Implement Shor's 9 Qubit Code

[adnathanail]

Description

Adds Shor's 9 Qubit Code as a benchmark

Fixes #790

Checklist:

• The pull request only contains commits that are focused and relevant to this change.
• I have added appropriate tests that cover the new/changed functionality.
• I have updated the documentation to reflect these changes.
• I have added entries to the changelog for any noteworthy additions, changes, fixes, or removals.
• I have added migration instructions to the upgrade guide (if needed).
• The changes follow the project's style guidelines and introduce no new warnings.
• The changes are fully tested and pass the CI checks.
• I have reviewed my own code changes.

[coderabbitai]

📝 Walkthrough

Summary by CodeRabbit

• New Features

  • Added Shor's 9 Qubit Code benchmark, supporting multiple logical qubits with flexible circuit sizes.

• Tests

  • Enhanced test coverage with structure validation and multi-qubit circuit testing for the new benchmark.

• Improvements

  • Optimized transpilation handling for Clifford+T gate synthesis.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

Walkthrough

Adds a new Shor's 9 Qubit Code benchmark and tests, introduces a CustomSolovayKitaev transpiler pass, and wires the pass into benchmark native-gate synthesis for Clifford+T targets. Changelog updated with the new benchmark entry and PR link.

Changes

Cohort / File(s)	Summary
Changelog CHANGELOG.md	Added Unreleased "Added" entry for "Shor's 9 Qubit Code benchmark" and PR link [#803].
Shor benchmark src/mqt/bench/benchmarks/shors_nine_qubit_code.py	New benchmark module: helpers for encoding/decoding, syndrome extraction, correction; builds single-logical-qubit circuit (17 qubits), validates num_qubits (≥17, divisible by 17), supports multiple logical qubits, adds barrier and measure_all, registers create_circuit(num_qubits).
Tests tests/test_bench.py	Added SPECIAL_QUBIT_COUNTS["shors_nine_qubit_code"] = 17. New tests test_shors_nine_qubit_code_circuit_structure and test_shors_nine_qubit_code_multiple_logical_qubits. Extended test_wrong_circuit_size with Shor case and added runtime skip in test_quantumcircuit_levels for devices with insufficient qubits. Imported IfElseOp for assertions.
Custom Solovay–Kitaev pass src/mqt/bench/custom_solovay_kitaev.py	New CustomSolovayKitaev class (subclass of Qiskit SolovayKitaev) overriding run to synthesize non-parameterized single-qubit gates in-DAG, skipping unsupported ops and replacing nodes with approximation DAGs.
Benchmark generation wiring src/mqt/bench/benchmark_generation.py	Switched from Qiskit's SolovayKitaev to local CustomSolovayKitaev when target.description == "clifford+t" in native-gate synthesis path.

Sequence Diagram(s)

sequenceDiagram
    actor User
    participant BenchmarkGen as BenchmarkGeneration
    participant PassMgr as PassManager
    participant CSVK as CustomSolovayKitaev
    participant Transpiler as Transpiler/DAG
    User->>BenchmarkGen: request native-gate synthesis (clifford+t)
    BenchmarkGen->>PassMgr: create PassManager with CustomSolovayKitaev
    PassMgr->>CSVK: run pass on DAG
    CSVK->>Transpiler: inspect op nodes, call internal SK to synthesize approximations
    CSVK->>Transpiler: replace op nodes with approximation DAGs
    PassMgr->>BenchmarkGen: return transformed DAG/circuit
    BenchmarkGen->>User: return native-gate circuit

Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

Poem

🐰
Nine qubits curl beneath my paw,
I sniff each stitch, then hop and draw,
Mid-circuit measures, fixes true,
I nudge a Z, a clever X too,
Hooray — the benchmark blooms anew 🥕

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Out of Scope Changes check	⚠️ Warning	All changes are directly related to implementing Shor's 9 Qubit Code. However, a CustomSolovayKitaev pass was introduced which appears unrelated to the core benchmark implementation.	Clarify why CustomSolovayKitaev was added and whether it relates to the Shor's code benchmark or represents a separate enhancement. If unrelated, consider splitting into a separate PR.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely summarizes the main change: implementing Shor's 9 Qubit Code as a new benchmark.
Description check	✅ Passed	The PR description follows the template structure with a description summary, issue reference (#790), and a completed checklist matching the required sections.
Linked Issues check	✅ Passed	The PR fully addresses issue #790's requirements: implements Shor's 9-qubit code with 9 physical qubits, demonstrates mid-circuit measurement-based error detection/correction, and applies corrective gates after measurement.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing touches

• 📝 Generate docstrings

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[coderabbitai]

Actionable comments posted: 6

🤖 Fix all issues with AI agents

In `@src/mqt/bench/benchmarks/shors_nine_qubit_code.py`:
- Around line 260-263: Fix the typo in the encoding docstring by replacing the
incorrect ket string "|--->⟩" with the correct "|---⟩" in the Encoding section;
locate the docstring that describes "Phase-flip encoding" (the text containing
"|0⟩ → |+++⟩ and |1⟩ → |--->⟩") and edit that string to remove the extra ">" so
it reads "|0⟩ → |+++⟩ and |1⟩ → |---⟩".
- Around line 84-110: The phase-flip syndrome circuit in
_get_nine_qubit_shors_code_phase_flip_syndrome_extraction_circuit uses
ancilla-controlled CNOTs but lacks an explicit note of which X‑parity
stabilizers are being measured; update the function docstring or add a brief
inline comment near the uses of phase_flip_syndrome and logical_qubit to state
that the circuit measures X⊗X⊗X parity on each 3‑qubit block (qubits 0–2, 3–5,
6–8) using ancillas phase_flip_syndrome[0] and phase_flip_syndrome[1], and
clarify the ancilla measurement mapping (01 → block 1, 10 → block 2, 11 → block
3, 00 → no error) so the CX patterns are immediately understandable.
- Around line 113-144: Update the docstring of
_apply_nine_qubit_shors_code_bit_flip_correction to explicitly state the assumed
Qiskit little-endian bit order (ClassicalRegister index 0 = least-significant
bit) and explain how the integer syndrome patterns (e.g., 0b000001, 0b000010,
etc.) map to the three-bit syndrome values for each 3-qubit block so future
maintainers understand the mapping and conditional X corrections.

In `@src/mqt/bench/custom_solovay_kitaev.py`:
- Around line 62-63: Fix the awkward comment in
src/mqt/bench/custom_solovay_kitaev.py by rephrasing the sentence that currently
reads "as we know it comes from a Qiskit gate, as this we know it will generate"
to a single clear clause; for example, change the comment inside the
input-matrix check block to: "we do not check the input matrix because it comes
from a Qiskit gate and therefore will be a valid SU(2) matrix" so it reads
grammatically correct and unambiguous.
- Around line 52-54: The docstring in src/mqt/bench/custom_solovay_kitaev.py
incorrectly states that TranspilerError is raised when a gate lacks to_matrix,
but the implementation silently continues for such gates; update the docstring
(or remove the Raises section) to reflect the actual behavior by removing the
TranspilerError claim and explaining that gates without a to_matrix method are
skipped, referencing the to_matrix check and TranspilerError symbol so reviewers
can locate the change.
- Around line 68-72: The code directly accesses the private attribute self._sk
and calls self._sk.run(...) then uses dag.substitute_node_with_dag(...); add a
clear FIXME comment above this usage that explains why accessing self._sk is
necessary (e.g., to call the Solovay–Kitaev implementation for approximation)
and include a removal trigger note "Remove once qiskit minimum version is bumped
to 2.0.0" so future maintainers know when to refactor; keep the comment adjacent
to the lines invoking self._sk, reference self._sk, _sk.run, recursion_degree
and dag.substitute_node_with_dag in the comment, and don’t change behavior—this
is purely documentation for the fragile private-attribute dependency.

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI agents

In `@src/mqt/bench/benchmarks/shors_nine_qubit_code.py`:
- Around line 306-313: The current validation only checks divisibility by 17 and
allows invalid values like 0 or negative multiples; update the guard around
num_qubits so it requires num_qubits to be a positive multiple of 17 (e.g.,
num_qubits >= 17 and num_qubits % 17 == 0) and raise a clear ValueError
otherwise; adjust the existing validation block that computes num_logical_qubits
and calls _create_single_logical_qubit_circuit to rely on this stricter check.