Dear All,

I am running a computationally intensive simulation on Stata-16 MP-32 core on a high-powered computing cluster. I am finding that with each rep the program is generating the exact same data when I set the observations N=10,000, but not when I set the number of obs to 1,000. However, having a larger dataset is useful for the approach I am studying.

I replicated the problem on my desktop with Stata-14 MP 4 core with obs=10,000 and matsize=10,000, but could solve the problem by setting the matsize = 11,000. With Stata 16 it doesn't seem possible to set the matsize, as matsize should automatically be the max (over 65,000 with this flavor).

As far as resources go on the HPCC, I am using 8 cpus and memory of 100 GB (more than my desktop has).

Does anyone have a suggestion of how I might be able to solve this issue?

Thanks!

Michael