uarray - Efficient and Generic Array Computation
summary
Claims a practical system uarray that rewrites array expressions via the Mathematics of Arrays to eliminate temporaries and unneeded work, then emits code for different backends. Demonstrates the outer product then indexing example where a naive O(n^2) allocation becomes linear work by symbolic reduction, shows generated NumPy-compatible code and discusses targeting other runtimes via XND and backend registration for GPU, TF, and sparse implementations.
AI-written summary · about
chapters (8)
- 0:00uarray introduction
- 3:30Fragmented array ecosystems
- 8:50XND data protocol
- 13:43Universal array vision
- 15:32uarray outer product demo
- 21:57Compiling high-level ops
- 26:54Open questions
- 35:58Future directions
related talks
36muarray: Separating interface from implementation
27mAPL-Inspired Techniques for Advanced NumPy
32mImmutable Arrays Help Mutating Arrays More Efficiently
02h 50mFunctional Array Funhouse Intensive - Part 3 of 5
30mThe evolution of array computing in Python
58mFunctional Array Funhouse Intensive - Part 2 of 5
15mCompilation of Sparse Array Programming Models
28mA Hitchhiker's Guide to the Array API Standard Ecosystem