo umbrella sampling. These FELs are used to quantify the phase behavior and onsets of phase transitions of the LJ$_{\mathrm{38}}$ cluster. [Preview Abstract] |
Friday, March 7, 2014 1:03PM - 1:15PM |
Z26.00010: Solving Hydrogen Systems in a Gaussian-Sinc Mixed Basis Jonathan Jerke, C.J. Tymczak, Young Lee We introduce a Gaussian-Sinc electronic structure mixed basis calculation scheme. We solve the Schr\"{o}dinger's wave equation with a uniform magnetic field for a single electron in Hydrogen Systems in three dimensions. The scheme is inherently unbiased since most of the field is digital under the Sinc basis and the Gaussian are only meant to capture the cusp of atomic states. The entire scheme is translational invariant and the potentials are calculated properly and are necessarily off diagonal. In the absence of magnetic fields the scheme is variational bound. We generally find under arbitrary configurations of protons that we can achieve four significant digits after the decimal. [Preview Abstract] |
Friday, March 7, 2014 1:15PM - 1:27PM |
Z26.00011: Kinetic barriers for Cd and Te adatoms on Cd and Te terminated CdTe (111) surface using {\it ab~ initio} simulations Ebadollah Naderi, Sachin P. Nanavati, Chiranjib Majumder, S.V. Ghaisas In the present work we have calculated using density functional theory (DFT), diffusion barrier potentials on both the CdTe (111) surfaces, Cd terminated (A-type) \& Te terminated (B-type). We employ nudge elastic band method (NEB) for obtaining the barrier potentials. The barrier is computed for Cd and for Te adatoms on both A-type and B-type surfaces. We report two energetically favourable positions along the normal to the surface, one above and other below the surface. The one above the surface has binding energy slightly more the one below. According to the results of this work, binding energy (in all cases) for adatoms are reasonable and close to experimental data. The barrier potential for hopping adatoms (Cd and Te) on both the surfaces is less than 0.35 eV. Apart from these most probable sites, there are other at least two sites on both the types of surfaces which are meta stable. We have also computed barriers for hopping to and from these meta stable positions. The present results can shade light on the defect formation mechanism in CdTe thin films during growth. [Preview Abstract] |
Friday, March 7, 2014 1:27PM - 1:39PM |
Z26.00012: The CONV-3D code for DNS CFD calculation Vladimir Chudanov The CONV-3D code for DNS CFD calculation of thermal and hydrodynamics on Fast Reactor with use of supercomputers is developed. This code is highly effective in a scalability at the high performance computers such as ``Chebyshev'', ``Lomonosov'' (Moscow State University, Russia), Blue Gene/Q(ALCF MIRA, ANL). The scalability is reached up to 10$^{6}$ processors [1]. The code was validated on a series of the well known tests in a wide range of Rayleigh (10$^{6}$-10$^{16}$) and Reynolds (10$^{3}$-10$^{5}$. Such code was validated on the blind tests OECD/NEA of the turbulent intermixing in horizontal subchannels of the fuel assembly at normal pressure and temperature (Matis-H), of the flows in T-junction and the report IBRAE/ANL was published [2]. The good coincidence of numerical predictions with experimental data was reached, that specifies applicability of the developed approach for a prediction of thermal and hydrodynamics in a boundary layer at small Prandtl that is characteristic of the liquid metal reactors.\\[4pt] [1] V.V. Chudanov et al. The national supercomputer forum (NSCF-2013). Pereslavl-Zalesskii, 26-29 November, 2013.\\[0pt] [2] A.V. Obabko, P.F. Fischer, et al. ISBN 978-953-51-0987-7, Published: February 13, 2013 under CC BY 3.0 license. DOI: 10.5772/53143. 2013. [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