Could mrgsim_q() have the obsonly option?

[benjaminrich]

It would be nice to be able to use obsonly=TRUE with mrgsim_q().

[kylebaron]

Hi @benjaminrich -

Yeah, I can add that. Can you try qsim() in the meantime? It has that argument. I had been developing those functions in parallel trying to reduce overhead on very small / quick turnaround problems. I think they would be hard to differentiate right now.

Also - would you mind giving me a (ballpark) idea how you're using it and how much benefit you think you're getting? Just a guestimate ... "seems like it's helping" or "can't tell, but using it anyway" etc.

Thanks,
Kyle

[benjaminrich]

I see. I was not aware of qsim() but indeed it looks like it would suit my purpose. Thanks.

As to the increase in speed, I cannot give even a ballpark figure right now unfortunately. But give me a bit of time and I will try to.

[kylebaron]

Ok; no problem. Sometimes I just use it too without regard to any savings; just curious about it.

If you are using v1.0.0 that does include some refactor of the standard mrgsim() simulation path that also closes the gap; I was sort of satisfied that the difference was fairly small and could only be detectable for very quick simulation that was repeated a very large number of times. So if you need the other features, I'd switch to that interface, esp if you don't know there is a benefit.

Kyle