Quantum Sudoku: collapsing, entangling, and sliding through universes

By Baraban, LLM Kun|

Quantum Sudoku: collapsing, entangling, and sliding through universes

This is a thinking toy. I’m not proving physics with Sudoku; I’m using Sudoku to feel what the math is doing: superposition, collapse, entanglement, GHZ/W-style tripartite structure, Schrödinger’s box, decoherence, Many-Worlds, and “the future creating the past.” In the draft below I’ll put demo parameters instead of illustrations. Later we’ll swap them for live snippets, gifs, or short videos.

Part I — The quantum cell

1) Observation and collapse (certainty first)

Start with a particle that’s already collapsed. In Sudoku terms: a cell with exactly one admissible state. Clicking just confirms what was already determined.

2) Superposition of two states

The cell is {1,2} until measured. Same setup, different runs, different outcomes.

State is {1,2} until measured.

3) Two independent particles

Two cells, both in superposition, but not entangled. Each collapses on its own timeline.

Two cells, both {1,2}, but not entangled.

Part II — The rules of the universe (why correlations appear)

Sudoku’s laws are local: every row, column, and section must contain each number exactly once. Once those constraints are on, cells affect each other — that’s how independent dice become correlated qubits.

4) Turning on the laws

Even a 2×2 world shows how local consistency creates nonlocal consequences.

Observe how local rules enforce global order.

5) Hidden order behind the chaos

A standard 9×9 Sudoku has a definite solution. The “superposition” is just ignorance — the answer is fixed before you start, you just haven’t worked it out yet. This is the hidden-variables case.

Hidden determinism: one unique 9×9 solution exists.

Part III — Entanglement and partial observation

6) Two entangled particles (perfect anticorrelation)

Two cells in the same section, constrained to be different. Click one and the other is immediately forced. Measurement on one determines the other.

Two cells share one section and must be different (values {1,2}). Measure one → the other is forced.

7) Entanglement with more than two states

Let each cell have four possibilities. Measuring one cell narrows the other’s candidates without fully resolving it — partial collapse.

Left allowed: Right allowed:
Partial measure ⇒ choose 2 at random; other cell gets the complement. Same section ⇒ final values must differ.

7.1) Multipartite entanglement (three particles, new phenomena)

Two bodies can correlate; three bodies can do qualitatively new things. With 3×1, one section, hints {1,2,3}, we can show three different “tripartite flavors”:

All-Distinct (Sudoku-native constraint). Any click forces the remaining two to be the two other numbers (still undecided who is who until the next click).

GHZ-like (all-match). Only triplets (1,1,1), (2,2,2), (3,3,3) allowed. Measure one → all three lock.

W-like (single excitation). Only permutations of (3,1,1). If you see a 3 anywhere, the others must be 1s; if you see a 1, the lone 3 is still “somewhere else” (residual uncertainty survives a measurement).

7.1a — All-Distinct

Allowed L: Allowed M: Allowed R: Rule: three different numbers from {1,2,3}. First measurement prunes the others; they remain undecided until clicked.

7.1b — GHZ-like

Allowed L: Allowed M: Allowed R: Rule: only (1,1,1), (2,2,2), (3,3,3). Measure any cell ⇒ all three lock to the same value.

7.1c — W-like

Allowed L: Allowed M: Allowed R: Rule: permutations of (3,1,1). Seeing a 3 fixes the other two to 1; seeing a 1 leaves a single unseen 3 “somewhere else.”

Optional hooks

// GHZ-like: after a value appears in any cell, set the others to the same value.
enforceGHZ(board_7b);

// W-like: enforce permutations of (3,1,1); if a cell becomes 3, set the others to 1.
// If a cell becomes 1, restrict others to {1,3} without fully collapsing them.
enforceW(board_7c);

8) Partial observation and decoherence

Each click removes half the candidates from both cells without fixing a value. That’s decoherence: the state becomes more classical — fewer possibilities — without fully collapsing.

Top hints: {1,2,3,4} · Bottom hints: {1,2,3,4} · Weak measurement removes half the possibilities without picking a definite value.

Part IV — Multiverse goggles

9) Many-Worlds vs true randomness vs hidden variables

Two boards, same entangled pair. First click yields 1 on one board, 2 on the other. Three interpretations:

  • Hidden state: deterministic — the answer was fixed, we just didn’t know it.
  • True randomness: the outcome was genuinely undetermined until measured.
  • Many-Worlds: both outcomes happen; you end up in one branch.
Click a cell to measure → reality branches into two.
🌌 Universe A
🌌 Universe B

10) Schrödinger’s Sudoku (box inside a box)

Wrap a small Sudoku inside a “room.” Until we open the room, the inner board stays unobserved (our outer variables treat it as a single superposed object). Open the box → collapse inside becomes visible. Close/erase records → restore uncertainty (quantum eraser flavor).

📦 Box is closed — contents in superposition

Part V — The tree of futures (and the future that writes the past)

11) Branching futures while solving

An unsolved puzzle is a branching tree. Each reveal cuts through possibility space. If a valid solution exists, branches eventually converge. If not, you reach a contradiction — local moves that looked fine turn out to be globally inconsistent.

Demo params

initSudoku("#p5_demo11", {
  subRows: 3, subCols: 3, singleSection: false,
  clickToSetAnswer: true, autoHints: true, autosolver: false
});

12) From future to past: restore superposition from a known finale

Run it backwards. Start from the fully solved state and erase values at random. Certainty decays back into superposition. A fully known board is a dead end — the only way to continue is to introduce new unknowns.

Demo params

initSudoku("#p5_demo12", {
  subRows: 3, subCols: 3, singleSection: false,
  clickToSetAnswer: true, autoHints: true, autosolver: true,
  correctValues: /* fullSolution map injected at runtime */,
});
// UI: Entropy slider => erase N random cells; Step Forward => one-hint fill; Step Back => erase 1

Part VI — Epilogue: sliding between universes; scripts and cycles

13) Sliding between nearby universes

Three identical boards, same seed, different click orders. Small input differences produce different outcomes — that’s the branching structure made tangible.

Demo params

["#p6_demo13a", "#p6_demo13b", "#p6_demo13c"].forEach(sel =>
  initSudoku(sel, { subRows: 2, subCols: 2, singleSection: false, clickToSetAnswer: true })
);

14) Scripted universes and rebirth

If a unique solution exists, the endpoint is fixed. You can traverse it either direction: solving (uncertainty → certainty) or erasing (certainty → uncertainty). After a full solve, the board is done. The only move left is to start over with a new one.

Demo params

// Chain: demo11 (branching) -> demo12 (final→uncertainty) -> back to Part I (single-cell soup)

Notes on the analogy

  • Real quantum theory is math (Hilbert spaces, operators, POVMs, phases). Sudoku gives intuition for the shape of it: constraints → correlations, observation → pruning the possibility space, partial information → decoherence.
  • Hidden-state Sudoku is not quantum indeterminacy — contrasting the two is the point. It maps interpretations.
  • The standard Sudoku used here has a fixed final state — the answer exists before you look, you’re just revealing it. That’s the hidden-variables interpretation. Two richer variants point toward something more quantum:
    • Truly random collapse with constraint preservation — no answer is set upfront; each cell collapses randomly, but the collapse propagates to keep the board consistent (like the Wave Function Collapse algorithm). The final state emerges from the process, not from a pre-existing solution.
    • Probabilistic convergence — there is a target final state, but cells don’t collapse cleanly in one observation. Each measurement pushes the cell toward the answer with some probability, landing in a weak superposition. You need to observe many times before the full state resolves. This is closer to the feel of repeated quantum measurement driving a system toward an eigenstate.
  • Tripartite cases (All-Distinct vs GHZ-like vs W-like) are especially useful. Pairwise intuition breaks; you need to think globally.

What’s next

  • Swap the parameter blocks for live demos driven by sudoku.js.
  • Add tiny “state overlays” (counts of hints per cell, entropy meter).
  • Record short clips for each concept (15–30s) so the rhythm of collapse/decohere/entangle is obvious even without clicking.

If you want, I’ll stitch this into a single page with Next/Prev buttons and load/unload the appropriate initSudoku(...) per section so the article literally plays like a guided tour.

Last modified February 28, 2026: Refined text, removed obvious AI speak (ba57025)
Citations
Baraban & Kun L. (2025). Quantum Sudoku: collapsing, entangling, and sliding through universes.https://KintaroAI.com/blog/2025/10/08/quantum-sudoku-collapsing-entangling-and-sliding-through-universes/ (KintaroAI)
@misc{baraban2025quantumsudokucollapsingentanglingandslidingthroughuniverses,
    author = {Baraban and LLM Kun},
    title = {Quantum Sudoku: collapsing, entangling, and sliding through universes},
    year = {2025},
    url = {https://KintaroAI.com/blog/2025/10/08/quantum-sudoku-collapsing-entangling-and-sliding-through-universes/},
}