On May 27, 2015, at 9:31 PM, Daniel Roßberg <danielmrossb...@gmail.com> wrote:
> As an addition to Sean's e-mail: I'm a little bit in favor for option 1. I > implemented once a function similar to option 2 and must learn that using > random numbers can lead to a random result. I.e. two calls of the function > give two different results. Although both are inside the desired accuracy > it's a permanent issue to explain why they are different. > I’m also (slightly) more in favor of option 1 simply because it’s lower risk. However, my previously unqualified “random sampling” wasn't the whole picture. “Quasi-random" for starters, but seeded with error/convergence tracking too. Repeatable results and unbiased sampling should be a requirement upheld with whatever approach is taken. That’s in part why I said it’d probably take a couple weeks. ;) Cheers! Sean
------------------------------------------------------------------------------
_______________________________________________ BRL-CAD Developer mailing list brlcad-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/brlcad-devel
