Overview

MENOS is a quantum circuit simulation compiler and runtime. It takes a circuit, works out how it could be executed, estimates the cost of each option, and runs the cheapest one that fits your constraints.

What it is

• A simulator. It produces measurement counts, exact statevectors, and exact expectation values of Pauli observables.
• A compiler. A circuit is lowered to an internal representation (KIR), validated, and reduced to the part that can influence the output you asked for (light-cone reduction).
• A planner. A cost model prices every feasible execution plan in a single shared unit, and the cheapest plan wins.
• A runtime. The chosen plan is executed, optionally monitored against a time budget and replanned when a run overruns.
• A multi-engine system. Eight execution strategies sit behind one interface: dense, stabilizer, Pauli propagation, matrix product state, tensor network, cut and recombine, density matrix, and trajectory. The dense engine also runs on an NVIDIA GPU.
• An exact noisy simulator. Qiskit Aer noise models are simulated exactly by the density-matrix engine, or, when you opt in, sampled by trajectories with a reported standard error. See Noise Models.
• A Qiskit integration. MENOSBackend is a Qiskit BackendV2; MENOSEstimator is a BaseEstimatorV2.

What it is not

Being precise about scope is part of the design, so this list is as important as the one above.

• It is not a large-scale noisy simulator by default. Noise is simulated exactly with a density matrix, which limits it to about 14 qubits on 16 GiB of RAM. Trajectory sampling reaches statevector sizes, but only when you opt in, and its results are statistical. Only verified noise types are accepted. See Noise Models.
• It does not support mid-circuit measurement, reset, or classical control. Measurements must be terminal. See Unsupported Circuits.
• It provides no SamplerV2 primitive. Counts come from MENOSBackend.run.
• It is not approximate unless you ask for approximation. The only approximation the Python API can ask for is trajectory sampling of noise, which is flagged. See Exactness.
• It is not a general quantum SDK. It has no circuit-construction library of its own; you build circuits with Qiskit (or the small native menos.Circuit API) and MENOS executes them.

A minimal tour

from qiskit import QuantumCircuit
from menos.qiskit import MENOSBackend

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

counts = MENOSBackend().run(qc, shots=1000, seed_simulator=7).result().get_counts()
assert set(counts) <= {"00", "11"}

The result is an ordinary Qiskit Result. To see how MENOS ran it, use the native API, where every result carries a human-readable plan report:

from menos import Circuit, Opcode, RunOptions, run

circuit = Circuit(2, 2)
circuit.append(Opcode.H, [0])
circuit.append(Opcode.CX, [0, 1])
circuit.append_measure(0, 0)
circuit.append_measure(1, 1)

options = RunOptions()
options.shots = 100
options.seed = 1
print(run(circuit, options).explain())

The report lists every candidate strategy, which won, and why. See Plan Explanation.

Next

• Why MENOS: the design motivation.
• Architecture: how the pieces fit together.
• Installation: get it running.