Re: [PATCH] Fix biased shuffle by avoiding already "struck" elements

[Sergei Trofimovich]

On Wed, 19 Jun 2024 10:55:53 -0400
Dmitry Goncharov <dgoncharov@users.sf.net> wrote:

> On Tue, Jun 18, 2024 at 5:38 PM Sergei Trofimovich <slyich@gmail.com> wrote:
> > After the change probabilities are not as biased:
> >
> >             0     1     2     3
> >         _____ _____ _____ _____
> >     0 | 24.99 24.99 25.01 25.01
> >     1 | 24.99 25.04 24.99 24.99
> >     2 | 25.01 25.00 25.00 24.99
> >     3 | 25.01 24.98 25.00 25.01
> >  
> 
> Looks like you wrote a program to figure out these probabilities.
> Should such a program be included as a part of make test suite?

I wrote a one-off `a.c` printer (attached).

It might be tricky to write safe ranges for probabilities: 4% is
"obviously" biased, but 0.02% is trickier (PRNG / modulo bias,
truncation, or expected deviation).

Glancing at tests/ all the tests exercise user-facing `make` API. What
would be the best way to validate probabilities?

-- 

  Sergei

a.c
Description: Text Data