bug-datamash
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Change of random seed


From: Shawn Wagner
Subject: Re: Change of random seed
Date: Fri, 5 Aug 2022 23:45:18 -0700

Heh. Another half-finished project I had was porting random number code from another project that uses a bunch of OS specific ways to get a strong random seed - getentropy() on Linux and OpenBSD, arc4random on other BSDs (Actually using that for all random number generation if available), some Windows API, /dev/urandom, etc.... Looks like the gnulib getrandom does pretty much the same thing, so I can scrap that work.

On Fri, Aug 5, 2022 at 11:23 PM Tim Rice <trice@posteo.net> wrote:
Hey all,

Just a heads up that I've slightly refactored the init_random function into a separate source file and changed the seed to use getrandom(2).

Looking over the getrandom source in gnulib, I believe this should be compatible with platforms other than GNU/Linux. However, if it gives you a problem, please let me know.

Although a small change, this foreshadows a larger body of work which people may find interesting. I intend to implement a `rand` program capable of simulating random numbers drawn from popular probability distributions.

This is intended to be analagous to functions like runif, rnorm and rexp in the R programming language. In future, you will be able to do things like `rand norm --mean=3 --stdev=1.2 10` or `rand exp --rate=5 1` from your shell, where the R equivalents would be `rnorm(10, 3, 1.2)` and `rexp(1,5)`.

So, the first step is to set up the skeleton with a couple of extra source files, and switch to a more standard way of seeding random numbers :)

~ Tim


reply via email to

[Prev in Thread] Current Thread [Next in Thread]