Transpiler pass that converts a generic circuit to PBC

[ShellyGarion]

Summary

close #15451

For each of the standard 1-qubit and 2-qubit gates, that do not have parameters, or have only one parameter,
we currently map the gate to a list of equivalent Pauli rotations and a global phase.
Each element of the list is of the form ((Pauli string, phase rescale factor, [qubit indices]), global phase).
For gates that didn't have a phase (e.g. X)
the phase rescale factor is simply the phase of the rotation gate. The convention is
original_gate = PauliEvolutionGate(pauli, phase) * e^{i global_phase * phase}
The transpiler pass will output the corresponding PauliEvolutionGate.

Details and comments

[coveralls]

Pull Request Test Coverage Report for Build 22140508833

Warning: This coverage report may be inaccurate.

This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.

• For more information on this, see Tracking coverage changes with pull request builds.
• To avoid this issue with future PRs, see these Recommended CI Configurations.
• For a quick fix, rebase this PR at GitHub. Your next report should be accurate.

Details

• 260 of 272 (95.59%) changed or added relevant lines in 6 files are covered.
• 1198 unchanged lines in 38 files lost coverage.
• Overall coverage increased (+0.03%) to 87.996%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
crates/transpiler/src/passes/pbc_transformation.rs	249	261	95.4%

Files with Coverage Reduction	New Missed Lines	%
crates/cext/build.rs	1	96.77%
crates/cext/src/transpiler/passes/basis_translator.rs	1	94.74%
crates/cext/src/transpiler/passes/consolidate_blocks.rs	1	96.15%
crates/cext/src/transpiler/passes/elide_permutations.rs	1	95.45%
crates/cext/src/transpiler/passes/inverse_cancellation.rs	1	90.0%
crates/cext/src/transpiler/passes/split_2q_unitaries.rs	1	96.55%
crates/cext/src/transpiler/transpile_layout.rs	1	99.37%
crates/synthesis/src/clifford/bm_synthesis.rs	2	99.16%
crates/cext/src/transpiler/passes/remove_identity_equiv.rs	3	85.0%
crates/cext/src/circuit_library/iqp.rs	4	82.35%

Totals
Change from base Build 21468038257:	0.03%
Covered Lines:	100547
Relevant Lines:	114263

---

💛 - Coveralls

[alexanderivrii]

Thanks Shelly for working on this. Some quick comments:

• The pass should handle circuits with standard Z-measurements. Of course, it would leave them unchanged.
• Maybe it can also trivially handle barriers.
• Can we find a better name than PBCTransformation (since LitinskiTransformation is also a PBC transformation), maybe something like DirectPBCTranslation or TranslateToPBC?

[alexanderivrii]

Huge thanks Shelly for working on this, this pass is shaping up quite nicely. I haven't fully checked the heart of the pass (i.e. which Pauli-based decompositions we construct for different gates, and whether these can be improved), for now I have just left a few high-level comments that may help you finalize the rust code;

• please see my previous comments here, I don't think all are fully addressed yet
• we should support parameterized gates (as we have methods to multiply a parameter expression by a float factor)
• we should try to deduplicate the code (combine common logic when possible)
• we should avoid panic! and unreachable! when it makes sense

[alexanderivrii]

Thanks for the hard work, Shelly. If it's not too much to ask, could you please update using PauliEvolutionGate with PauliProductRotationGate, as done in #15753 for the Litinski transform.

[ShellyGarion]

failing tests are due to an error with binding parameters for parmatrized PauliProductRotationGate.
should be fixed after #15763 is merged.

[alexanderivrii]

Thanks Shelly for the quick updates to use PPRs. I have a few minor nitpicks about the comments + the docstrings, otherwise it is completely ready.

[alexanderivrii]

Could you please update this comment something like that this writes a standard gate $S$ as a product of evolutions $S = e^{-i \phi_1 P_1} \dots e^{i \phi_k P_k}$, but for pauli rotation gates the convention is to divide the phase by $2$.

[ShellyGarion]

done in 737156f

[alexanderivrii]

Same here: add a comment about the factor of 2 for pauli product rotations.

[ShellyGarion]

done in done in 737156f

[alexanderivrii]

same comment here: the pass converts a circuit to a circuit, not to a list.

[ShellyGarion]

done in 737156f

[alexanderivrii]

Thanks for the quick updates, LGTM!