banctilrobitaille/torch-vectorized
Fast analytical implementation of batch eigen-decomposition for 3x3 symmetric matrices with Pytorch. > 250x faster than regular Pytorch implementation of batch eigen-decomposition on GPU.
PythonMIT
Issues
- 0
Possible to extend to 4x4 matrices?
#9 opened by Jessie-jx - 1
Getting negative eigenvalues for PDS matrices
#8 opened by zhesu1 - 1
- 4
back-propagation issue
#5 opened by zhan-xu - 0
Not supporting 0s only 3x3 matrix
#2 opened by banctilrobitaille