Modeling mesoscopic superconductivity with high-performance computing
Title: Modeling mesoscopic superconductivity with high-performance computing
SNIC Project: SNIC 2021/22-155
Project Type: SNIC Small Compute
Principal Investigator: Pascal Stadler <pascal.stadler@chalmers.se>
Affiliation: Chalmers tekniska högskola
Duration: 2021-02-26 – 2021-05-01
Classification: 10304
Keywords:

Abstract

One of the most powerful tools to study mesoscopic superconducting devices is the quasiclassical (QC) theory of superconductivity. However, there is a gap in the QC literature regarding an explicit, and thorough, explanation of how to properly implement and solve the relevant equations numerically. This knowledge is, at best, fragmented over different papers published over several decades. Since no open-source libraries exist, researchers and students spend valuable time re-implementing the same algorithms, often in a sub-optimal manner. Furthermore, the inherently selfconsistent equations are difficult to implement numerically, and computationally intensive in the mesoscopic regime. To remedy all these issues, we exploit the fact that the problem is parallelizable, and present a High-Performance Computing framework with highly optimized numerical implementations, running on state-of-the-art hardware in the form of Graphics Processing Units. The goal of the project is to run our software on HPC and study different finite size systems with various geometries, order parameters and magnetic fields.