Add support for H-boxes with arbitrary complex labels #400
Conversation
There was a problem hiding this comment.
Pull request overview
This PR adds support for H-boxes with arbitrary complex labels, similar to the existing support for Z-boxes with complex labels. Previously, H-boxes could only represent phases as multiples of π. This enhancement allows H-boxes to represent matrices with arbitrary complex entries in the bottom-right position, enabling more general quantum operations.
Changes:
- Added utility functions (
get_h_box_label,set_h_box_label,is_standard_hbox,hbox_has_complex_label) for managing H-box complex labels - Updated tensor computation to handle H-boxes with complex labels
- Enhanced JSON and TikZ serialization to support round-trip conversion of H-box complex labels
- Modified multiple rewrite rules to correctly distinguish between standard Hadamard gates (label=-1) and H-boxes with arbitrary complex labels
- Added comprehensive test coverage for the new functionality
Reviewed changes
Copilot reviewed 17 out of 17 changed files in this pull request and generated 3 comments.
Show a summary per file
| File | Description |
|---|---|
| pyzx/utils.py | Added four new helper functions for managing H-box complex labels, following the same pattern as Z-box labels |
| pyzx/tensor.py | Modified H_to_tensor to accept optional complex label parameter and updated tensorfy_naive to use labels when present |
| pyzx/tikz.py | Enhanced TikZ import/export to parse and serialize complex labels for H-boxes, hiding standard Hadamard labels (-1) |
| pyzx/drawing.py | Updated matplotlib and JSON visualization to display complex labels for H-boxes |
| pyzx/graph/jsonparser.py | Added JSON serialization support for H-box labels in both new and legacy formats |
| pyzx/graph/base.py | Updated documentation for add_vertex to clarify usage of label functions |
| pyzx/rewrite_rules/*.py | Updated 7 rewrite rule files to use is_standard_hbox instead of direct phase checks, ensuring rules only apply to standard Hadamard gates |
| tests/test_tensor.py | Added tests for H_to_tensor with labels and tensorfy with complex-labeled H-boxes |
| tests/test_jsonparser.py | Added comprehensive tests for JSON and TikZ round-trip with H-box labels |
| tests/test_hbox.py | Added extensive tests for helper functions and rewrite rules with complex labels |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
dlyongemallo
force-pushed
the
167-H-box_arbitrary_complex_labels
branch
from
af0c73f to
d1364d0
Compare
| newv = g.add_vertex(g.type(v[j]), qubit=q, row=r) | ||
| g.set_phase(newv, g.phase(v[j])) | ||
| # Copy complex label if H-box has one. | ||
| if g.type(v[j]) == VertexType.H_BOX and hbox_has_complex_label(g, v[j]): |
There was a problem hiding this comment.
v[i] and v[j] are always Z and X spiders here, and there phase I believe is always a Pauli, so this case here never applies.
There was a problem hiding this comment.
What about the check on line 121 below? I think it has the same applicability as this check here, so if this one is removed, that one should as well, for consistency.
dlyongemallo
force-pushed
the
167-H-box_arbitrary_complex_labels
branch
from
d1364d0 to
cccbf40
Compare
| newv = g.add_vertex(g.type(v[j]), qubit=q, row=r) | ||
| g.set_phase(newv, g.phase(v[j])) | ||
| # Copy complex label if H-box has one. | ||
| if g.type(v[j]) == VertexType.H_BOX and hbox_has_complex_label(g, v[j]): |
There was a problem hiding this comment.
What about the check on line 121 below? I think it has the same applicability as this check here, so if this one is removed, that one should as well, for consistency.
| etab[upair(n1,n2)] = [0, 0] | ||
| etab[upair(n1,n2)][0] += 1 | ||
|
|
||
| if g.type(v1) == VertexType.H_BOX or g.type(v2) == VertexType.H_BOX: # x-h bialgebra |
There was a problem hiding this comment.
I am referring to this check for H_BOX.
Based on the git history, it was added in commit 9dacf64 to editor_actions.py, before this was refactored here in commit ef23f18. In this line there was originally a check for X-H algebra as well.
So the code is inconsistent right now. check_bialgebra only allows Z-X pairs, but unsafe_bialegra (partially) handles H_BOX scalars. Should I remove both references (the existing one and the one I added) to H_BOX in unsafe_bialgebra? Or was the removal of X-H from check_bialgebra during the refactoring changes an accidental oversight?
2 participants
Fixes #167.