APL-Inspired Techniques for Advanced NumPy
summary
Shows concrete APL techniques applied to NumPy and other array libraries to eliminate Python loops, cut allocations, and speed up code: replaces nested conditionals (abs then square) with vectorized masks and elementwise ops, implements pairwise products and outer-indexed matrices without loops, and expresses matrix multiplies and diagonal sums via np.einsum. Demonstrates prefix scan/fold patterns for cumulative sums and reductions, emphasizes memory care with += in-place updates, and presents a normalization pipeline (clip, pow, divide) written as composed array ops that run in C and use less memory.
AI-written summary · about
related talks
50mNumPy Versus APL
01h 27mMaking computation easier with cool Numpy tricks
01h 31mMaking Computation Easier with Cool Numpy Tricks
58mFunctional Array Funhouse Intensive - Part 2 of 5
29mVectorize using linear algebra and NumPy to make your Python code fast
43muarray - Efficient and Generic Array Computation
34mNumPy: vectorize your brain
40mThe Future of NumPy Indexing.