You're in academia. The goal isn't to write fast code it is to create (and publish!) knowledge. If it is easier for you to create knowledge in Python than in C

You're in academia. The goal isn't to write fast code it is to create (and publish!) knowledge. If it is easier for you to create knowledge in Python than in C you should use that. In HPC exact performance on a specific device with a specific runtime is uninteresting. What is interesting is how well the solution scales, what its bottle-necks and constraints are.