1001 dalmiterations, shinystan


#1

I ran a new model for 1001 iterations just to have something to plot and when I launch “shinystan” it … crashes.
Ok, maybe 1 sampling iterations is a corner case so I edit my CmdStan output .csv file to pretend it has 999 warmup and 2 sampling iterations and shinystan starts up. That’s not ideal but ok.

Then I go to the ‘explore’ view and for the bivariate view I can’t include sampling iterations. … is that for real or am I just being thick and missing where the checkbox is? In any case would be useful to allow exploratory plotting on sampling iterations. I guess I’m using CmdStan so I can just further edit the input to pretend that I have lots of sampling iterations but not everyone is going to want to do that.


#2

This should be a bug report for shinystan. @jonah is the one to handle it.


#3

Didn’t know if they consider these bugs so I posted instead.


#4

If it crashes when there are 1001 iterations, but not when there are 999 and there’s no magic number requirement in the doc, then it’s a bug.


#5

I believe this applies only to code written in programming languages not named “R”.