wai-quantum 0.6.9: cut-rank amplitudes with the circuit's own global phase #350

Merged

dcharlot merged 1 commit from wai/quantum-exact-phase into main 2026-10-20 04:49:01 -04:00

Conversation 0 Commits 1 Files changed 18 +561 -51

dcharlot commented 2026-10-20 03:33:40 -04:00

Owner

Copy link

wai-quantum 0.6.9. The service's cutrank amplitudes now carry the circuit's own global phase, so they equal the state vector's exactly rather than up to a shared phase.

What changes

• PhasedTableau (quantum_tableau) is a stabilizer tableau with its global phase.
  • It tracks one basis state r of the support and its exact amplitude ⟨r|ψ⟩ = 2^{-k/2} e^{iπm/4}.
  • Paulis, phase gates, CX, CY, CZ and SWAP move r and the phase directly.
  • For a Hadamard on qubit a, the amplitude of r ⊕ e_a is c · ⟨r|ψ⟩, with c read from the stabilizer whose X part is exactly e_a. That stabilizer is found by reducing the stabilizers, about n³/64 word operations; c = 0 when there is none.
  • √X = H S H exactly.
• ProductMeasured::with_last (same graph form, order and steps; other closing gates) and ProductMeasured::scale (every amplitude times a factor, folded into one qubit's weights).
• The service scales cutrank amplitudes by e^{iΘ} a/b:
  • e^{iΘ} is the phase the lowering to Clifford gates drops, read gate by gate from each gate's matrix against its lowering's;
  • a is the tracked amplitude;
  • b is the graph form's amplitude at the same basis state, with identity closing gates.
  • Past about a minute of tracking (counted from the Hadamards), the result says global_phase is not fixed, as before.
  • auto now tries cutrank for amplitudes, before tensor_network.

Evidence

• The reference amplitude equals the dense state vector's after every gate of 40 random Clifford circuits (2 to 8 qubits, over 2,000 checks, phase included).
• with_last and scale agree with fresh builds.
• Service amplitudes equal the state vector's directly, with no phase aligned, on random Clifford-then-layer circuits of 6, 9 and 11 qubits. They also do on a circuit whose Clifford part uses rz, rx, ry, u, p, cp, rzz, rxx, iswap, sx, sxdg, cy, swap, crz and cu at Clifford angles.
• Mutants caught: a wrong phase in the Hadamard update, the lowering phase dropped, and the tracked amplitude ignored.
• Checked on the build host:
  • --lib quantum_tableau (11), quantum_cutrank (8), --test bit_identity (23);
  • the service's tests (64), clippy, and the full smoke.sh against a local server;
  • check-lockfiles.
• The public-text checks are clean.

🤖 Generated with Claude Code

**wai-quantum 0.6.9.** The service's `cutrank` amplitudes now carry the circuit's own global phase, so they equal the state vector's exactly rather than up to a shared phase. ### What changes - **`PhasedTableau`** (`quantum_tableau`) is a stabilizer tableau with its global phase. - It tracks one basis state `r` of the support and its exact amplitude `⟨r|ψ⟩ = 2^{-k/2} e^{iπm/4}`. - Paulis, phase gates, CX, CY, CZ and SWAP move `r` and the phase directly. - For a Hadamard on qubit `a`, the amplitude of `r ⊕ e_a` is `c · ⟨r|ψ⟩`, with `c` read from the stabilizer whose X part is exactly `e_a`. That stabilizer is found by reducing the stabilizers, about `n³/64` word operations; `c = 0` when there is none. - `√X = H S H` exactly. - **`ProductMeasured::with_last`** (same graph form, order and steps; other closing gates) and **`ProductMeasured::scale`** (every amplitude times a factor, folded into one qubit's weights). - **The service** scales `cutrank` amplitudes by `e^{iΘ} a/b`: - `e^{iΘ}` is the phase the lowering to Clifford gates drops, read gate by gate from each gate's matrix against its lowering's; - `a` is the tracked amplitude; - `b` is the graph form's amplitude at the same basis state, with identity closing gates. - Past about a minute of tracking (counted from the Hadamards), the result says `global_phase` is not fixed, as before. - `auto` now tries `cutrank` for amplitudes, before `tensor_network`. ### Evidence - **The reference amplitude** equals the dense state vector's after every gate of 40 random Clifford circuits (2 to 8 qubits, over 2,000 checks, phase included). - **`with_last` and `scale`** agree with fresh builds. - **Service amplitudes equal the state vector's directly**, with no phase aligned, on random Clifford-then-layer circuits of 6, 9 and 11 qubits. They also do on a circuit whose Clifford part uses `rz`, `rx`, `ry`, `u`, `p`, `cp`, `rzz`, `rxx`, `iswap`, `sx`, `sxdg`, `cy`, `swap`, `crz` and `cu` at Clifford angles. - **Mutants caught:** a wrong phase in the Hadamard update, the lowering phase dropped, and the tracked amplitude ignored. - **Checked on the build host:** - `--lib quantum_tableau` (11), `quantum_cutrank` (8), `--test bit_identity` (23); - the service's tests (64), clippy, and the full `smoke.sh` against a local server; - `check-lockfiles`. - The public-text checks are clean. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

dcharlot added 1 commit 2026-10-20 03:33:40 -04:00

wai-quantum 0.6.9: cut-rank amplitudes with the circuit's own global phase ... cfa797f856

PhasedTableau tracks one basis state of a stabilizer state's support and its
exact amplitude (2^{-k/2} e^{iπm/4}) through every gate; a Hadamard reads the
stabilizer whose X part is the one qubit. ProductMeasured gains with_last and
scale. The service's cutrank method scales its amplitudes by e^{iΘ} a/b (the
lowering's dropped phase, the tracked amplitude, the graph form's at the same
basis state), so they carry the circuit's own global phase; auto tries cutrank
for amplitudes. Tests against the dense state vector, phase included; three
mutants caught. Version 0.6.9 in every lockfile that pins it.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

dcharlot merged commit f85c7cdad5 into main 2026-10-20 04:49:01 -04:00

dcharlot referenced this pull request from a commit 2026-10-20 04:49:02 -04:00

Merge pull request 'wai-quantum 0.6.9: cut-rank amplitudes with the circuit's own global phase' (#350) from wai/quantum-exact-phase into main

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels

No items

No labels

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

Transaction-Science/open-standards!350

No description provided.