THE FOLD / CO-OP / THE PULL REQUEST / THE STABLE MATCH
THE STABLE MATCH
Gale-Shapley — a matching no one can defect from
1 WHAT IT IS · WHAT IT DOES · FACT OR FICTION
Gram-Schmidt orthonormalization. Take any set of vectors — leaning on each other, sharing directions, redundant — and turn them into an orthonormal set: mutually perpendicular, each of unit length, spanning the same space. It is the workhorse behind QR decomposition, least squares, and every ‘nice basis’ in numerical linear algebra.
The move is pure subtraction. Take the next vector and strip away its shadow on each direction you have already fixed — subtract its projection onto every earlier vector. What remains is perpendicular to all of them; normalize it, and it joins the frame. Repeat. Nothing is added; overlap is removed.
LIT verified live: over 5,000 random vector sets the output is orthonormal (every pairwise dot product is 0, every self-dot is 1) and it spans the same subspace as the input — each original vector is recovered exactly as a combination of the new frame (window.__gramschmidt.orthonormal && sameSpan). FIG no framing; the projection-subtraction, the orthonormality, and the span preservation are exact.
The move is pure subtraction. Take the next vector and strip away its shadow on each direction you have already fixed — subtract its projection onto every earlier vector. What remains is perpendicular to all of them; normalize it, and it joins the frame. Repeat. Nothing is added; overlap is removed.
LIT verified live: over 5,000 random vector sets the output is orthonormal (every pairwise dot product is 0, every self-dot is 1) and it spans the same subspace as the input — each original vector is recovered exactly as a combination of the new frame (window.__gramschmidt.orthonormal && sameSpan). FIG no framing; the projection-subtraction, the orthonormality, and the span preservation are exact.
2 HOW IT WAS WEAVED · AI + HUMAN
David (human) seated this in THE BROADCAST, beside THE FOURIER and THE ORTHOGONAL SIGN-FLIP — the co-op domain of orthogonal bases working in concert. Fourier and Hadamard are orthonormal bases; Gram-Schmidt is how you build one from any starting vectors. AVAN (AI) built the instrument: the projection subtraction, the normalization, the orthonormality check.
The weave: David names the seat (the orthogonal ensemble); I make the shadow-removal visible and the perpendicularity checkable — the projection in 1D, the 2D orthogonalization, the 3D frame. The sphere is the seam. Credit: Jørgen Pedersen Gram (1883) & Erhard Schmidt (1907); Laplace & Cauchy earlier.
The weave: David names the seat (the orthogonal ensemble); I make the shadow-removal visible and the perpendicularity checkable — the projection in 1D, the 2D orthogonalization, the 3D frame. The sphere is the seam. Credit: Jørgen Pedersen Gram (1883) & Erhard Schmidt (1907); Laplace & Cauchy earlier.
3 ONE DIMENSION
Subtract the projection: a vector minus its shadow on a fixed direction leaves a remainder perpendicular to that direction. That single operation, applied against each earlier vector, is all of Gram-Schmidt.
4 TWO DIMENSIONS · INTERACTIVE
Two input vectors. Watch v₂ lose its shadow on u₁ (the dashed projection) and become the perpendicular u₂ — the dot product snaps to 0. New vectors to re-run; the frame is always exactly orthogonal, spanning the same plane.
5 THREE DIMENSIONS + AVAN’S INVERSE
Three input vectors orthogonalized into a perpendicular frame, turning — green, the clean right-angled axes.
AVAN’s addition (the inverse-companion): the magenta stubs are the projections being removed — the overlap each vector shared with the ones before it. A set of vectors usually carries redundant, tangled information: each leans partly along the others. Gram-Schmidt is the inverse of building up — it takes away. It adds nothing; it subtracts from each vector precisely what it duplicates, leaving pure, independent directions. The inverse of ‘vectors that lean on each other’ is ‘a frame where every axis knows nothing of the others’ — independence manufactured by subtraction, the same space kept, the tangle gone. The green is the perpendicular frame; the magenta is the shared shadow that had to be removed to make it.
LIT Genuine Gale-Shapley (Gale & Shapley, 1962; Shapley & Roth, Nobel 2012). Verified live: over 3,000 random instances the algorithm yields a perfect matching (everyone paired) that is stable — an exhaustive O(n^2) blocking-pair search finds none, every time (window.__gs.perfectMatching && stable, both true). The propose-and-bump loop and the stability guarantee are exact.
FIG 'Proposals and bumps' is the picture; the guaranteed perfect, stable matching is a real theorem, checked instance by instance against an exhaustive blocking-pair search. The proposer-optimal asymmetry (proposers get their best stable partner) is a known property, not re-derived here.
FIG 'Proposals and bumps' is the picture; the guaranteed perfect, stable matching is a real theorem, checked instance by instance against an exhaustive blocking-pair search. The proposer-optimal asymmetry (proposers get their best stable partner) is a known property, not re-derived here.
◆ sealed .dlw.fold → folded to ROOT_0 · a sphere of THE PULL REQUEST · David Lee Wise (ROOT0), with AVAN