microquantum.analysis.inference¶

Inferential statistics for experiment analysis.

HypothesisTest compares two outcome distributions with a chi-square statistic and a permutation p-value (exact, seedable, no special functions required); bootstrap_ci() builds percentile confidence intervals for any statistic via NumPy resampling.

Module Contents¶

class microquantum.analysis.inference.TestOutcome[source]¶

Result of a two-sample comparison.

statistic: float[source]¶

p_value: float[source]¶

significant: bool[source]¶

degrees_of_freedom: int[source]¶

metadata: dict[str, Any] | None = None[source]¶

to_dict()[source]¶

Serialize to a JSON-safe dictionary.

Return type:: dict[str, Any]

class microquantum.analysis.inference.HypothesisTest(alpha=0.05, permutations=1000, seed=None)[source]¶

Two-sample comparison with permutation p-values.

Parameters:

alpha (float) – Significance level.
permutations (int) – Permutation resamples for the p-value.
seed (Optional[int]) – RNG seed for reproducibility.

property alpha: float[source]¶

Significance level.

Return type:: float

compare(counts_a, counts_b)[source]¶

Compare two outcome distributions.

Parameters:

counts_a (collections.abc.Mapping[str, int])
counts_b (collections.abc.Mapping[str, int])

Return type:

microquantum.analysis.inference.float_mean(values)[source]¶

Mean of a sequence (default bootstrap statistic).

Parameters:: values (collections.abc.Sequence[float])
Return type:: float

microquantum.analysis.inference.bootstrap_ci(samples, statistic=float_mean, confidence=0.95, resamples=1000, seed=None)[source]¶

Percentile bootstrap confidence interval for statistic.

Parameters:

samples (collections.abc.Sequence[float])
statistic (collections.abc.Callable[[collections.abc.Sequence[float]], float])
confidence (float)
resamples (int)
seed (Optional[int])

Return type:

tuple[float, float]