Cqt 455 implement analyzer pass in open squirrel by SoufiTNO · Pull Request #682 · QuTech-Delft/OpenSquirrel

SoufiTNO · 2026-05-05T12:33:33Z

Adds a new ''Analyzer'' pass type and the first concrete implementation, "'CircuitAnalyzer'', which computes structural metrics describing a quantum circuit. Metrics are organized into four categories :

Size: number of qubits, gates, two-qubit gates, two-qubit gate percentage, depth
Interaction graph (IG): average shortest path, std of adjacency, diameter, central dominance, average degree, number of maximal cliques, clustering coefficient
Gate dependency graph (GDG): critical path length, path length mean/std, fraction of gates on critical path
Density: density score, idling score

- Implement Analyzer ABC in opensquirrel/passes/analyzer/general_analyzer.py - Implement CircuitAnalyzer with size, IG, GDG, and density metrics - Update Circuit.analyze() return type to dict[str, Any] - Add 14 tests in tests/passes/analyzer/ - Update CHANGELOG

rares1609

Very cool pass! Its definitely helpful to have these metrics easily accessible.

I've only left a couple of minor comments, not approving/requesting changes specifically. Nice work!

rares1609 · 2026-05-18T09:13:39Z

 - `CNOT2CZDecomposer` decomposer pass
 - `RoutingChecker` routing pass
 - Restore SGMQ notation for barrier groups in cQASMv1 Exporter
+- `CircuitAnalyzer` analyzer pass for computing structural circuit metrics (size, interaction graph, gate dependency graph, density)


This would need to be added to the #Added section at line 15 in the file.

rares1609 · 2026-05-18T09:17:10Z

+def test_size_metrics_on_empty_circuit(analyzer: CircuitAnalyzer, empty_circuit: Circuit) -> None:
+    result = analyzer.analyze(empty_circuit)
+    assert result["n_qubits"] == 3
+    assert result["n_gates"] == 0
+    assert result["n_two_qubit_gates"] == 0
+    assert result["two_qubit_pct"] == pytest.approx(0.0, abs=1e-9)
+    assert result["depth"] == 0
+
+
+def test_size_metrics_on_ghz(analyzer: CircuitAnalyzer, ghz_circuit: Circuit) -> None:
+    result = analyzer.analyze(ghz_circuit)
+    assert result["n_qubits"] == 3
+    assert result["n_gates"] == 3
+    assert result["n_two_qubit_gates"] == 2
+    assert result["two_qubit_pct"] == pytest.approx(2 / 3, abs=1e-3)
+    assert result["depth"] == 3


Maybe you could group these two tests with using @pytest.mark.parametrize?

rares1609 · 2026-05-18T09:17:38Z

+def test_depth_with_parallel_gates(analyzer: CircuitAnalyzer, parallel_circuit: Circuit) -> None:
+    """Two independent gates can run in the same time-step, so depth is 1."""
+    result = analyzer.analyze(parallel_circuit)
+    assert result["depth"] == 1
+    assert result["n_gates"] == 2
+
+
+def test_depth_with_sequential_gates(analyzer: CircuitAnalyzer, sequential_circuit: Circuit) -> None:
+    """Gates that share qubits must serialise, so depth equals gate count."""
+    result = analyzer.analyze(sequential_circuit)
+    assert result["depth"] == 3
+    assert result["n_gates"] == 3


Same comment as above

SoufiTNO · 2026-05-18T12:23:13Z

"Hey Chris, thanks for the tip! I ve grouped the size, depth, IG, and GDG tests using @pytest.mark.parametrize. I also modified the CHANGELOG.md, and made sure all the linter checks are green. Thanks for the review !"

elenbaasc · 2026-05-19T11:57:36Z

"Hey Chris, thanks for the tip! I ve grouped the size, depth, IG, and GDG tests using @pytest.mark.parametrize. I also modified the CHANGELOG.md, and made sure all the linter checks are green. Thanks for the review !"

No thanks to me! It was Rares (@rares1609) who did the review 😄! Thanks for making the changes

SoufiTNO · 2026-05-19T12:28:05Z

Oops , my bad! Massive thanks to Rares for the review 🙌

elenbaasc

Nice work!

Just a few minor comments, nothing too serious :)

elenbaasc · 2026-05-19T12:04:15Z

+
+    * Size: number of qubits, gates, two-qubit gates, two-qubit gate percentage, depth.
+    * Interaction graph (IG): metrics derived from the qubit interaction graph,
+      where nodes are qubits and edges are two-qubit gates.


Suggested change

where nodes are qubits and edges are two-qubit gates.

where nodes correspond to qubits and edges correspond to two-qubit gates.

elenbaasc · 2026-05-19T12:11:59Z

+        for gate in gate_statements:
+            qubit_indices = list(gate.qubit_indices)


Suggested change

for gate in gate_statements:

qubit_indices = list(gate.qubit_indices)

for gate_statement in gate_statements:

qubit_indices = list(gate_statement.qubit_indices)

elenbaasc · 2026-05-19T12:13:57Z

+            if not qubit_indices:
+                continue


This would be strange. All gates should have qubit indices...

Suggested change

if not qubit_indices:

continue

elenbaasc · 2026-05-19T12:14:23Z

+            new_layer = max(layer[i] for i in qubit_indices) + 1
+            for i in qubit_indices:
+                layer[i] = new_layer


Suggested change

new_layer = max(layer[i] for i in qubit_indices) + 1

for i in qubit_indices:

layer[i] = new_layer

new_layer = max(layer[qubit_index] for qubit_index in qubit_indices) + 1

for qubit_index in qubit_indices:

layer[qubit_index] = new_layer

elenbaasc · 2026-05-19T12:37:13Z

+        empty: dict[str, Any] = {
+            "ig_avg_shortest_path": 0.0,
+            "ig_std_adjacency": 0.0,
+            "ig_diameter": 0,
+            "ig_central_dominance": 0.0,
+            "ig_avg_degree": 0.0,
+            "ig_n_maximal_cliques": 0,
+            "ig_clustering_coefficient": 0.0,
+        }
+
+        weighted_edges = circuit.interaction_graph
+        if not weighted_edges:
+            return empty


Suggested change

empty: dict[str, Any] = {

"ig_avg_shortest_path": 0.0,

"ig_std_adjacency": 0.0,

"ig_diameter": 0,

"ig_central_dominance": 0.0,

"ig_avg_degree": 0.0,

"ig_n_maximal_cliques": 0,

"ig_clustering_coefficient": 0.0,

}

weighted_edges = circuit.interaction_graph

if not weighted_edges:

return empty

weighted_edges = circuit.interaction_graph

if not weighted_edges:

return {

"ig_avg_shortest_path": 0.0,

"ig_std_adjacency": 0.0,

"ig_diameter": 0,

"ig_central_dominance": 0.0,

"ig_avg_degree": 0.0,

"ig_n_maximal_cliques": 0,

"ig_clustering_coefficient": 0.0,

}

elenbaasc · 2026-05-19T12:49:15Z

+    @staticmethod
+    def _build_gate_dependency_graph(gate_statements: list[Gate]) -> nx.DiGraph:
+        """Build a DAG where edge (i, j) means gate i must run before gate j."""
+        gdg: nx.DiGraph = nx.DiGraph()


I prefer to write this out (also in other places).

Suggested change

gdg: nx.DiGraph = nx.DiGraph()

gate_dependency_graph: nx.DiGraph = nx.DiGraph()

elenbaasc · 2026-05-19T12:51:42Z

+                if longest_from[successor] + 1 > longest_from[node]:
+                    longest_from[node] = longest_from[successor] + 1
+
+        return longest_to, longest_from


Why would you want them in this order (convention?)? Seems more intuitive to go from 'from' to 'to', and not from 'to' to 'from' 😅.

elenbaasc · 2026-05-19T12:53:44Z

+        mean_length = sum(path_lengths) / len(path_lengths)
+        variance = sum((x - mean_length) ** 2 for x in path_lengths) / len(path_lengths)


Surely there are built-in ways to determine the mean and variance of the path_lengths, right?

elenbaasc · 2026-05-19T13:26:53Z

+            for gate in gate_statements:
+                for qubit_index in gate.qubit_indices:


Suggested change

for gate in gate_statements:

for qubit_index in gate.qubit_indices:

for gate_statement in gate_statements:

for qubit_index in gate_statement.qubit_indices:

elenbaasc · 2026-05-19T13:28:09Z

+from opensquirrel import CircuitBuilder
+from opensquirrel.circuit import Circuit


Suggested change

from opensquirrel import CircuitBuilder

from opensquirrel.circuit import Circuit

from opensquirrel import CircuitBuilder, Circuit

…p the logic

SoufiTNO · 2026-05-20T12:18:00Z

Hey Chris , thank you for the review !
I ve implemented all of the requested changes. Please let me know if there are any other adjustments required before we merge it.

elenbaasc and others added 2 commits April 30, 2026 10:29

Setup implementation analyzer pass.

deb10d7

SoufiTNO requested a review from elenbaasc May 5, 2026 12:33

rares1609 reviewed May 18, 2026

View reviewed changes

SoufiTNO added 3 commits May 18, 2026 13:49

Refactor analyzer tests with parametrize and update changelog

667c410

Fix

cce1c1a

Fix pytest parametrize syntax

2ee1fca

elenbaasc reviewed May 19, 2026

View reviewed changes

I refactored CircuitAnalyzer to use instance attributes and cleaned u…

4078ae5

…p the logic

	where nodes are qubits and edges are two-qubit gates.
	where nodes correspond to qubits and edges correspond to two-qubit gates.

		for gate in gate_statements:
		qubit_indices = list(gate.qubit_indices)

	gdg: nx.DiGraph = nx.DiGraph()
	gate_dependency_graph: nx.DiGraph = nx.DiGraph()

		mean_length = sum(path_lengths) / len(path_lengths)
		variance = sum((x - mean_length) ** 2 for x in path_lengths) / len(path_lengths)

		for gate in gate_statements:
		for qubit_index in gate.qubit_indices:

		from opensquirrel import CircuitBuilder
		from opensquirrel.circuit import Circuit

	from opensquirrel import CircuitBuilder
	from opensquirrel.circuit import Circuit
	from opensquirrel import CircuitBuilder, Circuit

Conversation

SoufiTNO commented May 5, 2026

Uh oh!

rares1609 left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

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!

SoufiTNO commented May 18, 2026

Uh oh!

elenbaasc commented May 19, 2026

Uh oh!

SoufiTNO commented May 19, 2026

Uh oh!

elenbaasc left a comment

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!

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!

Choose a reason for hiding this comment

Uh oh!

SoufiTNO commented May 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

rares1609 left a comment •

edited

Loading