Here's a presentation from last year where a Python data scientist compares a Python and Nim implementation for a problem, with the Python version calling out to Numpy. There are performance comparisons at the end and his Nim version was faster so Nim should be usable for scientific programming:
The big issue Nim faces isn't performance but rather the relative community sizes, and thus how many libraries are available (and also how much help you might find when you run into problems).
https://archive.fosdem.org/2022/schedule/event/nim_hpcfrompy...
The big issue Nim faces isn't performance but rather the relative community sizes, and thus how many libraries are available (and also how much help you might find when you run into problems).