Use np.add.reduce(...) instead of np.sum(...) for small performance gains #2972
Description
OpenPNM/openpnm/core/_mixins.py
Line 205 in 7fd91d9
Hi, thank you for the clean and well-organized code!
In this line:
num_hits = np.sum(arr, axis=0)If this appears in a performance-critical section or inside a loop, consider replacing it with:
num_hits = np.add.reduce(arr, axis=0)np.sum(...) internally uses np.add.reduce(...), but includes additional logic (e.g., for dtype, keepdims, type promotion). If you don’t rely on those extras, using np.add.reduce(...) directly avoids that overhead and offers a small performance improvement, especially for large arrays or many iterations.
This is a micro-optimization—feel free to keep the current version if readability is preferred. Happy to open a PR if you’d like!