✨ Add optional MQT Core compilation by simon1hofmann · Pull Request #1027 · munich-quantum-toolkit/bench

simon1hofmann · 2026-09-12T11:37:05Z

🤖 AI text below 🤖

Description

Add optional MQT Core compilation through compiler="mqt" or --compiler mqt. Qiskit remains the default, and both compilers return QuantumCircuit.

The mqt extra pins Core 4.1.1 and Qiskit 2.5.x. The base package retains Qiskit 2.2.3 support. Native compilation keeps the input width and ignores device placement; mapped compilation enforces device width and connectivity. Both preserve measurement destinations. QASM/QPY output records compiler provenance alongside benchmark options and seeds.

Core uses a fixed pipeline (opt_level=2), seed 10, and four mapping trials. Pass CompilationOptions in Python or use --compiler-seed and --mapping-trials in the CLI. Setting one CLI control retains the other default. Invalid CLI combinations produce usage errors.

QIR export is stacked separately in #1094.

Validation

Full suite with Core 4.1.1 / Qiskit 2.5.2 on Python 3.13: 874 passed.
Full suite without Core / Qiskit 2.6.0 on Python 3.13: 707 passed.
Full suite at the Qiskit 2.2.3 floor on Python 3.11: 707 passed.
Each suite reports four expected failures for the known Qiskit fixed-width carry bug. Installations without Core skip its test module.
All lint hooks passed, including type checking with the optional extra.
Earlier documentation validation rendered the docs with notebook execution disabled; its 29 warnings also occurred on the preceding PR head.

Updated against current main. Fresh CI is pending.

Codex assisted with implementation, updates, and checks.

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.

If PR contains AI-assisted content:

Any agent that created, edited, or submitted GitHub content was explicitly authorized for that scope, as required by our AI Usage Guidelines.
Every agent-authored or agent-edited public text body begins with the visible disclosure 🤖 *AI text below* 🤖 (titles are exempt).
I have disclosed AI assistance in the PR description.
I confirm that I have personally reviewed and understood all AI-generated content, and accept full responsibility for it.

codecov · 2026-09-12T11:42:02Z

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

Assisted-by: GPT-6 via Codex

Include the MQT Core coverage report in the combined CI upload. Assisted-by: GPT-6 via Codex

🤖 *AI text below* 🤖 Pin Core main plus the controlled-composite fix and expose CompilationOptions with a fixed seed and mapping trial count, including mirror recompilation. Remove circuit preprocessing and parameter remapping now handled by Core. Configure MLIR for CI source builds and cover option forwarding, reproducibility, and parameter-vector identity. Assisted-by: GPT-6 via Codex

🤖 *AI text below* 🤖 Use the Core fix that handles composite controls within the existing decomposition pass and avoids extra single-qubit merging. Assisted-by: GPT-6 via Codex

🤖 *AI text below* 🤖 The reusable lint job configures sccache without installing it when stub checks are disabled, so building Core for ty fails. Run the existing ty hook in the Core job, which already provides MLIR and builds the dependency. Require that job whenever either tests or lint are requested. Assisted-by: GPT-6 via Codex

🤖 *AI text below* 🤖 Pin the integrated Core changes for native trapped-ion gates and generic fixed RX/RY/RZ capabilities while preserving controlled-composite support. Restore Rigetti pulse names during target validation and cover symbolic parameters, native semantics, and rejected target definitions. Assisted-by: GPT-6 via Codex

🤖 *AI text below* 🤖 Use canonical native definitions so Core preserves numeric and symbolic pulses through repeated compilation. Omit capabilities wider than the input circuit at the native-gate level. Pin Core with the matching import and fixed-parameter compatibility fixes. Assisted-by: GPT-6 via Codex

🤖 *AI text below* 🤖 Pin the rebased integration branch with the fixed-parameter and native-gate CI tests. Keep the dependency lockfile and documented revision aligned. Assisted-by: GPT-6 via Codex

Extract the target-test isolation and device capability checks from #1027. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-6 via Codex

Cover bounded target aliases and legacy Qiskit serialization. Extract registry isolation from #1027. Remove the extra test file. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de>

Extracted from #1027; retain original authorship. Assisted-by: GPT-6 via Codex Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

* ✨ Add AQT IBEX device and native gate set Extracted from #1027; retain original authorship. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-6 via Codex * 📝 Reference #1084 in changelog * ✅ Isolate target registries and check device control-flow limits Extract the target-test isolation and device capability checks from #1027. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-6 via Codex * ✅ Keep AQT tests in the existing target test module * 📝 Keep migration guidance and changelog links in their sections --------- Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

* 🔥 Remove retired IBM device models Extracted from #1027; retain original authorship. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-6 via Codex * 📝 Reference #1089 in changelog * 📝 Keep migration guidance and changelog links in their sections * 📝 Distinguish gate-set and device names in CLI help * 🔥 Remove tests for retired IBM catalogue names --------- Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

* 📝 Fix API documentation markup and notebook builds Extracted from #1027; retain original authorship. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-6 via Codex * 📝 Reference #1088 in changelog * 📝 Keep migration guidance and changelog links in their sections * Preserve citations and signatures in Markdown documentation * Remove ineffective sphinx-llm version cap sphinx-llm 1.0 and 1.1 produce identical per-page Markdown. Version 1.1 exposes existing unsupported-node warnings; the cap does not restore missing content. Refs: NVIDIA/sphinx-llm#153 --------- Signed-off-by: simon1hofmann <119581649+simon1hofmann@users.noreply.github.com> Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

* ✨ Add IBM Nighthawk device model Extracted from #1027; retain original authorship. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-6 via Codex * 📝 Reference #1085 in changelog * ✅ Keep Nighthawk tests in the existing target test module * 📝 Keep migration guidance and changelog links in their sections --------- Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

* 🐛 Preserve IonQ and Rigetti native gate semantics Extracted from #1027; retain original authorship. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-6 via Codex * 📝 Reference #1086 in changelog * ✅ Keep native regression tests in the target test module Cover bounded target aliases and legacy Qiskit serialization. Extract registry isolation from #1027. Remove the extra test file. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> * 🐛 Preserve physical placements shared by native aliases Keep equivalent RX aliases on every supported physical site, including targets that also expose a standard X gate. Retain phase-correct local lowering and add mapped regressions. Assisted-by: GPT-6 via Codex * ✅ Check mixed aliases at native and mapped levels * Test minimum Qiskit support and complete mirror layouts * ✅ Verify native gate constraints and restore minimums session Cover arbitrary U/RX compilation at all optimization levels for native and mapped targets, and check session equivalences remain unchanged. Keep local lowering because preset passes do not consistently enforce fixed R/RX angles. Restore noxfile.py to main. Assisted-by: GPT-6 via Codex --------- Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

🤖 *AI text below* 🤖 Keep the merged device and native-alias changes from main while preserving Core/QIR and fractional-Heron support. Reconcile the changelog and retain the upstream placement and bounds regressions. Assisted-by: GPT-5 via Codex

* 🐛 Preserve IonQ and Rigetti native gate semantics Extracted from #1027; retain original authorship. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-6 via Codex * 📝 Reference #1086 in changelog * ✨ Add fractional Heron compilation targets Extracted from #1027; retain original authorship. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> Assisted-by: GPT-6 via Codex * 📝 Reference #1087 in changelog * ✅ Keep native regression tests in the target test module Cover bounded target aliases and legacy Qiskit serialization. Extract registry isolation from #1027. Remove the extra test file. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> * ✅ Keep fractional target tests in the existing test module Cover old Qiskit rejection and narrower caller-defined RZZ bounds. Co-authored-by: Simon Hofmann <simon.t.hofmann@tum.de> * 🐛 Preserve physical placements shared by native aliases Keep equivalent RX aliases on every supported physical site, including targets that also expose a standard X gate. Retain phase-correct local lowering and add mapped regressions. Assisted-by: GPT-6 via Codex * ✅ Check mixed aliases at native and mapped levels * 🐛 Serialize documentation builds that share notebook output Use the same notebook build ordering as #1088 to prevent concurrent HTML and Markdown builds from deleting each other’s example output. * Test minimum Qiskit support and complete mirror layouts * Test the fractional target compatibility error on current Qiskit * ♻️ Simplify fractional Heron registration and angle wrapping Use a shared packaging.version check while keeping Qiskit 2.1.2 support. Use theta parameters, simplify the local wrapping traversal and phase correction, and cover all optimization levels and large angles. Assisted-by: GPT-6 via Codex * ✅ Verify native gate constraints and restore minimums session Cover arbitrary U/RX compilation at all optimization levels for native and mapped targets, and check session equivalences remain unchanged. Keep local lowering because preset passes do not consistently enforce fixed R/RX angles. Restore noxfile.py to main. Assisted-by: GPT-6 via Codex * ⬆️ Require Qiskit 2.2.3 and simplify fractional targets Register fractional Heron targets unconditionally, remove obsolete compatibility guards and tests, and document the new Qiskit minimum and upgrade command. Assisted-by: GPT-6 via Codex --------- Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

Assisted-by: GPT-5 via Codex

Extract target documentation, remove redundant target tests and dead packaging metadata, and restore main’s documentation dependency. Assisted-by: GPT-5 via Codex

Include the complete Cepheus architecture and consolidated parameter guide. Preserve the optional compiler dependency pins and minimum-version base support. Assisted-by: GPT-6 via Codex

burgholzer · 2026-10-25T22:46:14Z

I haven't looked at this in detail yet, but one thing that has become clear to me:
Let's move to Qiskit 2.5+ for this change. Already from quickly glancing over the changes, maintaining compatibility with anything less will become painful.

I also stumbled over this separately in munich-quantum-toolkit/core#2850 where I now experimented with raising the lower bound as well. This would fit well with the PR here.

github-project-automation Bot added this to MQT Applications Sep 12, 2026

github-project-automation Bot moved this to In Progress in MQT Applications Sep 12, 2026

simon1hofmann changed the title ~~✨ Add MQT Core as an optional compiler~~ ✨ Add MQT Core compilation and QIR export Sep 12, 2026

simon1hofmann added feature New feature or request python Pull requests that update Python code mqt.bench Issues that affect mqt.bench labels Sep 17, 2026

simon1hofmann force-pushed the feat/mqt-core-compiler branch from 509d184 to df2aeca Compare September 17, 2026 08:49

simon1hofmann self-assigned this Sep 24, 2026

simon1hofmann force-pushed the feat/mqt-core-compiler branch from a51d1ac to 91949dd Compare October 2, 2026 11:21

burgholzer mentioned this pull request Oct 2, 2026

✨ Synthesize native parameterized gate sets munich-quantum-toolkit/core#2578

Merged

10 tasks done

burgholzer force-pushed the feat/mqt-core-compiler branch from 91949dd to 68d4f2a Compare October 2, 2026 14:06

burgholzer self-assigned this Oct 2, 2026

burgholzer force-pushed the feat/mqt-core-compiler branch 3 times, most recently from 93867e1 to 3c3ab51 Compare October 4, 2026 15:09

This comment has been minimized.

Sign in to view

simon1hofmann added 10 commits October 15, 2026 00:22

✨ Add MQT Core as an optional compiler

9befe62

Assisted-by: GPT-6 via Codex

📝 Reference compiler PR in changelog

8424058

Assisted-by: GPT-6 via Codex

✨ Add QIR export as LLVM text and bitcode

643f02f

Assisted-by: GPT-6 via Codex

✅ Cover compiler and export edge cases

2d62523

Include the MQT Core coverage report in the combined CI upload. Assisted-by: GPT-6 via Codex

⚡️ Pin Core with selective control decomposition

79fc8c5

🤖 *AI text below* 🤖 Use the Core fix that handles composite controls within the existing decomposition pass and avoids extra single-qubit merging. Assisted-by: GPT-6 via Codex

🔗 Refresh Core integration pin

c611b46

🤖 *AI text below* 🤖 Pin the rebased integration branch with the fixed-parameter and native-gate CI tests. Keep the dependency lockfile and documented revision aligned. Assisted-by: GPT-6 via Codex

simon1hofmann mentioned this pull request Oct 21, 2026

✨ Add fractional Heron compilation targets #1087

Merged

11 of 12 tasks

This was referenced Oct 21, 2026

📝 Fix API documentation markup and notebook builds #1088

Merged

🔥 Remove retired IBM device models #1089

Merged

✅ Isolate target registry tests #1090

Merged

simon1hofmann added a commit that referenced this pull request Oct 21, 2026

✅ Isolate target registry tests (#1090)

5a8c254

Extracted from #1027; retain original authorship. Assisted-by: GPT-6 via Codex Co-authored-by: Lukas Burgholzer <burgholzer@me.com>

simon1hofmann added 2 commits October 21, 2026 21:07

🔀 Update Core integration from main

72ffd39

Assisted-by: GPT-5 via Codex

♻️ Separate QIR export from Core integration

d933130

Assisted-by: GPT-5 via Codex

simon1hofmann mentioned this pull request Oct 21, 2026

✨ Add QIR export with Base and Adaptive profiles #1094

Open

11 of 12 tasks

simon1hofmann changed the title ~~✨ Add MQT Core compilation and QIR export~~ ✨ Add optional MQT Core compilation Oct 21, 2026

♻️ Keep Core integration focused

5bf43e3

Extract target documentation, remove redundant target tests and dead packaging metadata, and restore main’s documentation dependency. Assisted-by: GPT-5 via Codex

simon1hofmann mentioned this pull request Oct 21, 2026

✨ Add full Cepheus architecture and simplify target documentation #1095

Merged

12 tasks done

🔀 Merge current main into optional Core compiler

c814879

Include the complete Cepheus architecture and consolidated parameter guide. Preserve the optional compiler dependency pins and minimum-version base support. Assisted-by: GPT-6 via Codex

simon1hofmann added this pull request to stack #1097 October 22, 2026 11:21

simon1hofmann added 2 commits October 22, 2026 13:57

🔀 Update Core compiler after #1093 merge

d2d8f9a

✨ Add Core CLI controls and trim integration docs

60ba2f4

simon1hofmann marked this pull request as ready for review October 22, 2026 13:39

simon1hofmann requested a review from burgholzer October 22, 2026 13:39

simon1hofmann added 3 commits October 23, 2026 22:47

Merge main into MQT Core compiler branch

c743c00

Merge main into MQT Core compiler branch

13b500a

Merge CLI and export updates into MQT Core compiler branch

c909e73

Navigation Menu

Repository navigation

✨ Add optional MQT Core compilation - #1027

✨ Add optional MQT Core compilation#1027
simon1hofmann wants to merge 47 commits into
mainfrom
feat/mqt-core-compiler

simon1hofmann commented Sep 12, 2026 •

edited

Loading

Uh oh!

codecov Bot commented Sep 12, 2026 •

edited

Loading

Uh oh!

This comment has been minimized.

burgholzer commented Oct 25, 2026

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

simon1hofmann commented Sep 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Validation

Checklist

Uh oh!

codecov Bot commented Sep 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

This comment has been minimized.

burgholzer commented Oct 25, 2026

Uh oh!

This branch has not been deployed

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

simon1hofmann commented Sep 12, 2026 •

edited

Loading

codecov Bot commented Sep 12, 2026 •

edited

Loading