fix: Multiple fixes to the pytket encoder #1226

aborgna-q · 2025-11-06T14:46:50Z

This PR is a compilation of fixes to the pytket encoder/decoder. With these implemented, the example from CQCL/hugr#2667 should work correctly.

The fixes include (one per commit):

Marking wires that carry no value as unsupported, since they cannot be attached to a pytket operation. (They will instead be included in an opaque subgraph / barrier op).
Don't use SiblingSubgraph to define the unsupported subgraphs. That structure is quite opinionated and brittle;
- it doesn't support non-value edges in the boundaries
- it has a complex definition for mapping between boundary ports and actual node ports (unnecessary in this case)
- requires running an expensive convexity checker on each instantiation, even when we know the set of nodes is valid
- and other annoyances
Mark any node with order edges as unsupported (so we don't lose those the edges the pytket roundtrip)
Fix issue with bits not appearing in the input/output missing from the pytket circuit's list
Add an initial hardcoded translator between native booleans and tket booleans, with a TODO to make the system extensible.
Avoid QAllocing qubits that don't get consumed afterwards.
Ensure we give loaded output parameters the correct type.
Don't try to encode wires with complex types that include parameters like [float] / [float, rotation].
Store input parameters to extra_subgraph (that cannot be encoded in the pytket circuit)

The main result is a bunch of tests now passing in tests.rs.

…r list

codecov · 2025-11-06T14:50:33Z

Codecov Report

❌ Patch coverage is 78.99160% with 100 lines in your changes missing coverage. Please review.
✅ Project coverage is 78.71%. Comparing base (b270af8) to head (9600a72).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
...et/src/serialize/pytket/config/type_translators.rs	40.00%	39 Missing ⚠️
tket/src/serialize/pytket/decoder/wires.rs	51.94%	34 Missing and 3 partials ⚠️
tket/src/serialize/pytket/encoder.rs	88.57%	2 Missing and 10 partials ⚠️
tket/src/serialize/pytket/decoder.rs	90.69%	0 Missing and 4 partials ⚠️
tket/src/serialize/pytket/decoder/subgraph.rs	92.59%	1 Missing and 1 partial ⚠️
tket/src/serialize/pytket.rs	0.00%	0 Missing and 1 partial ⚠️
tket/src/serialize/pytket/circuit.rs	93.33%	0 Missing and 1 partial ⚠️
tket/src/serialize/pytket/extension/core.rs	0.00%	1 Missing ⚠️
tket/src/serialize/pytket/extension/prelude.rs	75.00%	1 Missing ⚠️
tket/src/serialize/pytket/opaque.rs	91.66%	0 Missing and 1 partial ⚠️
... and 1 more

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1226      +/-   ##
==========================================
- Coverage   78.78%   78.71%   -0.07%     
==========================================
  Files         159      160       +1     
  Lines       20324    20592     +268     
  Branches    19222    19490     +268     
==========================================
+ Hits        16012    16210     +198     
- Misses       3332     3392      +60     
- Partials      980      990      +10

Flag	Coverage Δ
python	`92.65% <ø> (ø)`
qis-compiler	`68.40% <ø> (ø)`
rust	`78.35% <78.99%> (-0.07%)`	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

hugrbot · 2025-11-06T14:51:44Z

This PR contains breaking changes to the public Rust API.
Please deprecate the old API instead (if possible), or mark the PR with a ! to indicate a breaking change.

cargo-semver-checks summary

    Building tket v0.16.0 (current)
     Built [  40.538s] (current)
   Parsing tket v0.16.0 (current)
    Parsed [   0.097s] (current)
  Building tket v0.16.0 (baseline)
     Built [  40.579s] (baseline)
   Parsing tket v0.16.0 (baseline)
    Parsed [   0.080s] (baseline)
  Checking tket v0.16.0 -> v0.16.0 (assume minor change)
   Checked [   0.076s] 159 checks: 158 pass, 1 fail, 0 warn, 41 skip

--- failure auto_trait_impl_removed: auto trait no longer implemented ---

Description:
A public type has stopped implementing one or more auto traits. This can break downstream code that depends on the traits being implemented.
      ref: https://doc.rust-lang.org/reference/special-types-and-traits.html#auto-traits
     impl: https://github.com/obi1kenobi/cargo-semver-checks/tree/v0.45.0/src/lints/auto_trait_impl_removed.ron

Failed in:
type EncodedCircuit is no longer UnwindSafe, in /home/runner/work/tket2/tket2/PR_BRANCH/tket/src/serialize/pytket/circuit.rs:39
type EncodedCircuit is no longer RefUnwindSafe, in /home/runner/work/tket2/tket2/PR_BRANCH/tket/src/serialize/pytket/circuit.rs:39

   Summary semver requires new major version: 1 major and 0 minor checks failed
  Finished [  82.869s] tket
  Building tket-qsystem v0.22.0 (current)
     Built [  42.169s] (current)
   Parsing tket-qsystem v0.22.0 (current)
    Parsed [   0.029s] (current)
  Building tket-qsystem v0.22.0 (baseline)
     Built [  42.252s] (baseline)
   Parsing tket-qsystem v0.22.0 (baseline)
    Parsed [   0.028s] (baseline)
  Checking tket-qsystem v0.22.0 -> v0.22.0 (assume minor change)
   Checked [   0.046s] 159 checks: 159 pass, 41 skip
   Summary no semver update required
  Finished [  86.199s] tket-qsystem

mark-koch

Nice! Not an expert on your pytket conversion work, but looks good from the surface 👍

Mostly just some doc nits and questions. Also I'm wondering why this doesn't count as a breaking change due to signature changes/the switch to OpaqueSubgraph?

tket/src/serialize/pytket/opaque/subgraph.rs

mark-koch · 2025-11-06T15:46:04Z

tket/src/serialize/pytket/opaque/subgraph.rs

+/// of subgraphs it can represent; in particular const edges and order edge are
+/// not allowed in the boundary.
+#[derive(Debug, Clone)]
+pub struct OpaqueSubgraph<N> {


Should we consider upstreaming this? In general, SiblingSubgraph seems quite brittle/not production ready 😅

Yeah, we could have a discussion on which bits of this are useful.

This structure is not specific to the encoder/decoder per-se, but some it's not always useful esp. if the plan is to do rewriting.

There are some plans to redefine the subgraph structures anyways.

tket/src/serialize/pytket/opaque/subgraph.rs

mark-koch · 2025-11-06T15:51:00Z

tket/src/serialize/pytket/encoder.rs

+                // Ignore nodes with order edges, as they cannot be represented in the pytket circuit.
+                if !self.has_order_edges(node, optype, circ) {


Coming from Guppy, this also includes ancilla allocation and measurement :/

Those would have appeared as input/output qubits to pytket anyways.

We should thing about alternatives in the future, but at least this change makes things correct.
(We'd need to device some way to track relations between pytket commands, which seems hard)

mark-koch · 2025-11-06T16:28:15Z

tket/src/serialize/pytket/config/type_translators.rs

+            return true;
+        }
+
+        // For now, we just hard-code this to the two kind of bits we support.


This only works if the sum-bool is used linearly, right? Is this a problem?

Or do we assume that pytket extraction only runs after bool linearisation?

Sum bool is marked as unsupported, so we won't see them here.

tket/src/serialize/pytket/tests.rs

mark-koch · 2025-11-06T16:49:10Z

tket/src/serialize/pytket/tests.rs

 // TODO: fix edge case: non-local edge from an unsupported node inside a nested CircBox
 // to/from the input of the head region being encoded...
-#[should_panic(expected = "Unsupported edge kind")]
-#[case::non_local(circ_non_local(), 1, CircuitRoundtripTestConfig::Default)]
+#[should_panic(expected = "has an unconnected port")]
+#[case::non_local(circ_non_local(), 2, CircuitRoundtripTestConfig::Default)]


Is this still the same edge case? I don't understand why the message is now "has an unconnected port"?

Thinking about it I'm not sure why it's still failing after some of these fixes 🤔
Let me double check :P

It was a problem with non-local edges from input nodes... It's fixed now :)

…meter

aborgna-q added 6 commits November 5, 2025 17:57

fix: Don't try to encode wires with no values

d576fd2

feat: Replace SiblingSubgraph with simpler OpaqueSubgraph

035c76b

Don't encode nodes with order edges

fb5278d

fix: Internal-only bits not appearing in pytket circuit's bit registe…

5de8616

…r list

feat: Initial piping for translating isomorphic types

56d2be5

fix: Allocate qbs declared by pytket only if consumed

431b8e6

aborgna-q requested a review from mark-koch November 6, 2025 14:57

aborgna-q marked this pull request as ready for review November 6, 2025 14:57

aborgna-q requested a review from a team as a code owner November 6, 2025 14:57

aborgna-q requested review from ss2165 and removed request for ss2165 November 6, 2025 14:57

aborgna-q added 2 commits November 6, 2025 15:46

fix: Make sure output params have the right type

85ccdee

Merge remote-tracking branch 'origin/main' into ab/fixes

dd6f1c3

mark-koch approved these changes Nov 6, 2025

View reviewed changes

cqc-alec approved these changes Nov 6, 2025

View reviewed changes

aborgna-q added 4 commits November 6, 2025 17:44

fix: Don't try to encode wires with complex types that include a para…

a585069

…meter

fix: encode input parameters to extra_subgraphs

2475a90

Review comments

118b86f

fix: Move non-local edges originating from decoded region's input

9600a72

aborgna-q added this pull request to the merge queue Nov 6, 2025

Merged via the queue into main with commit e0771f8 Nov 6, 2025
23 of 24 checks passed

aborgna-q deleted the ab/fixes branch November 6, 2025 19:04

hugrbot mentioned this pull request Nov 6, 2025

chore: release #1189

Draft

		// Ignore nodes with order edges, as they cannot be represented in the pytket circuit.
		if !self.has_order_edges(node, optype, circ) {

fix: Multiple fixes to the pytket encoder #1226

fix: Multiple fixes to the pytket encoder #1226

Uh oh!

Conversation

aborgna-q commented Nov 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

codecov bot commented Nov 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

hugrbot commented Nov 6, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mark-koch left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

aborgna-q commented Nov 6, 2025 •

edited

Loading

codecov bot commented Nov 6, 2025 •

edited

Loading

hugrbot commented Nov 6, 2025 •

edited

Loading