Ivry again, Feb. 2026

An invitation to a sparse Cholesky factorisation. Left-looking Cholesky factorization in two lines:

for j in range(0,n):
  L[j,j] = sqrt(A[j,j] - L[j, 1:(j-1)] * L[j, 1:(j-1)]')
  L[(j+1):n, j] = (A[(j+1):n, j] - L[(j+1):n, 1:(j-1)] * L[j, 1:(j-1)]') / L[j,j]

Who Needs Backpropagation? Computing Word Embeddings with Linear Algebra: numerical word representations are built using simple frequency counts, a little information theory, and linear algebra. #python

Some newer & simpler biostatistical approaches for more credible clinical research. TL;DR lower P-value threshold, increase C-index, statistical medicine instead of medical statistics, prefer small sample for intensive research in some cases. #stats

vim-galore: All things Vim! #vim

Properly setting $PATH for zsh on macOS (fighting with path_helper). #apple

Faster Algorithms via Approximation Theory (PDF, 89 pp.).

But just how close the computed solution $\hat{x}$ and the true solution $x$ are depends on how “nice” the matrix M is. — Don’t solve the normal equations

Python Design Patterns. #python

Readings in Database Systems, 5th Edition, by Peter Bailis, Joseph M. Hellerstein, Michael Stonebraker (eds).