NIKX-Tech/numx

linalg: QR decomposition (Householder)

Open

#82 opened on Jul 11, 2026

View on GitHub
 (0 comments) (0 reactions) (0 assignees)C (1 fork)github user discovery
enhancementgood first issue

Repository metrics

Stars
 (4 stars)
PR merge metrics
 (PR metrics pending)

Description

Add numx_qr_decompose to the linalg module: QR decomposition via Householder reflections, bounded by NUMX_MAX_MAT_ROWS/NUMX_MAX_MAT_COLS like the rest of linalg (see numx_lu_decompose and numx_cholesky_decompose, #51, for the established style of matrix decomposition functions in this codebase).

Follow CONTRIBUTING.md exactly: full Doxygen header, NULL-checks, numx_status_t return, all four deliverables (header, impl, test, docs page). Test against known QR examples and verify Q is orthogonal (Q^T Q = I) and QR reconstructs the original matrix, following the same reconstruction-test pattern #51 used for Cholesky (test_cholesky_decompose_residual_reconstruction in tests/test_linalg.c).

Contributor guide