THE FOLD / GRIND / THE MAINFRAME / THE SIMPLEX
THE SIMPLEX
the optimum lives on the boundary
1 WHAT IT IS · WHAT IT DOES · FACT OR FICTION
The simplex method solves linear programs — maximize a linear objective c·x subject to linear inequalities Ax ≤ b, x ≥ 0. Its key insight: the feasible region is a convex polytope, and the optimum is always at a vertex (a corner), never strictly inside. So simplex starts at one corner and walks along edges, each step to a neighbouring vertex that improves the objective, until no neighbour is better — that corner is optimal. It never wanders the interior; it hops corner to corner.
LIT verified live: over thousands of random linear programs, the simplex optimum equals the best value found by brute-force enumeration of every feasible vertex of the polytope (window.__simplex). FIG no framing; the tableau pivots and the exhaustive vertex search both run in-browser.
LIT verified live: over thousands of random linear programs, the simplex optimum equals the best value found by brute-force enumeration of every feasible vertex of the polytope (window.__simplex). FIG no framing; the tableau pivots and the exhaustive vertex search both run in-browser.
2 HOW IT WAS WEAVED · AI + HUMAN
David (human) seated this at the-mainframe — linear programming was the mainframe’s first killer app, planning and logistics at industrial scale, corner by corner. AVAN (AI) built the instrument: a simplex tableau with Bland’s anti-cycling rule and a brute-force vertex oracle to confirm the optimum.
Credit as content: George Dantzig (1947). The weave: David names the mainframe; I confirm simplex lands on the same optimum that exhaustively checking every vertex would — the answer is always at a corner.
Credit as content: George Dantzig (1947). The weave: David names the mainframe; I confirm simplex lands on the same optimum that exhaustively checking every vertex would — the answer is always at a corner.
3 ONE DIMENSION
A 2D feasible polygon cut out by the constraints; the objective pushes in one direction, and the optimum sits at the far corner.
4 TWO DIMENSIONS · INTERACTIVE
Generate a random linear program; the feasible region, the objective direction, and the optimal vertex are drawn — matched against brute-force vertex enumeration.
5 THREE DIMENSIONS + AVAN’S INVERSE
The green forward object: the optimal vertex the edge-walk arrives at.
AVAN’s addition (the inverse-companion): don’t search the interior — only the corners can win. The inverse of ‘optimise over the whole region’ is ‘a linear objective is maximised at a vertex, so the infinite interior is irrelevant.’ Magenta is the interior that never holds the optimum; green is the winning corner. The optimum lives on the boundary.
LIT Genuine simplex method (George Dantzig, 1947), with Bland's anti-cycling rule. Verified live: over 3000 random LPs, the simplex tableau optimum equals brute-force enumeration over every feasible vertex of the polytope (window.__simplex.matchesBrute).
FIG No framing: the tableau pivots and the exhaustive vertex search both run in-browser and agree. The AVAN inverse is honest — that a linear objective is always maximised at a vertex (so the infinite interior is irrelevant) is the fundamental theorem simplex exploits; magenta is the interior that never holds the optimum, green the winning corner. The optimum lives on the boundary.
FIG No framing: the tableau pivots and the exhaustive vertex search both run in-browser and agree. The AVAN inverse is honest — that a linear objective is always maximised at a vertex (so the infinite interior is irrelevant) is the fundamental theorem simplex exploits; magenta is the interior that never holds the optimum, green the winning corner. The optimum lives on the boundary.
◆ sealed .dlw.fold → folded to ROOT_0 · a sphere of THE MAINFRAME · David Lee Wise (ROOT0), with AVAN