Hosted by: Kerstin Kleese van Dam

With the advent of exascale computing the field of computational chemistry is on the verge of entering a new era of modeling. Large computing resources can enable researchers to tackle scientific problems that are larger and more realistic than ever before, and to include more of the complex dynamical behavior of nature. However, the future exascale architectures will be significantly different and require advances in algorithms and new programming paradigms. We will discuss some of the work on developing scalable algorithms for strongly correlated systems, simulations of complexes in dynamical environments, and complex spectra. Significant improvements will be reported in our development efforts of a full threaded plane wave ab initio molecular dynamics code in NWChem on Intel Phi platforms. Finally, we will demonstrate advances in the parallel communication layer Global Arrays utlizing LBNL's GasNET and barrier elision techniques. Bio: Bert de Jong leads the Computational Chemistry, Materials, and Climate Group at LBNL. He has a background in general chemistry, chemical engineering and high performance computational chemistry, with specialization and strong capabilities in modeling heavy element chemistry. He is a main developer of the NWChem software at the EMSL, one of four developers of the unique fully relativistic software MOLFDIR for quantum chemistry. Prior to joining Berkeley Lab, de Jong was at PNNL, where he lead the High Performance Software Development Group responsible for NWChem. He has published 89 journal papers, 14 conference papers and 7 book chapters and has given over 65 invited presentations and lectures at international conferences and universities.De Jong earned his doctorate in theoretical chemistry in 1998 from the University of Groningen in the Netherlands. He was a postdoctoral fellow at PNNL before transitioning to a staff member in 2000.