Qevedo

Open-source software for compiling and simulating quantum programs, fast, on your own machine.

pip install qevedo

Packages

qevedo-simulator

qvd, a fast quantum circuit simulator in Rust: state vector, stabilizer, MPS and near-Clifford methods, chosen per circuit.

Rustpip install qevedo-simulator

qevedo-compiler

QCC, a hardware-agnostic quantum circuit compiler. OpenQASM 2 and 3 in and out, optimized for a device profile. Try the decomposition playground.

Pythonpip install qevedo-compiler

openqasm

An OpenQASM 2 and 3 parser and semantic checker for Python.

Pythonpip install openqasm

qevedo

The simulator and the compiler in one install.

Pythonpip install qevedo

Example

from qevedo.compiler import compile_source, emit_qasm
from qevedo.simulator import run

compiled = compile_source(open("circuit.qasm").read())
result = run(emit_qasm(compiled), shots=1000)
print(result.counts, result.backend)

Benchmarks

qevedo-simulator

Same circuits, same machine: an Intel i9-14900K with 24 cores and 62 GB of RAM. Lower is better.

Bar chart: seconds per circuit for qvd, qsim and Qiskit Aer on random, QFT and GHZ circuits of 26 to 30 qubits — **State vector.** 3–4× faster than qsim and 7–12× faster than Qiskit Aer at 26–30 qubits.

Bar chart: seconds for 10,000 shots of rotated surface codes for qvd, Stim and Qiskit Aer — **Clifford circuits.** A distance-101 surface code, 20,401 qubits, samples 10,000 shots in 1.5 s, 8× faster than Stim.

Bar chart: seconds for gates and 1000 shots on MPS circuits for qvd, Qiskit Aer and quimb — **Matrix product states.** 1.6–14× faster than the faster of Qiskit Aer and quimb, with the same truncation.

Bar chart: seconds for 1000 shots of random Clifford+T circuits for qvd and Qiskit Aer's methods — **Clifford+T.** 200 qubits with 20 T gates sample 1000 shots in 0.11 s. Every Qiskit Aer method takes over 10 minutes or does not support the circuit.

Method, raw data and scripts to reproduce every number: qevedo-simulator/benchmarks.