Hi everyone,
I’m a computational science student investigating the effects that arise from lowering the floating-point precision of operations in scientific codes. For this research project, I have a system in place to take arbitrary parts of a code and automatically convert the precision of all floating-point operations at compile-time.
What I need help with is finding some process in OpenMC that can be expressed with an error threshold. Can you point me to a kernel in OpenMC that is bound to an error or residual term? Is it possible to express the overall error or residual of an OpenMC simulation in a single variable?
Thanks!