lomatching 0.2.0

Logical observable decoder for transversal logical gates in surface codes based on pymatching.

lomatching

Decoder for (fold-)transversal logical gates in surface codes based on MWPM.

Installation

This package is available in PyPI, thus it can be installed using

pip install lomatching

or alternatively, it can be installed from source using

git clone [e-mail]:MarcSerraPeralta/lomatching.git
pip install lomatching/

Example

import numpy as np
import stim

from surface_sim.setups import CircuitNoiseSetup
from surface_sim.models import CircuitNoiseModel
from surface_sim import Detectors
from surface_sim.experiments import schedule_from_circuit, experiment_from_schedule
from surface_sim.circuit_blocks.unrot_surface_code_css import gate_to_iterator
from surface_sim.layouts import unrot_surface_codes

from lomatching import MoMatching, get_reliable_observables, remove_obs_except

# circuit considered
circuit = stim.Circuit(
    """
    RX 0
    RZ 1
    TICK
    CNOT 0 1
    MX 0 1
    """
)

# generate encoded circuit
layouts = unrot_surface_codes(2, distance=3)
setup = CircuitNoiseSetup()
setup.set_var_param("prob", 1e-3)
model = CircuitNoiseModel.from_layouts(setup, *layouts)
detectors = Detectors.from_layouts("pre-gate", *layouts)

schedule = schedule_from_circuit(circuit, layouts, gate_to_iterator)
encoded_circuit = experiment_from_schedule(
    schedule, model, detectors, anc_reset=True, anc_detectors=None
)

# prepare inputs for MoMatching
stab_coords = [{} for _ in layouts]
for l, layout in enumerate(layouts):
    coords = layout.anc_coords
    stab_coords[l][f"Z"] = [v for k, v in coords.items() if k[0] == "Z"]
    stab_coords[l][f"X"] = [v for k, v in coords.items() if k[0] == "X"]

reliable_obs = get_reliable_observables(encoded_circuit)
encoded_circuit = remove_obs_except(encoded_circuit, reliable_obs)

decoder = MoMatching(encoded_circuit, stab_coords)

# run MoMatching
sampler = encoded_circuit.detector_error_model().compile_sampler()
syndromes, log_flips, _ = sampler.sample(shots=10)

predictions = decoder.decode_batch(syndromes)
log_errors = (predictions != log_flips).any(axis=1)

print("Logical error probability:", np.average(log_errors))

How do I cite lomatching?

When using lomatching for research, please cite:

@misc{serraperalta2025decoding,
      title={Decoding across transversal Clifford gates in the surface code}, 
      author={Marc Serra-Peralta and Mackenzie H. Shaw and Barbara M. Terhal},
      year={2025},
      eprint={2505.13599},
      archivePrefix={arXiv},
      primaryClass={quant-ph},
      url={https://arxiv.org/abs/2505.13599}, 
}