Case Study: Beyond Homogeneous Decomposition with Qbox Scaling Long-Range Forces on Massively Parallel Systems
Donald Frederick, LLNL – Presented at Supercomputing ‘11 Lawrence Livermore National Laboratory, P. O. Box 808, Livermore, CA 94551! LLNL-PRES-508651 Case Study: Outline
• Problem Descrip on • Computa onal Approach • Changes for Scaling
LLNL-PRES-508651 Computer simula ons of materials Computer simula ons are widely used to predict the proper es of new materials or understand the proper es of exis ng ones
LLNL-PRES-508651 Simula on of Materials from First-
First-principles methods: Principles Calculate proper es of a given material directly from fundamental physics equa ons.
• No empirical parameters Can make predic ons about complex or novel materials, under condi ons where experimental data is unavailable or inconclusive.
• Chemically dynamic As atoms move, chemical bonds can be formed or broken. Electron density surrounding • Computa onally expensive water molecules, calculated Solving quantum equa ons is me-consuming, limi ng from first-principles systems sizes (e.g. hundreds of atoms) and simula on mescales (e.g. picoseconds)
LLNL-PRES-508651 Quantum Mechanics is Hard Proper es of a many-body quantum system are given by Schrödinger's Equa on:
2 ⎡⎤−h 2 vv vv v ⎢⎥∇+V()V()V()ionrr + H + xc rrψεψiii () = () r ⎣⎦2m Exact numerical solu on has exponen al complexity in the number of quantum degrees of freedom, making a brute force approach imprac cal (e.g. a system with 10,000 electrons would take 2.69 x 1043 mes longer to solve than one with only 100 electrons!)
Approximate solu ons are needed: Quantum Monte Carlo stochas cally sample high-dimensional phase space O(N3) Density Func onal Theory use approximate form for exchange and correla on of electrons
Both methods are O(N3), which is expensive, but tractable
LLNL-PRES-508651 Density Func onal Theory Density func onal theory: total energy of system can be wri en as a func onal of the electron density
occ vv*2⎛⎞11 v vvion v ZZ ER{},{}ψψψρ= 2 drr ()−∇ () rdrrVr+ () () +IJ [ iI] ∑∑∫∫ i⎜⎟ i iJI⎝⎠22≠ RRIJ− v xc 1()vv vρ rʹ ++Edrrdr[]ρρ () ʹ vv 2 ∫∫rr− ʹ many-body effects are approximated by a density func onal exchange-correla on
Solve Kohn-Sham equa ons self-consistently to find the ground state electron density defined by the minimum of this energy func onal (for a given set of ion posi ons)
The total energy allows one to directly compare different structures and configura ons
LLNL-PRES-508651 Sta c Calcula ons: Total Energy Basic idea: For a given configura on of atoms, we find the electronic density which gives the lowest total energy
Bonding is determined by electronic structure, no bonding informa on is supplied a priori
LLNL-PRES-508651 First-Principles Molecular Dynamics (FPMD) For dynamical proper es, both ions and electrons have to move simultaneously and consistently: 1. Solve Schrodinger’s Equa on to find ground state electron density. 2. Compute inter-atomic forces. 3. Move atoms incrementally forward in me. 4. Repeat. Because step 1 represents the vast majority (>99%) of the computa onal effort, we want to choose me steps that are as large as possible, but small enough so that previous solu on is close to current one