Re: Sample implementation of make-ball-generator seems incorrect
Bradley Lucier 09 Feb 2024 16:13 UTC
On 2/9/24 12:26 AM, Linas Vepstas wrote:
>
> I lost track. Some questions:
> -- Is the original generator, as originally written, correct?
> -- Were all of the original tests correct?
In my recent changes, I added tests that failed with the old
make-ball-generator, and corrected make-ball-generator, which now passes
the tests.
make-sphere-generator (which, in 2020, you split into two, including
make-ellipsoid-generator) was failing some informal tests I cooked up in
August 2020, which you indeed fixed in 2020. My recent changes added
more fine-grained argument error checking and more statistically
justified versions of my informal 2020 tests that people might be able
to use in independent implementations.
Brad