[racket] Controlling the size of plot symbols

On 04/11/2014 02:04 AM, Konrad Hinsen wrote:
> Neil Toronto writes:
>
>   > On git HEAD this takes about 15-20 seconds on my computer, depending on
>   > random sphere placement and size. It takes 5-6 seconds for 81 samples
>   > instead of 121. For comparison, the following takes 7-10 seconds with
>   > 121 samples and 3-4 seconds with 81 samples.
>
> Thanks, I'll try that approach next week. I wouldn't use a
> sphere-based representation for more than about 100 atoms anyway, and
> I can probably reduce the number of samples even further. 30 samples
> looks good enough for interactive use.
>
> Is there any particular reason for choosing 81 or 121 samples?
> Are those magic numbers?

They're just fencepost counts; i.e. 1 + section count. 81 samples splits 
the isosurface plot into an 80x80x80 grid of cells. It's not necessary, 
more robust, nor even more efficient.

Neil ⊥