Bulletin of the American Physical Society
APS March Meeting 2017
Volume 62, Number 4
Monday–Friday, March 13–17, 2017; New Orleans, Louisiana
Session C26: Theoretical Chemical Physics |
Hide Abstracts |
Sponsoring Units: DCP Chair: Joel Yeun-Zhou, University of California, San Diego Room: 289 |
Monday, March 13, 2017 2:30PM - 2:42PM |
C26.00001: Experimental and Computational study of azobenzene and 2,2',6,6'-tetrafluoroazobenzene cation. Mohammadreza Rezaee, Peter B. Armentrout The electronic structure of the protonated azobenzene and it its derivative 2,2',6,6'-tetrafluoroazobenzene were studied using ab initio methods and the bond strength were measured utilizing the collision induced dissociation experiment. Several highly accurate multi-level schemes such as different variations of the Complete Basis Set (CBS) method and the Gaussian (G-n) theory along with DFT study employed to accurately compute the energies of the neutral and the parent cation as well as the fragment ions. The transition state were studied and the dissociation path was identified using B3LYP method along with aug-cc-pVTZ as the basis set. Thermochemical properties such as proton affinity, gas phase basicity and the bond dissociation energies were calculated. Molecular electrostatic potential analysis was performed to identify the charge distribution inside the molecule to study the effects of the protonation reaction. [Preview Abstract] |
Monday, March 13, 2017 2:42PM - 2:54PM |
C26.00002: Quantitative characterization of the errors of the 3d-transition-metal pseudopotentials in Diffusion Monte Carlo Allison Dzubak, Jaron Krogel, Fernando Reboredo Using a recently proposed extrapolation scheme and multideterminent wavefunctions, we estimate the errors of two families of non-local pseudopotentials of the first row transition metal atoms Sc-Zn. Jastrow sensitivities are presented to assess the quality of two sets of pseudopotentials with respect to localization error reduction. The locality approximation and T-moves scheme are also compared for accuracy of total energies. After removal of the locality and T-moves errors, we present the range of fixed-node energies between a single determinant description and a full valence multideterminant CAS expansion. The results presented here corroborate previous findings that the locality approximation is less sensitive to changes in the Jastrow than T-moves, yielding more accurate total energies. For both the locality approximation and T-moves, we find decreasing Jastrow sensitivity moving left to right across the series Sc-Zn. The recently generated pseudopotentials of Krogel et al. reduce the magnitude of the localization error compared with the pseudopotentials of Burkatzki et al. For the Sc-Zn atomic series with these pseudopotentials, and using up to three-body Jastrows we find that the fixed-node error is dominant over the localization error. [Preview Abstract] |
Monday, March 13, 2017 2:54PM - 3:06PM |
C26.00003: Asymptotic analysis of atomic correlation energies and the generalized gradient approximation Antonio Cancio, Kieron Burke, Tim Gould It has long been known that the non-relativistic ground-state energy in Thomas-Fermi theory becomes relatively exact in the high-density, large particle number limit typified by the atomic number $Z \to \infty$ limit of neutral atoms. The analysis of this limit provides a unified approach to the explicit construction of density functionals, inspiring advances in understanding kinetic and exchange energy functionals. Recent benchmark calculations of atomic correlation energies allow us to extend this analysis to correlation. Asymptotic extrapolation gives a correlation energy of the form $-AZ\log{Z} + BZ$ with $A$ a known universal quantity, and $B$ about 38 millihartrees. The PBE functional, derived in part from an analysis of the high density limit, has remarkably good scaling behavior with $Z$, but fails to predict this limit quantitatively. We re-parametrize the high density limit of the PBE for finite levels of inhomogeneity to construct an asymptotically corrected GGA. This reparametrization captures most but not all of the asymptotic trend in atomic benchmark data with the remainder mostly captured by the correct treatment of the density dependence of the gradient expansion for correlation. The results compare favorably at all $Z$ to empirical functionals. [Preview Abstract] |
Monday, March 13, 2017 3:06PM - 3:18PM |
C26.00004: Method for the Direct Solve of the Many-Body Schr\"{o}dinger Wave Equation Jonathan Jerke, C. J. Tymczak, Bill Poirier We report on theoretical and computational developments towards a computationally efficient direct solve of the many-body Schr\"{o}dinger wave equation for electronic systems. This methodology relies on two recent developments pioneered by the authors: 1) the development of a Cardinal Sine basis for electronic structure calculations [arXiv:1405.5073; Jerke, JCP 2015 143]; and 2) the development of a highly efficient and compact representation of multidimensional functions using the Canonical tensor rank representation developed by Belykin et. al. [SIAM 2005 26(6)] which we have adapted to electronic structure problems. We then show several relevant examples of the utility and accuracy of this methodology, scaling with system size, and relevant convergence issues of the methodology. [Preview Abstract] |
Monday, March 13, 2017 3:18PM - 3:30PM |
C26.00005: Hybrid Grid and Basis Set Approach to Quantum Chemistry DMRG Edwin Miles Stoudenmire, Steven White We present a new approach for using DMRG for quantum chemistry that combines the advantages of a basis set with that of a grid approximation. Because DMRG scales linearly for quasi-one-dimensional systems, it is feasible to approximate the continuum with a fine grid in one direction while using a standard basis set approach for the transverse directions. Compared to standard basis set methods, we reach larger systems and achieve better scaling when approaching the basis set limit. The flexibility and reduced costs of our approach even make it feasible to incoporate advanced DMRG techniques such as simulating real-time dynamics. [Preview Abstract] |
Monday, March 13, 2017 3:30PM - 3:42PM |
C26.00006: Leading approximations to local corrections II: The case with turning points Raphael Ribeiro, Kieron Burke Semiclassical uniform approximations are employed to study the leading energetic corrections to Thomas-Fermi theory for 1d noninteracting fermions coupled to a confining potential $v(x)$ in the semiclassical limit. Novel universal analytical results are given on the leakage of particle density beyond classical turning points and the resulting energetic consequences are derived in the semiclassical limit. These are confirmed by a systematic numerical study of the semiclassical limiting behavior of the global, regional and pointwise properties of fermions in a diverse set of potentials including double wells. Singular situations where the semiclassical approximation breaks down are verified. The connection to DFT developments will also be discussed.\\ \\ R.F Ribeiro and K. Burke. Leading corrections to local approximations II (with turning points) {\em arXiv}:1611.00881 [Preview Abstract] |
Monday, March 13, 2017 3:42PM - 3:54PM |
C26.00007: Computationally Efficient Characterization of Potential Energy Surfaces Based on Fingerprint Distances Bastian Schaefer, Stefan Goedecker Based on Lennard-Jones, Silicon, Sodium-Chloride and Gold clusters, it was found that uphill barrier energies of transition states between directly connected minima tend to increase with increasing structural differences of the two minima. Based on this insight it also turned out that post-processing minima hopping data at a negligible computational cost allows to obtain qualitative topological information on potential energy surfaces that can be stored in so called qualitative connectivity databases. These qualitative connectivity databases are used for generating fingerprint disconnectivity graphs that allow to obtain a first qualitative idea on thermodynamic and kinetic properties of a system of interest. [Preview Abstract] |
Monday, March 13, 2017 3:54PM - 4:06PM |
C26.00008: Deep learning and the electronic structure problem Kyle Mills, Michael Spanner, Isaac Tamblyn In the past decade, the fields of artificial intelligence and computer vision have progressed remarkably. Supported by the enthusiasm of large tech companies, as well as significant hardware advances and the utilization of graphical processing units to accelerate computations, deep neural networks (DNN) are gaining momentum as a robust choice for many diverse machine learning applications. We have demonstrated the ability of a DNN to solve a quantum mechanical eigenvalue equation directly, without the need to compute a wavefunction, and without knowledge of the underlying physics. We have trained a convolutional neural network to predict the total energy of an electron in a confining, 2-dimensional electrostatic potential. We numerically solved the one-electron Schr\"{o}dinger equation for millions of electrostatic potentials, and used this as training data for our neural network. Four classes of potentials were assessed: the canonical cases of the harmonic oscillator and infinite well, and two types of randomly generated potentials for which no analytic solution is known. We compare the performance of the neural network and consider how these results could lead to future advances in electronic structure theory. [Preview Abstract] |
Monday, March 13, 2017 4:06PM - 4:18PM |
C26.00009: Towards efficient coupled-cluster theories for periodic systems Theodoros Tsatsoulis, Felix Hummel, Andreas Grueneis Over the last few years, quantum-chemical correlation methods have been increasingly often applied on extended systems. In this work we consider an ab-initio description of the true many-body wave function. We explore canonical coupled-cluster theory within the projector-augmented-wave method in a plane-wave basis. A combination of Gaussian basis-functions with plane-waves, as well as a low-rank factorization of the Coulomb integrals results in an effective quantum-chemical scheme for extended systems. We demonstrate the capabilities of the methods by studying molecular interactions with periodic surfaces. [Preview Abstract] |
Monday, March 13, 2017 4:18PM - 4:30PM |
C26.00010: Computing observables and correlation functions of molecular systems with auxiliary-field quantum Monte Carlo Mario Motta, Shiwei Zhang The quantitative study of correlated materials requires accurate and efficient calculations of electronic density, forces and correlation functions. To achieve this goal, we formulated and implemented a back-propagation scheme \footnote{S. Zhang, J. Carlson and J. E. Gubernatis, {\em{Phys. Rev. B}} {\bf{55}}, 7464 (1997);} for auxiliary-field quantum Monte Carlo \footnote{S. Zhang and H. Krakauer, {\em{Phys. Rev. Lett.}} {\bf{90}}, 136401 (2003)} electronic structure calculations. We discuss the numerical stability and computational complexity of the technique, and assess its performance computing ground-state properties for a broad set of molecules, including constituents of the primordial terrestrial atmoshpere and medium-sized organic molecules. Accurate estimates for electronic density and dipole moment of molecular systems are obtained. [Preview Abstract] |
Monday, March 13, 2017 4:30PM - 4:42PM |
C26.00011: Defect formation energies and equations of state of Mn oxides using diffusion Monte Carlo Vinit Sharma, Jaron Krogel, Paul Kent, Fernando Reboredo Quantum Monte~Carlo (QMC) methods are the most accurate methods available for \textit{ab-initio} calculations of systems with more than 100 electrons. A family of projector methods allows the direct treatment of electron corrections by statistical sampling. Using the diffusion quantum Monte Carlo (DMC) method, as implemented in QMCPACK, we calculate the cohesive energy, formation energy and the structural parameters of MnO and MnO$_{\mathrm{2}}$ and the ternary perovskite LaMnO$_{\mathrm{3}}$. Next, we study the oxygen vacancy and cationic dopants in (La,$A)$MnO$_{\mathrm{3}}$ (A $=$ Ca, Sr, and Ba) which have been identified as suitable candidates for improving the electro-chemical properties of the parent material. The goals the present work are (a) to quantify the accuracy of the DMC method to study the transition metal oxides, (b) to establish the accuracy of different approximations of density functional theory as compared with DMC results, and (c) to demonstrate in another case that QMC methods are an accurate tool for the prediction of the properties of strongly correlated systems. [Preview Abstract] |
Monday, March 13, 2017 4:42PM - 4:54PM |
C26.00012: Fermi-orbital Descriptors across the Periodic Table Kushantha Withanage, Der-You Kao, Koblar Jackson The optimization of the Fermi-orbital descriptor (FOD) positions is required in the method of Fermi-L\”{o}wdin-orbital self-interaction correction (FLO-SIC) in order to minimize the self-interaction-corrected energy.\footnote{ M. R. Pederson, A. Ruzsinszky, and J. P. Perdew, J. Chem. Phys. \textbf{140}, 121103 (2014).} This optimization is carried out using the derivatives of the SIC energy with respect to the FODs. A recent publication\footnote{M. R. Pederson, J. Chem. Phys. \textbf{142}, 064112 (2015).} showed the optimal FODs for a set of closed shell atoms and that the total energy and ionization energies for these atoms were improved using SIC. Knowing how FODs evolve atom to atom, column to column and row to row over the periodic table is expected be very useful. Clear trends for the atoms may suggest strategies for creating useful starting FOD positions for systems involving many atoms. Such transferability is critical for the efficient application of FLO-SIC to large molecules and clusters. In this talk, we will present a simple unbiased method that we used to find the optimal FODs for atoms up to Z = 36 (Kr). We will present our results and discuss the evolution of and patterns in the FOD positions. [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