Horton

Re: [PATCH] random_normal function

Paul Ramsey <pramsey@cleverelephant.ca>

View thread
From: Paul Ramsey <pramsey@cleverelephant.ca>
To: "David G. Johnston" <david.g.johnston@gmail.com>
Cc: Pgsql Hackers <pgsql-hackers@postgresql.org>
Date: 2022-12-08T22:57:34Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Round off random_normal() test results one more decimal place.

  2. Remove pg_regress' never-documented "ignore" feature.

  3. Upgrade the random.sql regression test.

  4. Invent random_normal() to provide normally-distributed random numbers.

Attachments


> On Dec 8, 2022, at 2:40 PM, David G. Johnston <david.g.johnston@gmail.com> wrote:
> 
> On Thu, Dec 8, 2022 at 2:53 PM Paul Ramsey <pramsey@cleverelephant.ca> wrote:
> 
> random_normal(stddev float8 DEFAULT 1.0, mean float8 DEFAULT 0.0)
> 
> Any particular justification for placing stddev before mean?  A brief survey seems to indicate other libraries, as well as (at least for me) learned convention, has the mean be supplied first, then the standard deviation.  The implementation/commentary seems to use that convention as well.

No, I'm not sure what was going through my head, but I'm sure it made sense at the time (maybe something like "people will tend to jimmy with the stddev more frequently than the mean"). I've reversed the order

> Some suggestions:

Thanks! Taken :)

> And a possible micro-optimization...
> 
> + bool   rescale = true
> + if (PG_NARGS() = 0)
> +    rescale = false
> ...
> + if (rescale)
>     ... result = (stddev * z) + mean
> + else
> +      result = z

Feels a little too micro to me (relative to the overall cost of the transform from uniform to normal distribution). I'm going to leave it out unless you violently want it.

Revised patch attached.

Thanks!

P