Symbolic Code Generation for Tensorial PDEs with Applications to Black holes, Cosmic Strings, and Quantum Fields

Thursday, March 9, 2023, 9am US central time

Slides and recordings are available for download pdf (embedded visualizations: Slide 22, Slide 32, Slide 48). The recordings are available in webm and mp4 format,

Abstract

I will present the ongoing developments in code generation techniques for producing executable code blocks representing tensorial PDEs as encountered in numerical relativity. Specifically I will discuss code generation applications to AMReX based simulations of various systems including black hole mergers, cosmic string collapse, and more exoticly, lattice quantum field theories. I will demonstrate specific cases where symbolic tensorial expressions can be written in SymPy or Mathematica and then expanded into code blocks in C++ or Fortran. Simulations of black hole mergers and cosmic string collapse produced entirely from code generated applications will be shown with comparisons to standard testbeds in numerical relativity. Finally, a discussion of current and future developments will be presented.

Presenter: Adam Peterson

Adam Peterson

Adam Peterson is postdoctoral research scientist at the Center for Computational Sciences and Engineering (CCSE) in the Computing Sciences Area at the Lawrence Berkeley National Laboratory. He is a theoretical/computational physicist in high energy, cosmology, and condensed matter.

He is currently working on numerical relativity and interfacing the Einstein Toolkit with AMReX.