sicore 2.0.3

Core Package for Selective Inference

sicore package

This package provides core functions for selective inference.

Detailed API reference is here.

Installation

This package requires python 3.10 or higher and automatically installs any dependent packages. If you want to use tensorflow and pytorch's tensors, please install them manually.

$ pip install sicore

Uninstall :

$ pip uninstall sicore

Module Contents

The following modules can be imported by from sicore import *.

Selective Inference

• SelectiveInferenceNorm : Selective inference for the normal distribution.
• SelectiveInferenceChi : Selective inference for the chi distribution.
• SelectiveInferenceResult: Data class for the result of selective inference.

Evaluation

• rejection_rate(): Computes rejection rate from the list of SelectiveInferenceResult objects or p-values.

Figure

• pvalues_hist() : Draws a histogram of p-values.
• pvalues_qqplot() : Draws a uniform Q-Q plot of p-values.
• FprFigure: Draws a figure of the false positive rate.
• TprFigure: Draws a figure of the true positive rate.

Interval Operations

• RealSubset : Class for representing a subset of real numbers, which provides many operations with intuitive syntax.
• complement() : Take the complement of intervals.
• union() : Take the union of two intervals.
• intersection() : Take the intersection of two intervals.
• difference() : Take the difference of first intervals with second intervals.
• symmetric_difference() : Take the symmetric difference of two intervals.

Inequalities Solver

• polynomial_below_zero() : Compute intervals where a given polynomial is below zero.
• polytope_below_zero() : Compute intervals where a given polytope is below zero.
• degree_one_polynomials_below_zero: Compute intervals where given degree-one polynomials are all below zero.

Non-Gaussian Random Variables

• generate_non_gaussian_rv(): Generate a standardized random variable in a given rv_name family with a given Wasserstein distance from the standard gaussian distribution.

Constructor

• OneVector : Vector whose elements at specified positions are set to 1, and 0 otherwise.
• construct_projection_matrix() : Construct projection matrix from basis.

Others

Execute code test :

$ pytest tests/