Documentation

A Probability Library

Try it out

With docker installed run the command and browse the notebooks

sh demo_env.sh

Installation

TODO

...

Introduction

The fundamental building block is the RandomVariable

Random Variables can be created from any distribution using the "med" function. "med" is Swedish for with. All random variables has a sampling function and a pdf/pmf function.

import probpy as pp

rv = pp.normal.med(mu=0.0, sigma=1.0)

samples = rv.sample(size=5)
density = rv.p(samples)

print(samples)
print(density)

[ 0.56202855 -0.63491592 -1.11501445  0.10756346  1.40117568]
[0.34065788 0.32611746 0.21425955 0.39664108 0.14948112]

Conditional Distributions

Variables can be created with partial arguments to represent conditional distributions.

import probpy as pp
rv = pp.normal.med(sigma=1.0)

print(rv.sample(0.0, size=5))
print(rv.sample(4.0, size=5))

[ 0.24050695  0.19103947  1.01564618 -0.37190388 -0.04080893]
[5.24490748 3.72506806 3.59844073 4.71898881 2.47418571]

These variables can be used in many functions in this library. Things ranging from estimating parameter posteriors, predictive posteriors, integration, MCMC finding modes and more to come.

check out the notebooks for some usage examples. if you have docker installed just run

sh demo_env.sh