The Binomial distribution does not draw its PDF for alpha = 0

[mbaudin47]

What happened?

The Binomial distribution does not draw its PDF for n = 10 and alpha = 0. The script below generates the exception "The skewness is not defined for the Binomial distribution when p is 0 or 1".

How to reproduce the issue?

import openturns as ot
graph = ot.Binomial(10, 0).drawPDF()

A workaround is to use the Dirac(0) distribution, which draws the correct PDF.

import openturns as ot
graph = ot.Dirac(0).drawPDF()

Version

1.22

Operating System

unknown

Installation media

unknown

Additional Context

This bug was found by @adutfoy while working on quantiles of the Binomial distribution.

[regislebrun]

The exception is triggered by the computation of the default plot range, based on 2 quantiles. It involves the call to Binomial::computeScalarQuantile, which uses the Cornish Fisher expansion to find a reasonable starting point for a sequential search of the quantile value. This expansion is based on higher order moments, not defined for a Dirac distribution. It should be fixed when the Binomial::computeScalarQuantile() is fixed, or at least you should get a different exception ;-)