considerable time spent in `setPrior` for IHT and BAR

[egillax]

considerable time spent in `setPrior` for IHT and BAR

[msuchard]

Hi @egillax -- fantastic find!!! I agree that for some models we do not need to check for uniqueness. Can we add an extra boolean argument to the function to short-cut the find? In the patch, make sure to not to change the R API.

[egillax]

Hi @msuchard ,

I did dive into this a bit more. This only seems to affect the running time when models have low N (~ 1000) and high P (~ 30k). When I tested this on data with higher N I couldn't observe any speedup by skipping the uniqueness check, I assume because the model fitting itself takes bulk of the time.

I'm not sure if you want to optimize for the case of low N - which is very quick anyway - at the cost of more complexity to the code. If you do I can submit a PR. Otherwise we can close this issue.

[egillax]

I'm closing this. I don't think this will be useful for the type of data we normally use.