Matthew Fidler

I'm hopeful; Thanks for entertaining this.

@billdenney you could try bioconductor. They don't have the package size requirements and fit well in the CRAN system; I believe that CRAN and bioconductor are both installed in CRAN...

In case you need a list of supported functions for `nm-vars` there is one in the `nonmem2rx` parser syntax: https://github.com/nlmixr2/nonmem2rx/blob/d74a4eef3a9c0211636cd0931f8424a2bd64e195/inst/abbrev.g#L233 I think I captured them all, but am unsure.

No problem Kyle, thought the function names might be helpful (it took me a bit to find from the `NONMEM` manual). As far as the NONMEM translator, it translates directly...

This would reduce the probability of having similar seeds for each core, though with the size of the uint32 and the number of cores people typically have, this still would...

I looked at the examples in the sitmo package and they have all the seeds specified by the user for each core. I don't think they actually do anything for...

As I said, the probability is low. However, the probability with 1000 cores using the sequential solution is 0.

Wouldn't the probabily increase as you make multiple calls to the mvnfast? It generates a seed from the uniform distribution every time it starts the mvnfast number generation. Still be...

Indeed it is better. Now it only depends on how many times you call mvnfast. The only way to completely overcome it is to create your own seed setting routine...

If you are interested in this approach, it wouldn't be too hard to adapt...?