Merge pull request #124 from CQCL/3q-rz-bug

Exec order of sub commands bug fixes

.pre-commit-config.yaml

@@ -13,7 +13,7 @@ repos:
       - id: debug-statements
 
   - repo: https://github.com/crate-ci/typos
-    rev: v1.18.0
+    rev: v1.18.1
     hooks:
       - id: typos
 
@@ -25,7 +25,7 @@ repos:
         - black==23.10.1
 
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.2.0
+    rev: v0.2.1
     hooks:
       - id: ruff
         args: [--fix, --exit-non-zero-on-fix]

pytket/phir/phirgen_parallel.py

@@ -30,23 +30,21 @@
 logger = logging.getLogger(__name__)
 
 
-def exec_order_preserved_helper(
-    ordered_dict: OrderedDict["UnitID", int], group_number: int, qubit_last_group: int
+def exec_order_preserved(
+    group_exec_order: list[int], group_number: int, qubit_last_group: int
 ) -> bool:
-    """A helper to determine whether order is preserved when adding qubits to groups."""
-    # determine whether the current group number is later in execution
-    # than the last group in which a qubit was used
-    group_eligible = group_number > qubit_last_group
-    if not group_eligible:
+    """Determine whether order is preserved when adding qubits to groups."""
+    if (group_number not in group_exec_order) or (
+        qubit_last_group not in group_exec_order
+    ):
+        return True
+    if group_number <= qubit_last_group:
         return False
-    for group in ordered_dict.values():
-        if group == qubit_last_group:
-            order_preserved = False
-            break
-        if group == group_number:
-            order_preserved = True
-            break
-    return order_preserved
+    # if the group that the qubit is eligible for is later in the exec_order list than
+    # the last group in which it was used, it can be parallelized
+    return group_exec_order.index(group_number) > group_exec_order.index(
+        qubit_last_group
+    )
 
 
 def process_sub_commands(
@@ -61,28 +59,29 @@ def process_sub_commands(
     # and different gate types don't go in the same group
     # RZ gates go in (mod 3)=0 groups, R1XY gates go in (mod 3)=1 groups,
     # and all other gates will go in (mod 3)=2 groups
-    rz_group_number = -3  # will be set to 0 when first RZ gate is assigned (-3 + 3 = 0)
-    r1xy_group_number = (
-        -2  # will be set to 1 when first R1XY gate is assigned (-2 + 3 = 1)
-    )
-    other_group_number = (
-        -1  # will be set to 2 when first other gate is assigned (-1 + 3 = 2)
-    )
-    num_scs_per_qubit = {}
-
-    for qubit in sub_commands:
-        num_scs_per_qubit[qubit] = len(sub_commands[qubit])
-        # set every qubit's group id to be -4
-        # prevents KeyError in check for group number
-        # will get set to a valid group number the first time the qubit is used
-        qubits2groups[qubit] = -4
+    rz_group_number = -3  # set to 0 when first RZ gate is assigned (-3 + 3 = 0)
+    r1xy_group_number = -2  # set to 1 when first R1XY gate is assigned (-2 + 3 = 1)
+    other_group_number = -1  # set to 2 when first other gate is assigned (-1 + 3 = 2)
+    num_scs_per_qubit: dict["UnitID", int] = {}
+    group_exec_order: list[int] = []
+
+    for qubit, cmds in sub_commands.items():
+        num_scs_per_qubit[qubit] = len(cmds)
+
     max_len = max(num_scs_per_qubit.values())
 
     for index in range(max_len):
-        for qubit in sub_commands:
+        for qubit, cmds in sub_commands.items():
+            # make sure the qubits are inserted into
+            # qubits2groups in the order in which they appear
+            if qubit not in qubits2groups:
+                # set every qubit's group id to be a default value -4
+                # prevents KeyError in check for group number
+                # will get set to a valid group number the first time the qubit is used
+                qubits2groups[qubit] = -4
             # check to make sure you are not accessing beyond the end of the list
             if index < num_scs_per_qubit[qubit]:
-                sc = sub_commands[qubit][index]
+                sc = cmds[index]
                 gate = sc.op.type
                 match gate:
                     case tk.OpType.Rz:
@@ -96,18 +95,15 @@ def process_sub_commands(
                         valid_pll_op = False
                 # does a group exist for that gate type?
                 group_available = group_number in groups
-                # is that group later in execution than the
-                # most recent group for an op on that qubit?
-                order_preserved = exec_order_preserved_helper(
-                    qubits2groups, group_number, qubits2groups[qubit]
-                )
                 # is the group size still under the maximum allowed parallel ops?
                 group_size = len(groups[group_number]) if group_number in groups else 0
                 group_not_too_large = group_size < max_parallel_sq_gates
                 # is the op parallelizable (only RZ or R1XY)?
                 if (
                     group_available
-                    and order_preserved
+                    and exec_order_preserved(
+                        group_exec_order, group_number, qubits2groups[qubit]
+                    )
                     and group_not_too_large
                     and valid_pll_op
                 ):
@@ -126,6 +122,7 @@ def process_sub_commands(
                             other_group_number += 3
                             group_number = other_group_number
                     groups[group_number] = [sc]
+                    group_exec_order.append(group_number)
                     # prevent the group number from ever decrementing
                     if group_number > qubits2groups[qubit]:
                         qubits2groups[qubit] = group_number

requirements.txt

@@ -6,7 +6,7 @@ pre-commit==3.6.0
 pydata_sphinx_theme==0.15.2
 pytest==8.0.0
 pytket==1.24.0
-ruff==0.2.0
+ruff==0.2.1
 setuptools_scm==8.0.4
 sphinx==7.2.6
 wasmtime==17.0.0

tests/data/qasm/rz_exec_order_three_qubits.qasm

@@ -0,0 +1,22 @@
+OPENQASM 2.0;
+include "qelib1.inc";
+include "hqslib1_dev.inc";
+
+qreg q[3];
+creg c[3];
+
+ZZ q[1],q[2];
+
+ry(0.5*pi) q[0];
+ry(0.5*pi) q[2];
+
+barrier q[0],q[1],q[2];
+
+ry(3.5*pi) q[0];
+ry(0.5*pi) q[2];
+
+
+barrier q[0],q[1],q[2];
+measure q[0] -> c[0];
+measure q[1] -> c[1];
+measure q[2] -> c[2];

tests/test_parallelization.py

@@ -110,6 +110,29 @@ def test_single_qubit_circuit_with_parallel() -> None:
         )
 
 
+def test_three_qubit_rz_exec_order_preserved() -> None:
+    """Test that the order of gating is preserved in a 3 qubit circuit with RZ gates."""
+    phir = get_phir_json(QasmFile.rz_exec_order_three_qubits, rebase=True)
+    # verify that the parallel RZ gates are executed before the second R1XY gate
+    assert phir["ops"][8] == {
+        "qop": "R1XY",
+        "angles": [[0.5, 0.5], "pi"],
+        "args": [["q", 0]],
+    }
+    assert phir["ops"][10] == {
+        "block": "qparallel",
+        "ops": [
+            {"qop": "RZ", "angles": [[0.5], "pi"], "args": [["q", 1]]},
+            {"qop": "RZ", "angles": [[1.5], "pi"], "args": [["q", 2]]},
+        ],
+    }
+    assert phir["ops"][12] == {
+        "qop": "R1XY",
+        "angles": [[0.5, 0.5], "pi"],
+        "args": [["q", 2]],
+    }
+
+
 def test_two_qubit_exec_order_preserved() -> None:
     """Test that the order of gating in preserved in a 2 qubit circuit."""
     phir = get_phir_json(QasmFile.exec_order_two_qubits, rebase=True)

tests/test_utils.py

@@ -47,6 +47,7 @@ class QasmFile(Enum):
     classical_ordering = auto()
     single_qubit_parallel_test = auto()
     exec_order_two_qubits = auto()
+    rz_exec_order_three_qubits = auto()
     cond_barrier = auto()
     arbitrary_qreg_names = auto()