Linear Algebra in PyTorch
summary
PyTorch 1.2–1.3 adds native batching so batched QR, LU and other decompositions run as single ops instead of Python loops, and most linear-algebra ops are differentiable under autograd. New, clearer function names replace LAPACK abbreviations: LU, solve, lstsq, triangular_solve (with a get_info/get_enforce-style return for factorization metadata). The changes target practical workflows: fewer Python hacks, faster batched kernels, and backward-compatible APIs for common matrix factorizations and solves.
AI-written summary · about
chapters (5)
related talks
07mComplex Numbers in PyTorch
06mMaking PyTorch More "NumPy Compatible"
43mThe Linear Algebra of Deep Learning
32mAdvanced Techniques for Low-Rank Matrix Approximation
32mA Primer (or Refresher) On Linear Algebra for Data Science
34mThe Linear Algebra of Deep Learning
01h 31mPast, Present and Future of Randomized Numerical Linear Algebra II
12mLightning Talk: Simulating Quantum Systems with PyTorch