Add new multithreaded TwoQubitPeepholeOptimization pass

crates/pyext/src/lib.rs

@@ -68,6 +68,11 @@ fn _accelerate(m: &Bound<PyModule>) -> PyResult<()> {
     add_submodule(m, ::qiskit_transpiler::passes::high_level_synthesis_mod, "high_level_synthesis")?;
     add_submodule(m, ::qiskit_transpiler::passes::remove_diagonal_gates_before_measure_mod, "remove_diagonal_gates_before_measure")?;
     add_submodule(m, ::qiskit_transpiler::passes::remove_identity_equiv_mod, "remove_identity_equiv")?;
+    add_submodule(
+        m,
+        ::qiskit_transpiler::passes::two_qubit_peephole_mod,
+        "two_qubit_peephole",
+    )?;
     add_submodule(m, ::qiskit_accelerate::results::results, "results")?;
     add_submodule(m, ::qiskit_transpiler::passes::sabre::sabre, "sabre")?;
     add_submodule(m, ::qiskit_accelerate::sampled_exp_val::sampled_exp_val, "sampled_exp_val")?;

crates/synthesis/src/qsd.rs

@@ -106,7 +106,7 @@ pub fn quantum_shannon_decomposition(
         smallvec![],
         aview2(&qiskit_circuit::gate_matrix::CX_GATE),
         1.0,
-        "U",
+        EulerBasis::U,
         None,
     )?;
     let one_qubit_decomposer = one_qubit_decomposer_basis_set.unwrap_or(&default_1q_basis);

crates/synthesis/src/two_qubit_decompose/basis_decomposer.rs

@@ -10,6 +10,8 @@
 // copyright notice, and modified files need to carry a notice indicating
 // that they have been altered from the originals.
 
+use std::str::FromStr;
+
 use approx::{abs_diff_eq, relative_eq};
 use num_complex::{Complex64, ComplexFloat};
 use num_traits::Zero;
@@ -117,6 +119,12 @@ impl TwoQubitBasisDecomposer {
         __num_basis_gates(self.basis_decomposer.b, self.basis_fidelity, u)
     }
 
+    /// Is the gate super controlled
+    #[inline]
+    pub fn super_controlled(&self) -> bool {
+        self.super_controlled
+    }
+
     fn decomp1_inner(
         &self,
         target: &TwoQubitWeylDecomposition,
@@ -495,7 +503,7 @@ impl TwoQubitBasisDecomposer {
         gate_params: SmallVec<[f64; 3]>,
         gate_matrix: ArrayView2<Complex64>,
         basis_fidelity: f64,
-        euler_basis: &str,
+        euler_basis: EulerBasis,
         pulse_optimize: Option<bool>,
     ) -> PyResult<Self> {
         let ipz: ArrayView2<Complex64> = aview2(&IPZ);
@@ -604,7 +612,7 @@ impl TwoQubitBasisDecomposer {
             gate,
             gate_params,
             basis_fidelity,
-            euler_basis: EulerBasis::__new__(euler_basis)?,
+            euler_basis,
             pulse_optimize,
             basis_decomposer,
             super_controlled,
@@ -781,7 +789,7 @@ impl TwoQubitBasisDecomposer {
             gate_params?,
             gate_matrix.as_array(),
             basis_fidelity,
-            euler_basis,
+            EulerBasis::from_str(euler_basis).map_err(PyValueError::new_err)?,
             pulse_optimize,
         )
     }
@@ -1014,7 +1022,7 @@ pub fn two_qubit_decompose_up_to_diagonal(
         smallvec![],
         aview2(&CX_GATE),
         1.0,
-        "U",
+        EulerBasis::U,
         None,
     )?;
 

crates/synthesis/src/two_qubit_decompose/gate_sequence.rs

@@ -17,8 +17,8 @@ pub(crate) type TwoQubitSequenceVec = Vec<(PackedOperation, SmallVec<[f64; 3]>,
 
 #[derive(Clone, Debug)]
 pub struct TwoQubitGateSequence {
-    pub(crate) gates: TwoQubitSequenceVec,
-    pub(crate) global_phase: f64,
+    pub gates: TwoQubitSequenceVec,
+    pub global_phase: f64,
 }
 
 impl TwoQubitGateSequence {

crates/transpiler/Cargo.toml

@@ -31,6 +31,7 @@ thiserror.workspace = true
 bytemuck.workspace = true
 fixedbitset.workspace = true
 anyhow.workspace = true
+thread_local = "1.1"
 
 [dependencies.uuid]
 workspace = true

crates/transpiler/src/passes/consolidate_blocks.rs

@@ -31,6 +31,7 @@ use qiskit_circuit::interner::Interned;
 use qiskit_circuit::operations::StandardGate;
 use qiskit_circuit::operations::{ArrayType, Operation, Param, UnitaryGate};
 use qiskit_circuit::packed_instruction::PackedOperation;
+use qiskit_synthesis::euler_one_qubit_decomposer::EulerBasis;
 use qiskit_synthesis::linalg::nalgebra_array_view;
 use qiskit_synthesis::matrix::two_qubit::{
     blocks_to_matrix, get_1q_matrix_from_inst, get_2q_matrix_from_inst, get_matrix_from_inst,
@@ -145,7 +146,7 @@ fn get_decomposer_and_basis_gate(
                         SmallVec::default(),
                         get_matrix(&gate),
                         approximation_degree,
-                        "U",
+                        EulerBasis::U,
                         None,
                     )
                     .unwrap_or_else(|_| {
@@ -167,7 +168,7 @@ fn get_decomposer_and_basis_gate(
                 SmallVec::default(),
                 aview2(&CX_GATE),
                 1.0,
-                "U",
+                EulerBasis::U,
                 None,
             )
             .expect("Error while creating TwoQubitBasisDecomposer using a 'cx' gate."),

crates/transpiler/src/passes/high_level_synthesis.rs

@@ -770,7 +770,7 @@ fn extract_definition(op: &PackedOperation, params: &[Param]) -> PyResult<Option
                         SmallVec::new(),
                         aview2(&CX_GATE),
                         1.0,
-                        "U",
+                        EulerBasis::U,
                         None,
                     )?;
                     let two_qubit_sequence =

crates/transpiler/src/passes/mod.rs

@@ -50,6 +50,7 @@ mod schedule_analysis;
 mod split_2q_unitaries;
 mod substitute_pi4_rotations;
 mod synthesize_rz_rotations;
+mod two_qubit_peephole;
 pub mod unitary_synthesis;
 mod unroll_3q_or_more;
 pub mod vf2;
@@ -104,6 +105,7 @@ pub use schedule_analysis::scheduling_mod;
 pub use split_2q_unitaries::{run_split_2q_unitaries, split_2q_unitaries_mod};
 pub use substitute_pi4_rotations::{py_run_substitute_pi4_rotations, substitute_pi4_rotations_mod};
 pub use synthesize_rz_rotations::{py_run_synthesize_rz_rotations, synthesize_rz_rotations_mod};
+pub use two_qubit_peephole::{two_qubit_peephole_mod, two_qubit_unitary_peephole_optimize};
 pub use unitary_synthesis::{
     UnitarySynthesisConfig, UnitarySynthesisState, run_unitary_synthesis, unitary_synthesis_mod,
 };