Bulletin of the American Physical Society
APS March Meeting 2013
Volume 58, Number 1
Monday–Friday, March 18–22, 2013; Baltimore, Maryland
Session A24: Novel Technologies and Algorithms |
Hide Abstracts |
Sponsoring Units: DCOMP Chair: Brandon Cook, Oak Ridge National Laboratory Room: 326 |
Monday, March 18, 2013 8:00AM - 8:12AM |
A24.00001: Cubic-scaling algorithm and self-consistent mean field for the random-phase approximation with second-order screened exchange Jonathan Moussa The random-phase approximation including second-order screened exchange (RPA+SOSEX) is an accurate model of electron correlation energy with two caveats. Its accuracy depends on an arbitrary mean field choice and its scaling of $\mathcal{O}$($n^5$) operations and $\mathcal{O}$($n^3$) memory for $n$ electrons cannot compete with the $\mathcal{O}$($n^3$) operations and $\mathcal{O}$($n^2$) memory scaling of density functional theory (DFT). We rederive RPA+SOSEX as an approximation of the Brueckner doubles coupled-cluster (BCCD) equations, which produces a self-consistent mean field and other model corrections. In addition, we present a new algorithm for RPA+SOSEX that matches the scaling of DFT. We verify the accuracy of the new model on H$_2$ dissociation and the uniform electron gas and verify the reduced scaling of the new algorithm on H$_n$ rings. \\ \\ This work was supported by the Laboratory Directed Research and Development program at Sandia National Laboratories. Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC04-94AL85000. [Preview Abstract] |
Monday, March 18, 2013 8:12AM - 8:24AM |
A24.00002: Large-Scale Hybrid-DFT First-Principles Molecular Dynamics William Dawson, Francois Gygi The recursive subspace bisection algorithm[1] is used to accelerate the computation of the Hartree-Fock exchange operator in hybrid-DFT, First-Principles Molecular Dynamics (FPMD) simulations. This approach provides a set of maximally localized orbitals in domains of variable size and allows for a reduction of the number of computed exchange integrals with controlled accuracy. It does not require a priori assumptions about the localization of orbitals in limited domains and can be used with both occupied and empty orbitals, thus enabling computations of the HOMO-LUMO gap during hybrid-DFT FPMD simulations. We discuss algorithmic improvements of the method and demonstrate its use in hybrid-DFT FPMD simulations of water, solvated ions, and a liquid-solid interface in which maximally localized orbitals show a wide range of localization properties. [1] F.Gygi, Phys. Rev. Lett. 102, 166406 (2009). [2] F. Gygi and I. Duchemin, JCTC (submitted). [3] http://eslab.ucdavis.edu/software/qbox/ . [Preview Abstract] |
Monday, March 18, 2013 8:24AM - 8:36AM |
A24.00003: One-shot calculation of the Electronic Structure across the Metal Insulator Transitions in V$_{2}$O$_{3}$ by Hybrid Density Functional John Robertson, Yuzheng Guo We present the first calculation of the electronic structure of V$_{2}$O$_{3}$ in its different phases using the screened exchange (sX) hybrid functional [1]. The sX functional reproduces the observed band gaps, magnetic moments and photoemission spectra of the corundum paramagnetic metal (PM) phase, the monoclinic anti-ferromagnetic insulating (AFI) phase, and the corundum Cr-doped paramagnetic insulating (PI) phase. The PI phase has a 0.15eV band gap in good agreement with experiment. Using the generalised Kohn-Sham nature of the hybrid functional, a fully relaxed supercell model of the Cr-doped V$_{2}$O$_{3}$ PI phase is calculated, and it shows that the local strain field around Cr atoms is the driving force for the PI-PM transition. This illustrates that hybrid functionals that fix the exchange interaction can give a good, one-shot description of single particle spectra, and are efficient enough compared to DMFT or GW to treat the complex electron-lattice interactions that occur in the more interesting systems.\\[4pt] [1] S J Clark, J Robertson, Phys Rev B 82 085208 (2010) [Preview Abstract] |
Monday, March 18, 2013 8:36AM - 8:48AM |
A24.00004: Ab initio calculations of non-radiative carrier trapping due to deep impurity levels Lin-Wang Wang, Lin Shi Non-radiative carrier decay due to deep impurity levels in semiconductors is an important process which affects the efficiencies of devices from solar cells to light emitting diode. This process is due to multiple phonon emission. Despite of the fact the analytical formalisms have been derived long time ago, so far there is no direct ab initio calculations due to the high cost of calculating all the electron-phonon coupling constants. Here we introduce an algorithm which calculates all the electron-phonon coupling constants at once, hence allows the ab initio calculations of such processes. Another approximation is introduced to calculate the phonon modes of a given impurity system. We use a Zn$_{\mathrm{Ga}}$-V$_{\mathrm{N}}$ paired defect in GaN as an example to study this process. We found that while most of the promoting phonon modes (used to promote the transition with the electron-phonon coupling) come from the optical modes, the accepting phonon modes (used to satisfy the energy conservation) come mostly from the acoustic phonons. [Preview Abstract] |
Monday, March 18, 2013 8:48AM - 9:00AM |
A24.00005: Characterizing Oxidation State using Bader Analysis, Maximally Localized Wannier Functions and Atomic Orbitals Projection Kyle Reeves, Yosuke Kanai The concept of oxidation state of atoms in molecules and materials is widely used to predict and understand chemical and physical properties. This concept is perhaps driven more empirically than by any rigorous criteria differentiating one oxidation state from another. Within the oxidation state framework, an integer number of electrons is assigned to the nuclei within a system. In practice, a distribution of electron density makes it difficult to quantify such discrete assignments without some ambiguities. We explore three different charge analysis approaches in density functional theory calculations for addressing the oxidation state of important organometallic molecules [Ru(bpy)$_{\mathrm{3}}$]$^{\mathrm{2+}}$ and [Ru(bpy)$_{\mathrm{3}}$]$^{\mathrm{3+}}$, which are widely used for solar energy conversion applications. Bader charge analysis, Wannier function analysis, and atomic orbital projection are employed in this work. Given the highly-localized nature of the d-electrons of the ruthenium atom, the charge analysis methods are also compared with Hubbard-U correction. We also discuss how the solvation by water molecules influences the oxidation state characterization for these organometallic complexes. [Preview Abstract] |
Monday, March 18, 2013 9:00AM - 9:12AM |
A24.00006: Revised Basin Hopping Monte Carlo Algorithm Applied for Nanoparticles Juarez L. F. Da Silva, Gustavo G. Rondina The Basin Hopping Monte Carlo (BHMC) algorithm has been very successful in obtaining the atomic structure of nanoparticles (NPs), however, its application for unbiased randomly initialized NPs have been restricted to few hundreds atoms employing empirical pair-potentials (EPP) and for small clusters employing first-principles interacting potentials based on density functional theory (DFT). In this talk, we will present our suggestions for bringing improvements to the the BHMC algorithm, which successfully extend its application for relatively large systems employing EPP and DFT potenticals. Using our implementation from scratch, we have found all the reported putative minimum energy configurations for Lennard-Jones and Sutton-Chen EPPs (N = 2 - 147, 200, 250, 300, ..., 1000). We addressed also binary systems described by the Lennard-Jones or Sutton-Chen empirical potentials, and excellent results have been obtained. Finally, our revised BHMC implementation was combined with DFT potentials (FHI-AIMS), which was employed to study the atomic structure of Al clusters from 2 - 55 atoms in the neutral and charged states. Thus, our results indicate the our suggestions provide an important contribution to improve the quality of the BHMC results employing EPP or DFT potentials. [Preview Abstract] |
Monday, March 18, 2013 9:12AM - 9:24AM |
A24.00007: Generic parallel Wang-Landau sampling for complex systems Ying Wai Li, Thomas Vogel, David P. Landau, Thomas W\"ust We introduce a parallel realization for Wang-Landau sampling in Monte Carlo simulations based on a replica-exchange framework. The key idea is to split the entire energy range of the system under consideration into several smaller, overlapping sub intervals. The survey of configurational phase space can then be distributed over multiple processors, with exchanges of random walkers taking place in the overlapping energy windows. To demonstrate the robustness and advantages of this parallel scheme for the simulations of complex systems, we have applied it to protein adsorption problems using the HP lattice protein model\footnote{K. A. Dill, Biochemistry \textbf{24}, 1501 (1985).}. The method gives significant speed-up and achieves strong scaling on small computer architectures like multi-core processors, with a possible improvement in accuracy. We believe that it could be potentially beneficial for large-scale petaflop machines. [Preview Abstract] |
Monday, March 18, 2013 9:24AM - 9:36AM |
A24.00008: ABSTRACT WITHDRAWN |
Monday, March 18, 2013 9:36AM - 9:48AM |
A24.00009: A Scientific Cloud Computing Platform for Condensed Matter Physics K. Jorissen, W. Johnson, F. D. Vila, J. J. Rehr Scientific Cloud Computing (SCC) makes possible calculations with high performance computational tools, without the need to purchase or maintain sophisticated hardware and software. We have recently developed an interface dubbed SC2IT [1] that controls on-demand virtual Linux clusters within the Amazon EC2 cloud platform [2]. Using this interface we have developed a more advanced, user-friendly SCC Platform configured especially for condensed matter calculations. This platform contains a GUI, based on a new Java version of SC2IT, that permits calculations of various materials properties. The cloud platform includes Virtual Machines preconfigured for parallel calculations and several precompiled and optimized materials science codes for electronic structure and x-ray and electron spectroscopy. Consequently this SCC makes state-of-the-art condensed matter calculations easy to access for general users. Proof-of-principle performance benchmarks [1] show excellent parallelization and communication performance. [1] K. Jorissen, F.D. Vila, and J.J. Rehr, Comp. Phys. Comm. 183 1911 (2012) [2] http://aws.amazon.com and http://www.feffproject.org/scc [Preview Abstract] |
Monday, March 18, 2013 9:48AM - 10:00AM |
A24.00010: Anderson Localization: Dynamical Cluster Approximation - Typical Medium Theory Perspective Chinedu Ekuma, Ziyang Meng, Hanna Terletska, Juana Moreno, Mark Jarrell, Vladimir Dobrosavljevic Mean field theories like the coherent potential approximation (CPA) and its cluster extensions, including the dynamical cluster approximation (DCA), fail to describe the Anderson localization transition in disordered systems. This failure is intrinsic to these theories as the algebraically averaged quantities used in them always favor the metallic state, and hence cannot describe the localization transition. Here we extend the Typical Medium Theory (TMT), which replaces the average quantities with their corresponding typical (geometrically averaged) equivalents, to its cluster form such that non-local correlations can be incorporated systematically. We apply our method to study the localization phenomena in various dimensions. Such an approach opens a new avenue to study localization effect both in model and in real materials. [Preview Abstract] |
Monday, March 18, 2013 10:00AM - 10:12AM |
A24.00011: Gradient-Stable Linear Time Steps for Phase Field Models Benjamin Vollmayr-Lee Phase field models, which are nonlinear partial-differential equations, are a widely used for modeling the dynamics and equilibrium properties of materials. Unfortunately, time marching the equations of motion by explicit methods is usually numerically unstable unless the size of the time step is kept below a lattice-dependent threshold. Consequently, the amount of numerical computation is determined by avoidance of the instability rather than by the natural time scale of the dynamics. This can be a severe overhead. In contrast, a gradient stable method ensures a decreasing free energy, consistent with the relaxational dynamics of the continuous time model. Eyre's theorem proved that gradient stable schemes are possible, and Eyre presented a framework for constructing gradient-stable, semi-implicit time steps for a given phase-field model. Here I present a new theorem that provides a broader class of gradient-stable steps, in particular ones in which the implicit part of the equation is linear. This enables use of fast Fourier transforms to solve for the updated field, providing a considerable advantage in speed and simplicity. Examples will be presented for the Allen-Cahn and Cahn-Hilliard equations, an Ehrlich-Schwoebel-type interface growth model, and block copolymers. [Preview Abstract] |
Monday, March 18, 2013 10:12AM - 10:24AM |
A24.00012: Total energy and force calculations for correlated materials Ivan Leonov, Vladimir I. Anisimov, Dieter Vollhardt We present a computational scheme for the investigation of complex materials with strongly interacting electrons which is able to treat atomic displacements, and hence structural relaxation, caused by electronic correlations. It combines \textit{ab initio} band structure and dynamical mean-field theory and is implemented with the linear response formalism regarding atomic displacements. We employ this approach to compute the equilibrium crystal structure and phase stability of a couple of correlated electron materials, such as elemental hydrogen, SrVO$_3$, and KCuF$_3$. Our results show an overall good agreement between the total energy and force computations of the equilibrium atomic position for these materials. The approach presented here allows one to study the structural properties of materials with strongly correlated electrons such as lattice instabilities observed at correlation induced metal-insulator phase transitions from first principles. [Preview Abstract] |
Monday, March 18, 2013 10:24AM - 10:36AM |
A24.00013: Self-consistent implementation of the vector disordered local moment method for magnetic alloys and its applications to magnetic thermodynamics Kirill Belashchenko, Bhalchandra Pujari, Paul Larson, Vladimir Antropov, Mark van Schilfgaarde We describe an implementation of the coherent potential approximation within the LMTO formalism, which combines chemical and magnetic disorder treated within the vector disordered local moment model. It allows for arbitrary degree of magnetic order for Heisenberg spins specified by axially symmetric spin distribution functions. The atomic charges and potentials are determined self-consistently, and the transverse constraining fields are included as required by density functional theory. Total energies and spectral functions are available, and the spin distribution functions can be used as variational parameters to determine the magnetic state at the given temperature by minimizing the free energy. The performance of this method is illustrated using several examples. The predictions of the Curie temperatures by different approximations for several materials (such as Fe, Co, Gd, FePt, FePd, CoPt) are compared, including the effect of the constraining fields. We also discuss competing magnetic interactions in the Fe$_{1-x}$Mn$_x$Pt alloy, which is known from experiment to present five magnetic phases, including two noncollinear ones. We construct the magnetic phase diagram using the variational minimization of the free energy and obtain the correct sequence of phases. [Preview Abstract] |
Monday, March 18, 2013 10:36AM - 10:48AM |
A24.00014: Large-scale Atomic Effective Pseudopotential Method for the Electronic Structure of Semiconductor Nanostructures Gabriel Bester, R. Cardenas, F. Zirkelbach, P.-Y. Prodhomme, P. Han, R. Cherian In the {\em Large-scale Atomic Effective Pseudopotential Method}, the Schr\"odinger equation of an electronic system is solved within an effective single-particle approach. {\em Atomic Effective Pseudopotentials} are utilized, which are derived from screened local effective crystal potentials obtained from self-consistent density functional theory calculations on elongated and slightly deformed bulk structures. A self-consistency cycle is not required, which reduces the computational effort. Furthermore, iterative solvers can be used to focus only on a few eigenstates of interest, e.g., states in the vicinity of the band gap of a semiconductor. Hence, this approach is particularly well suited for first-principles investigations of the electronic structure of nanostructures consisting of up to ten thousands of atoms, when the knowledge of the total energy of the system is not required. The treatment includes semi-local pseudopotentials (Kleinman Bylander separable form in real space) as well as the spin-orbit interaction. The obtained single-particle wavefunctions are then used to treat excited state properties by means of a configuration interaction approach. We will illustrate the capabilities of the method on some selected semiconductor nanostructures. [Preview Abstract] |
Monday, March 18, 2013 10:48AM - 11:00AM |
A24.00015: Fidelity susceptibility of one-dimensional models with twisted boundary conditions Diptiman Sen, Manisha Thakurathi, Amit Dutta It is well-known that the ground state fidelity of a quantum many-body system can be used to detect its quantum critical points (QCPs). If $g$ denotes the parameter in the Hamiltonian with respect to which the fidelity is computed, we find that for one-dimensional models with a large but finite size, the fidelity susceptibility $\chi_F$ can detect a QCP provided that the correlation length exponent satisfies $\nu < 2$. We then show that $\chi_F$ can be used to locate a QCP even if $\nu \ge 2$ if we introduce boundary conditions labeled by a twist angle $N\theta$, where $N$ is the system size. If the QCP lies at $g=0$, we find that if $N$ is kept constant, $\chi_F$ has a scaling form given by $\chi_F \sim \theta^{-2/\nu} f(g/\theta^{1/\nu})$ if $\theta \ll 2\pi/N$. We illustrate this in a tight-binding model of fermions with a spatially varying chemical potential with amplitude $h$ and period $2q$ in which $\nu = q$. Finally we show that when $q$ is very large, the model has two QCPs at $h=\pm 2$ which cannot be detected by studying the energy spectrum but are clearly detected by $\chi_F$. The peak value and width of $\chi_F$ scale as non-trivial powers of $q$ at these QCPs. We argue that these QCPs mark a transition between extended and localized states at the Fermi energy. [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