Recursive Factorization of the Inverse Overlap Matrix in Linear-Scaling Quantum Molecular Dynamics Simulations.

We present a reduced complexity algorithm to compute the inverse overlap factors required to solve the generalized eigenvalue problem in a quantum-based molecular dynamics (MD) simulation. Our method is based on the recursive, iterative refinement of an initial guess of Z (inverse square root of the overlap matrix S). The initial guess of Z is obtained beforehand by using either an approximate divide-and-conquer technique or dynamical methods, propagated within an extended Lagrangian dynamics from previous MD time steps. With this formulation, we achieve long-term stability and energy conservation even under the incomplete, approximate, iterative refinement of Z. Linear-scaling performance is obtained using numerically thresholded sparse matrix algebra based on the ELLPACK-R sparse matrix data format, which also enables efficient shared-memory parallelization. As we show in this article using self-consistent density-functional-based tight-binding MD, our approach is faster than conventional methods based on the diagonalization of overlap matrix S for systems as small as a few hundred atoms, substantially accelerating quantum-based simulations even for molecular structures of intermediate size. For a 4158-atom water-solvated polyalanine system, we find an average speedup factor of 122 for the computation of Z in each MD step.

[1]  Michele Benzi,et al.  A Sparse Approximate Inverse Preconditioner for the Conjugate Gradient Method , 1996, SIAM J. Sci. Comput..

[2]  Anders M N Niklasson,et al.  Generalized extended Lagrangian Born-Oppenheimer molecular dynamics. , 2014, The Journal of chemical physics.

[3]  Emanuel H. Rubensson,et al.  Graph-based linear scaling electronic structure theory. , 2016, The Journal of chemical physics.

[4]  Joost VandeVondele,et al.  Linear Scaling Self-Consistent Field Calculations with Millions of Atoms in the Condensed Phase. , 2012, Journal of chemical theory and computation.

[5]  J. C. Slater,et al.  Simplified LCAO Method for the Periodic Potential Problem , 1954 .

[6]  Emanuel H. Rubensson,et al.  Recursive inverse factorization. , 2008, The Journal of chemical physics.

[7]  Sándor Suhai,et al.  A Self‐Consistent Charge Density‐Functional Based Tight‐Binding Method for Predictive Materials Simulations in Physics, Chemistry and Biology , 2000 .

[8]  A M N Niklasson,et al.  Efficient parallel linear scaling construction of the density matrix for Born-Oppenheimer molecular dynamics. , 2015, Journal of chemical theory and computation.

[9]  Stephen L. Olivier,et al.  Task Parallel Incomplete Cholesky Factorization using 2D Partitioned-Block Layout , 2016, ArXiv.

[10]  Ester M. Garzón,et al.  Improving the Performance of the Sparse Matrix Vector Product with GPUs , 2010, 2010 10th IEEE International Conference on Computer and Information Technology.

[11]  Emanuel H. Rubensson,et al.  Interior Eigenvalues from Density Matrix Expansions in Quantum Mechanical Molecular Dynamics , 2014, SIAM J. Sci. Comput..

[12]  Matt Challacombe,et al.  Time-reversible Born-Oppenheimer molecular dynamics. , 2006, Physical review letters.

[13]  Matt Challacombe,et al.  A simplified density matrix minimization for linear scaling self-consistent field theory , 1999 .

[14]  C. Negre,et al.  Atomistic structure dependence of the collective excitation in metal nanoparticles. , 2008, The Journal of chemical physics.

[15]  Car,et al.  Orbital formulation for electronic-structure calculations with linear system-size scaling. , 1993, Physical review. B, Condensed matter.

[16]  M. Challacombe.  A general parallel sparse-blocked matrix multiply for linear scaling SCF theory , 2000 .

[17]  Michiaki Arita,et al.  Stable and Efficient Linear Scaling First-Principles Molecular Dynamics for 10000+ Atoms. , 2014, Journal of chemical theory and computation.

[18]  Gustavo E. Scuseria,et al.  Linear scaling conjugate gradient density matrix search as an alternative to diagonalization for first principles electronic structure calculations , 1997 .

[19]  M. Oviedo,et al.  Dynamical simulation of the optical response of photosynthetic pigments. , 2010, Physical chemistry chemical physics : PCCP.

[20]  Emanuel H. Rubensson,et al.  Systematic sparse matrix error control for linear scaling electronic structure calculations , 2005, J. Comput. Chem..

[21]  D. Bowler,et al.  O(N) methods in electronic structure calculations. , 2011, Reports on progress in physics. Physical Society.

[22]  Anders M N Niklasson,et al.  Extended Born-Oppenheimer molecular dynamics. , 2008, Physical review letters.

[23]  Sándor Suhai,et al.  Self-consistent-charge density-functional tight-binding method for simulations of complex materials properties , 1998 .

[24]  David R. Bowler,et al.  Stable and Efficient Linear Scaling First-Principles Molecular Dynamics for 10,000+ atoms , 2014 .

[25]  Emanuel H. Rubensson,et al.  Truncation of small matrix elements based on the Euclidean norm for blocked data structures , 2009, J. Comput. Chem..

[26]  J. Olsen,et al.  Linear-scaling symmetric square-root decomposition of the overlap matrix. , 2007, The Journal of chemical physics.

[27]  P. Löwdin.  Quantum theory of cohesive properties of solids , 2001 .

[28]  Bálint Aradi,et al.  Extended Lagrangian Density Functional Tight-Binding Molecular Dynamics for Molecules and Solids. , 2015, Journal of chemical theory and computation.

[29]  M. Oviedo,et al.  A theoretical study of the optical properties of nanostructured TiO2 , 2013, Journal of physics. Condensed matter : an Institute of Physics journal.

[30]  Peter Pulay,et al.  Fock matrix dynamics , 2004 .

[31]  Nicolas Bock,et al.  Extended Lagrangian Born-Oppenheimer molecular dynamics with dissipation. , 2009, The Journal of chemical physics.

[32]  Thomas D. Kühne,et al.  Linear-scaling self-consistent field theory based molecular dynamics: application to C60buckyballs colliding with graphite , 2018, Molecular Simulation.

[33]  D. Remler,et al.  Molecular dynamics without effective potentials via the Car-Parrinello approach , 1990 .

[34]  A. Niklasson.  Iterative refinement method for the approximate factorization of a matrix inverse , 2004 .

[35]  Stewart A. Adcock,et al.  Molecular Dynamics: Survey of Methods for Simulating the Activity of Proteins , 2006 .

[36]  Taisuke Ozaki,et al.  Efficient recursion method for inverting an overlap matrix , 2001 .

[37]  Emanuel H. Rubensson,et al.  Density matrix purification with rigorous error control. , 2008, The Journal of chemical physics.

[38]  J. M. Dickey,et al.  Computer Simulation of the Lattice Dynamics of Solids , 1969 .

[39]  Jürg Hutter,et al.  Car–Parrinello molecular dynamics , 2012 .

[40]  Martin Karplus,et al.  Lagrangian formulation with dissipation of Born-Oppenheimer molecular dynamics using the density-functional tight-binding method. , 2011, The Journal of chemical physics.

[41]  Anders M. N. Niklasson,et al.  Wave function extended Lagrangian Born-Oppenheimer molecular dynamics , 2010 .

[42]  Michael Methfessel,et al.  Crystal structures of zirconia from first principles and self-consistent tight binding , 1998 .

[43]  Martin Head-Gordon,et al.  Fast Sparse Cholesky Decomposition and Inversion using Nested Dissection Matrix Reordering. , 2011, Journal of chemical theory and computation.

[44]  Matthias Krack,et al.  Efficient and accurate Car-Parrinello-like approach to Born-Oppenheimer molecular dynamics. , 2007, Physical review letters.

[45]  Sihong Shao,et al.  Analysis of Time Reversible Born-Oppenheimer Molecular Dynamics , 2013, Entropy.

[46]  Emanuel H. Rubensson,et al.  A hierarchic sparse matrix data structure for large‐scale Hartree‐Fock/Kohn‐Sham calculations , 2007, J. Comput. Chem..

[47]  Peter L. Freddolino,et al.  Force field bias in protein folding simulations. , 2009, Biophysical journal.

[48]  Anders M.N. Niklasson.  Expansion algorithm for the density matrix , 2002 .

[49]  S. Goedecker.  DECAY PROPERTIES OF THE FINITE-TEMPERATURE DENSITY MATRIX IN METALS , 1998 .