Error when trying to calculate the number of errors

[golanor]

Following either stim's or pymatching's tutorial on counting the number of errors, I get the following error

ValueError('resize only works on single-segment arrays')

Here is my code:

import stim
import pymatching

# Base stim circuit
base_circuit = """
R 0 1 2 3 4
X_ERROR(0.001) 0 1 2 3 4
TICK
CX 0 1 2 3
DEPOLARIZE2(0.001) 0 1 2 3
DEPOLARIZE1(0.001) 4
TICK
CX 2 1 4 3
DEPOLARIZE1(0.001) 0
DEPOLARIZE2(0.001) 2 1 4 3
TICK
X_ERROR(0.001) 1 3
M 1 3
TICK
DEPOLARIZE1(0.001) 0 2 4
DETECTOR(1,0) rec[-2]
DETECTOR(3,0) rec[-1]

REPEAT 5 {
    R 1 3
    X_ERROR(0.001) 1 3
    DEPOLARIZE1(0.001) 0 2 4
    CX 0 1 2 3
    DEPOLARIZE2(0.001) 0 1 2 3
    DEPOLARIZE1(0.001) 4
    CX 2 1 4 3
    DEPOLARIZE1(0.001) 0
    DEPOLARIZE2(0.001) 2 1 4 3
    X_ERROR(0.001) 1 3
    M 1 3
    DEPOLARIZE1(0.001) 0 2 4
    SHIFT_COORDS(0, 1) # increase t of future detectors
    DETECTOR(1, 0) rec[-2] rec[-4]
    DETECTOR(3, 0) rec[-1] rec[-3]
}

X_ERROR(0.001) 0 2 4
M 0 2 4
SHIFT_COORDS(0, 1)
DETECTOR(1, 0) rec[-2] rec[-3] rec[-5]
DETECTOR(3, 0) rec[-1] rec[-2] rec[-4]
# Choose the measurement(s) that make up the logical observables:
OBSERVABLE_INCLUDE(0) rec[-1]
"""

sc = stim.Circuit(base_circuit)
dem = sc.detector_error_model()
matcher = pymatching.Matching.from_detector_error_model(dem)
sampler = sc.compile_detector_sampler()
syndrome, actual_observables = sampler.sample(shots=500, separate_observables=True)
predicted_observables = matcher.decode_batch(syndrome)  # this line fails
num_errors = np.sum(np.any(predicted_observables != actual_observables, axis=1))

[oscarhiggott]

This is a result of an incompatibility between python and numpy v2 (see issue #102). Currently pymatching is pinned to numpy==1.* until the issue is resolved.