Sinter

1.15.0 verified Fri May 01 auth: no python

Sinter is a Python library for sampling and decoding quantum error correction circuits using Stim, with integrated support for PyMatching and various decoders. Current version 1.15.0; released roughly monthly. It provides CLI tools and a Python API for collecting statistics, plotting error curves, and managing batch jobs.

pip install sinter

Common errors

error ModuleNotFoundError: No module named 'sinter' ↓

cause Sinter is not installed or installed in a different environment.

fix

Run pip install sinter in your active Python environment.

error AttributeError: module 'sinter' has no attribute 'collect' ↓

cause Outdated version of sinter (<0.1). The API changed significantly in earlier versions.

fix

Upgrade to latest version: pip install --upgrade sinter.

error RuntimeError: Failed to decode: decoder 'pymatching' not available ↓

cause pymatching library is not installed.

fix

Install pymatching: pip install pymatching.

Warnings

breaking In sinter v1.10+, the `sinter.plot.error_rate_vs_error_probability` function changed its signature: the `collection` argument is now positional-only and `show` parameter is removed. ↓

fix Use `sinter.plot.error_rate_vs_error_probability(sample, ax=ax)` instead of passing collection as keyword or using `show=True`.

deprecated `sinter.collect()` parameter `num_workers` is deprecated in favor of `num_processes`. ↓

fix Replace `num_workers=4` with `num_processes=4`.

gotcha If you don't install pymatching, sinter will raise a runtime error when trying to use 'pymatching' decoder. It is not installed by default. ↓

fix Install pymatching via `pip install pymatching` or use a different decoder like 'fusion_blossom' that doesn't require pymatching.

Imports

sinter
```
import sinter
```
sinter.Collection
```
from sinter import Collection
```
sinter.Task
```
from sinter import Task
```
sinter.Stats
```
from sinter import Stats
```

Quickstart

Create a repetition code circuit, sample it with PyMatching decoder, and plot error rate vs physical error probability.

import sinter
import stim

tasks = [
    sinter.Task(
        circuit=stim.Circuit.generated(
            'repetition_code:memory',
            rounds=100,
            distance=5
        ),
        decoder='pymatching',
    )
]

sample = sinter.collect(
    num_workers=4,
    max_errors=1_000,
    start_batch_size=100,
    tasks=tasks,
)

print(sample)

# Plot results
fig, ax = sinter.plot.error_rate_vs_error_probability(sample)
fig.savefig('plot.png')