Circuit Inputs¶

Clifft's primary circuit input format follows Stim but adds gates for non-Clifford operations and additional noise models. Clifft supports nearly all native Stim operations, including Clifford gates, noise, measurements, detectors, observables, and repeat blocks. This is the most capable input path and works with every Clifft workflow.

OpenQASM 2 support is built into clifft. Qiskit and Cirq support comes from separately released companion packages. The input format is independent of the simulation workflow and CPU execution settings.

Starting point	Package and entry point	Important limits
Stim with Clifft extensions	`clifft.compile(text)`	Broadest feature support; see Supported Gates.
OpenQASM 2 text	`clifft.compile(text, input_format="qasm2")`	Unitary circuits only; no measurement, reset, classical control, or custom gate declarations.
Qiskit `QuantumCircuit`	`clifft-qiskit`	Terminal-measurement sampling; rejects mid-circuit measurement, feedforward, reset, and other non-unitary operations.
Cirq `cirq.Circuit`	`clifft-cirq`	Parameter-resolved qubit circuits; no device, timing, calibration, qudit, arbitrary classical-control, or noise-channel semantics.

Stim with Clifft extensions¶

Existing Stim circuits can be passed directly to clifft.compile(). Add Clifft operations such as T when the circuit needs non-Clifford behavior:

import clifft

program = clifft.compile("""
    H 0
    CX 0 1
    T 1
    M 0 1
""")

result = clifft.sample(program, shots=1_000, seed=42)
print(result.measurements[:5])

Use this path for detectors, observables, post-selection, importance sampling, leakage, loss, or other Clifft-specific workflows. Call clifft.parse() or clifft.parse_file() only when the parsed circuit needs to be inspected before compilation.

OpenQASM 2¶

Select OpenQASM explicitly; Clifft does not guess the format from the text:

import clifft

source = """
OPENQASM 2.0;
include "qelib1.inc";
qreg q[2];
h q[0];
t q[1];
cx q[0], q[1];
"""

program = clifft.compile(source, input_format="qasm2")
probabilities = clifft.basis_probabilities(program, ["00", "11"])

The importer supports the standard qelib1.inc unitary vocabulary, register broadcasting, barriers as no-ops, comments, and finite constant angle expressions. It rejects classical registers, measurements, resets, conditions, custom or opaque gates, and nonstandard includes rather than assigning them approximate semantics.

Use clifft.parse_qasm2() or clifft.parse_qasm2_file() when the imported circuit needs to be inspected. OpenQASM support does not require Qiskit.

Qiskit¶

Install clifft-qiskit and run a supported circuit through its Qiskit BackendV2 provider:

pip install clifft-qiskit

from qiskit import QuantumCircuit
from clifft_qiskit import ClifftProvider

qc = QuantumCircuit(2, 2)
qc.h(0)
qc.cx(0, 1)
qc.measure([0, 1], [0, 1])

backend = ClifftProvider().get_backend("clifft")
counts = backend.run(qc, shots=1_000).result().get_counts()
print(counts)

The adapter returns Qiskit-style results for supported terminal-measurement circuits. See the clifft-qiskit repository for its current basis, decomposition behavior, and package-specific limits.

Cirq¶

Install clifft-cirq to convert or sample a supported Cirq circuit:

pip install clifft-cirq

import cirq
import clifft_cirq

q0, q1 = cirq.LineQubit.range(2)
circuit = cirq.Circuit(
    cirq.H(q0),
    cirq.CNOT(q0, q1),
    cirq.measure(q0, q1, key="m"),
)

sampler = clifft_cirq.ClifftSampler(seed=123)
result = sampler.run(circuit, repetitions=1_000)
print(result)

The converter supports common one-, two-, and three-qubit gates that map to Clifft. See the clifft-cirq repository for its current operations, conversion metadata, and package-specific limits.