1 points | by marcoloco232323 16 hours ago
1 comments
Hi HN! I built dimtensor after wasting too many hours debugging physics simulations where the bug was a unit mismatch buried in the code.
It wraps NumPy/PyTorch/JAX arrays with physical units and catches dimensional errors immediately:
velocity = DimArray([10], units.m / units.s) acceleration = DimArray([9.8], units.m / units.s**2) velocity + acceleration # DimensionError!
Docs: https://marcoloco23.github.io/dimtensor
Happy to answer questions!
Hi HN! I built dimtensor after wasting too many hours debugging physics simulations where the bug was a unit mismatch buried in the code.
It wraps NumPy/PyTorch/JAX arrays with physical units and catches dimensional errors immediately:
Key difference from Pint/Astropy: native PyTorch autograd and JAX JIT support. Gradients flow through unit-aware operations.Docs: https://marcoloco23.github.io/dimtensor
Happy to answer questions!