CLOSE
PRINT
Aiichiro Nakano
Associate Professor
University of Southern California
Ultrascale Atomistic Simulations of Nanomechanical Properties
Classical and quantum mechanical simulations involving multimillion- to-multibillion atoms provide powerful tools to study nanomechanical properties of materials. We have developed enabling computational techniques for such ultrascale atomistic simulations.
Based on an embedded divide-and-conquer (EDC) framework, we have designed linear-scaling algorithms for a hierarchy of molecular dynamics (MD) simulations: 1) space-time multiresolution MD (MRMD); 2) first principles-based fast reactive force-field (F-ReaxFF) MD; 3) an EDC density functional theory (DFT) algorithm on adaptive multigrids for quantum mechanical MD; and 4) a hierarchical MD/QM simulation approach that invokes accurate simulations only when and where high fidelity is required. To map these O(N) algorithms onto massive parallel computers with deep memory hierarchies, we have developed a tunable hierarchical cellular decomposition (THCD) framework, which achieves performance tunability through a hierarchy of parameterized cell data/computation structures and adaptive load balancing through wavelet-based computational-space decomposition. We have also developed: 1) octree-based probabilistic visibility culling for interactive visualization of billion-atom datasets; and 2) graph- based mining to detect topological anomalies in chemical bond networks. Scalability tests on 1920 Itanium2 processors have achieved unprecedented scales of reactive atomistic simulations--0.56 billion- atom F-ReaxFF and 1.4 million-atom (0.12 trillion electronic degrees- of-freedom) EDC-DFT simulations--in addition to 18.9 billion-atom MRMD simulation, with parallel efficiency as high as 0.953.
I will also discuss the application of large atomistic simulations to the study of hypervelocity impact damage, nano-energetic materials, and nanoindentation of superhard nanophase ceramics.
BIO
Aiichiro Nakano is an Associate Professor at the University of Southern California. He has published many articles on visualization, distributed computing for molecular dynamics and quantum mechanics, nanosystems, multiscale methods, rigid body dynamics, and billion atom simulations for materials research. He is a recipient of a National Science Foundation Career Award and received the Best Technical Paper Award at the IEEE/ACM Supercomputing 2001 Conference. He received his PhD at the University of Tokyo, Japan, in 1989.