RNG collision for seeds 181200 and 361380

[martijnr-lapp]

I just ran 26 bootstraps with unique seeds. The results were only 24 unique batches.

An example of a collision was for the seeds 181200 and 361380

It is not the same issue as #108, but similar. As I couldn't reopen that one, I created a new issue.

[rikardn]

I cannot reproduce this. Which version of PsN are you using? Did you use Math::Random or Math::Random::Free ? You can check which one you have installed with

perl -e "use Math::Random::Free"
perl -e "use Math::Random"

Math::Random::Free is now used by default unless the use_old_math_random configuration option is set. This should only be used when attempting to reproduce old results.

[martijnr-lapp]

Thanks for your reply.

We use PsN 5.5.0, Perl 5.38.2 with the Math::Random::Free library and NONMEM 7.6.0, ran via a Docker container on Ubuntu 24.04.

[rikardn]

I guess collisions would be more probable with few bootstrap samples and few individuals in the dataset. Is this the case for you?

[martijnr-lapp]

The dataset contains several thousands of subjects so I guess we can exclude that. Next to that, on the sample level results are also identical.