465599 Efficient K-Point Grid Generation through the Use of Informatics

Thursday, November 17, 2016: 2:18 PM
Yosemite A (Hilton San Francisco Union Square)
Pandu Wisesa and Tim Mueller, Materials Science and Engineering, Johns Hopkins University, Baltimore, MD

Many methods for calculating the physical properties of crystalline materials depend on the evaluation of an integral over the Brillouin zone which is generally approximated by sampling points in reciprocal space known as k-points. By choosing an efficient grid of k-points one can significantly reduce the cost of a calculation. However, simpler grids which may lead to higher computational cost tend to be favored due to the time and effort required to find the most efficient k-point grids. To resolve this problem, we have created a publicly available k-point grid server that is capable of generating highly efficient k-point grids in a fraction of a second. This server is backed by a database of approximately 60,000 highly efficient k-point grids. We present an overview of our method and benchmark results, and we discuss how our tool can be integrated in a computing environment.

Extended Abstract: File Uploaded