Fix controlled GF2MulK decomposition when control is 0

[rupasri118]

Summary

Fixes #1951.

The permutation step in SynthesizeLRCircuit was implemented using direct Soquet rewiring. When GF2MulK was controlled, this permutation remained unconditional, causing the decomposed circuit to modify the target register even when the control qubit was 0.

This change represents the permutation using TwoBitSwap, making it an explicit Bloq that can be correctly controlled during decomposition.

Changes

• Replace direct Soquet swapping with TwoBitSwap in SynthesizeLRCircuit.
• Add a regression test for controlled GF2MulK decomposition.
• Verify that:
  • ctrl=0 leaves the input unchanged.
  • ctrl=1 performs multiplication by the specified constant.

Testing

• test_multiply_by_constant_controlled_decomposition — passed
• test_synthesize_lr_circuit — 8 passed
• gf2_multiplication_test.py excluding notebook tests — 78 passed before a long-running test was interrupted.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[rupasri118]

Updated the controlled GF2 multiplication decomposition to use explicit TwoBitSwap operations for controlled permutations while preserving wire reordering for uncontrolled synthesis.

Validation:

• Controlled multiplication decomposition test: passed.
• FLASQ Karatsuba span test: passed.
• Multiplication test suite: 299 passed, 3 skipped, 3 expected failures.
• Notebook test: passed after installing Graphviz.

[rupasri118]

Hi @Strilanc ! I've been working on a fix for this issue and have pushed the changes along with regression tests. Could you please review the proposed approach and let me know if any changes are needed? Thank you for your time and guidance!