Split out quantum-info and synthesis crates from accelerate #14342

mtreinish · 2025-05-10T01:46:11Z

Summary

This commit completes the decomposition of thw qiskit-accelerate crate into separate compilation units based on functionality. There are two new creates created in this commit, qiskit-synthesis and qiskit-quantum-info. The logical split between the crates is synthesis is the crate used to synthesize mathematical operations into quantum circuits, while the quantum-info crate is used to contain the data structures and functions used to model quantum information. In an earlier draft synthesis was just a submodule inside the quantum-info crate, but before finalizing this commit it was split out because the external dependencies of the functionality were distinct enough that keeping them seperate provide real advantages. Concretely this was primarily driven by being able to exclude pulling in faer and faer-ext when building the cext crate, since cext only depends currently on qiskit-quantum-info and qiskit-circuit.

Details and comments

This commit finishes the migration to the new internal dependency structure so that accelerate is towards the end of the tree and nothing depends on it accept for the Python interface crate. Here is the new dependency tree of qiskit-* crates of qiskit-pyext:

qiskit-pyext
├── qiskit-accelerate
    ├── qiskit-transpiler
├── qiskit-cext
    ├── qiskit-quantum-info
    ├── qiskit-circuit
├── qiskit-transpiler
    ├── qiskit-synthesis
    ├── qiskit-quantum-info
    ├── qiskit-circuit
├── qiskit-synthesis
    ├── qiskit-quantum-info
    ├── qiskit-circuit
├── qiskit-quantum-info
    ├── qiskit-circuit
├── qiskit-circuit

qiskit-bot · 2025-05-10T01:46:16Z

One or more of the following people are relevant to this code:

@Qiskit/terra-core
@levbishop

This commit completes the decomposition of thw qiskit-accelerate crate into separate compilation units based on functionality. There are two new creates created in this commit, qiskit-synthesis and qiskit-quantum-info. The logical split between the crates is synthesis is the crate used to synthesize mathematical operations into quantum circuits, while the quantum-info crate is used to contain the data structures and functions used to model quantum information. In an earlier draft synthesis was just a submodule inside the quantum-info crate, but before finalizing this commit it was split out because the external dependencies of the functionality were distinct enough that keeping them seperate provide real advantages. Concretely this was primarily driven by being able to exclude pulling in faer and faer-ext when building the cext crate, since cext only depends currently on qiskit-quantum-info and qiskit-circuit. This commit finishes the migration to the new internal dependency structure so that accelerate is towards the end of the tree and nothing depends on it accept for the Python interface crate. qiskit-pyext ├── qiskit-accelerate ├── qiskit-transpiler ├── qiskit-cext ├── qiskit-quantum-info ├── qiskit-circuit ├── qiskit-transpiler ├── qiskit-synthesis ├── qiskit-quantum-info ├── qiskit-circuit ├── qiskit-synthesis ├── qiskit-quantum-info ├── qiskit-circuit ├── qiskit-quantum-info ├── qiskit-circuit ├── qiskit-circuit

coveralls · 2025-05-10T02:42:26Z

Pull Request Test Coverage Report for Build 15170092988

Details

24 of 29 (82.76%) changed or added relevant lines in 4 files are covered.
9 unchanged lines in 3 files lost coverage.
Overall coverage increased (+0.01%) to 88.317%

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
crates/transpiler/src/passes/barrier_before_final_measurement.rs	0	1	0.0%
crates/quantum_info/src/sparse_pauli_op.rs	5	9	55.56%

Files with Coverage Reduction	New Missed Lines	%
crates/qasm2/src/expr.rs	1	94.23%
crates/qasm2/src/lex.rs	3	91.98%
crates/circuit/src/symbol_expr.rs	5	75.31%

Totals
Change from base Build 15169979939:	0.01%
Covered Lines:	78455
Relevant Lines:	88833

💛 - Coveralls

ShellyGarion · 2025-05-10T15:04:06Z

Thanks for re-organizing!

isometry.rs - should be moved to the circuit_library?
uc_gate.rs - should be moved to the circuit_library?

mtreinish · 2025-05-10T22:25:20Z

Thanks for re-organizing!

* `isometry.rs` - should be moved to the `circuit_library`?

* `uc_gate.rs` - should be moved to the `circuit_library`?

I left them where they were because it's not a full synthesis implementation. I only ported some of the numerics from the isometry code last year (which is both of the files). It was just enough to avoid buggy pieces in Apple's accelerate BLAS implementation that were causing spurious failures on arm64 macOS. I think we should be saving the synthesis crate for full synthesis algorithms, not partial things like these. If we port the full isometry code to Rust then I'd say we should move it over to qiskit-synthesis.

…um-info

ShellyGarion

Thanks for re-organizing the code. I only have a few questions:

why rayon_ext.rs was moved to quantum_info?
why test.rs was moved to quantum_info? perhaps it would be good to update the file name to something more meaningful.
pauli_exp_val.rs and sampled_exp_val.rs - should they be moved to quantum_info?
optimize_1q_gates - should it be moved to qiskit_transplier?

mtreinish · 2025-05-20T14:04:38Z

Thanks for re-organizing the code. I only have a few questions:

* why `rayon_ext.rs` was moved to quantum_info?

This is an implementation of traits for the rayon data parallelism library that is used in the sparse pauli op code. It's only used there and is specific to that code. It doesn't make sense to live anywhere else.

* why `test.rs` was moved to quantum_info? perhaps it would be good to update the file name to something more meaningful.

This is specifically test code for the above rayon module. I can rename it but I'd rather do that in a follow up just to try and minimize the diff here since the more code we change in this PR makes potential future issues difficult to bisect or revert.

* `pauli_exp_val.rs` and `sampled_exp_val.rs` - should they be moved to quantum_info?

pauli_exp_val.rs falls into the same category as isometry. It's just a small function accelerating a larger set of functionality. It was only ported to rust because it was written in Cython at the time we started using rust.

sampled_exp_val.rs I didn't put in quantum info because it's a standalone function that lives in qiskit/result in python. It operates on a distribution dictionary and operator string and returns the computed expectation value. I can move it if you think that makes sense, but it doesn't really interact with the quanutm info objects directly.

* `optimize_1q_gates` - should it be moved to qiskit_transplier?

This is a similar reason to isometry. It accelerates some of the internal quaternion calculations used in the transpiler pass. It's not a standalone pass and doesn't work without python which is part of what accelerate is for.

ShellyGarion

Thanks for the explanation. I see now that the code in accelerate is not ready to be moved to other parts.
One last follow-up question:
should sampled_exp_val.rs be moved into accelerate/results?

mtreinish · 2025-05-21T14:55:07Z

Potentially, the directory based split never really caught on in accelerate, so we never really have add a 1:1 mapping of the Python directory structure in the rust crates. That being said it does seem to make sense there since it's about result post-processing. I'd suggest we do that in a separate PR to minimize the diff here about the crate re-organization.

ShellyGarion

Thanks for the explanations. I've approved your PR. Let's merge it carefully, since everyone will have to rebase their PRs afterwards :)

…um-info

…4342) * Split out quantum-info and synthesis crates from accelerate This commit completes the decomposition of thw qiskit-accelerate crate into separate compilation units based on functionality. There are two new creates created in this commit, qiskit-synthesis and qiskit-quantum-info. The logical split between the crates is synthesis is the crate used to synthesize mathematical operations into quantum circuits, while the quantum-info crate is used to contain the data structures and functions used to model quantum information. In an earlier draft synthesis was just a submodule inside the quantum-info crate, but before finalizing this commit it was split out because the external dependencies of the functionality were distinct enough that keeping them seperate provide real advantages. Concretely this was primarily driven by being able to exclude pulling in faer and faer-ext when building the cext crate, since cext only depends currently on qiskit-quantum-info and qiskit-circuit. This commit finishes the migration to the new internal dependency structure so that accelerate is towards the end of the tree and nothing depends on it accept for the Python interface crate. qiskit-pyext ├── qiskit-accelerate ├── qiskit-transpiler ├── qiskit-cext ├── qiskit-quantum-info ├── qiskit-circuit ├── qiskit-transpiler ├── qiskit-synthesis ├── qiskit-quantum-info ├── qiskit-circuit ├── qiskit-synthesis ├── qiskit-quantum-info ├── qiskit-circuit ├── qiskit-quantum-info ├── qiskit-circuit ├── qiskit-circuit * Update dependency package in cbindgen config * Remove unused requirements * Bump cargo.lock

mtreinish added the Changelog: None Do not include in changelog label May 10, 2025

mtreinish requested a review from a team as a code owner May 10, 2025 01:46

mtreinish added the Rust This PR or issue is related to Rust code in the repository label May 10, 2025

mtreinish added this to the 2.1.0 milestone May 10, 2025

mtreinish force-pushed the split-synthesis-quantum-info branch from e36d991 to 6166bc0 Compare May 10, 2025 01:49

mtreinish added 2 commits May 9, 2025 22:01

Update dependency package in cbindgen config

ca19d33

Remove unused requirements

3b16fae

eliarbel assigned mtreinish and ShellyGarion May 15, 2025

Merge remote-tracking branch 'origin/main' into split-synthesis-quant…

f7dbf41

…um-info

mtreinish force-pushed the split-synthesis-quantum-info branch from 05fdb5c to f7dbf41 Compare May 17, 2025 11:58

mtreinish added 2 commits May 19, 2025 17:35

Merge branch 'main' into split-synthesis-quantum-info

9eebbe2

Bump cargo.lock

5129cc6

ShellyGarion reviewed May 20, 2025

View reviewed changes

mtreinish removed their assignment May 20, 2025

ShellyGarion self-requested a review May 21, 2025 08:10

ShellyGarion reviewed May 21, 2025

View reviewed changes

ShellyGarion previously approved these changes May 21, 2025

View reviewed changes

Merge remote-tracking branch 'origin/main' into split-synthesis-quant…

b50b51d

…um-info

mtreinish dismissed ShellyGarion’s stale review via b50b51d May 21, 2025 18:43

ShellyGarion approved these changes May 22, 2025

View reviewed changes

mtreinish added this pull request to the merge queue May 22, 2025

Merged via the queue into Qiskit:main with commit 22367fc May 22, 2025
24 checks passed

jakelishman mentioned this pull request May 29, 2025

Add QubitSparsePauli and QubitSparsePauliList classes to quantum_info #14283

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Split out quantum-info and synthesis crates from accelerate #14342

Split out quantum-info and synthesis crates from accelerate #14342

Uh oh!

mtreinish commented May 10, 2025 •

edited

Loading

Uh oh!

qiskit-bot commented May 10, 2025

Uh oh!

coveralls commented May 10, 2025 •

edited

Loading

Uh oh!

ShellyGarion commented May 10, 2025

Uh oh!

mtreinish commented May 10, 2025 •

edited

Loading

Uh oh!

ShellyGarion left a comment

Uh oh!

mtreinish commented May 20, 2025

Uh oh!

ShellyGarion left a comment

Uh oh!

mtreinish commented May 21, 2025

Uh oh!

ShellyGarion left a comment

Uh oh!

Uh oh!

Uh oh!

Split out quantum-info and synthesis crates from accelerate #14342

Split out quantum-info and synthesis crates from accelerate #14342

Uh oh!

Conversation

mtreinish commented May 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Details and comments

Uh oh!

qiskit-bot commented May 10, 2025

Uh oh!

coveralls commented May 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pull Request Test Coverage Report for Build 15170092988

Details

💛 - Coveralls

Uh oh!

ShellyGarion commented May 10, 2025

Uh oh!

mtreinish commented May 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ShellyGarion left a comment

Choose a reason for hiding this comment

Uh oh!

mtreinish commented May 20, 2025

Uh oh!

ShellyGarion left a comment

Choose a reason for hiding this comment

Uh oh!

mtreinish commented May 21, 2025

Uh oh!

ShellyGarion left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

mtreinish commented May 10, 2025 •

edited

Loading

coveralls commented May 10, 2025 •

edited

Loading

mtreinish commented May 10, 2025 •

edited

Loading