search

pbs-assess / herringsr

:paperclip: :fish: Herring Science Response document
0 stars 1 forks source link

Autocorrelation plots only show 100 values #117

Closed grinnellm closed 2 years ago

grinnellm commented 2 years ago

Our autocorrelation plots only show 100 values but I expect them to have 4,000 points like the MCMC trace plots. 5,000,000 chain length, 1,000th thinning, and 1,000 burn-in makes 4,000 values.

We are on the herring branch of gfiscamutils.

The plot is made here.

image

cgrandin commented 2 years ago

I think you just change the lag.max argument..

https://github.com/pbs-assess/gfiscamutils/blob/d41552634a5b4c54fe6be2ce371ea1b6968c6ade/R/figures-mcmc-diagnostics.R#L304

cgrandin commented 2 years ago

And if you're calling it from make.autocor.plot() it's hard coded in there: https://github.com/pbs-assess/gfiscamutils/blob/d41552634a5b4c54fe6be2ce371ea1b6968c6ade/R/figures-mcmc-diagnostics.R#L282

That should be passed in the elippsis ... or as an argument in make.autocor.plot()

grinnellm commented 2 years ago

Thanks for this @cgrandin. It could be that I don't understand what the plot is supposed to show.. is it supposed to show all 4,000 MCMCs (after thinning and burn-in accounted for)?

grinnellm commented 2 years ago

I'm pretty sure I misunderstood what the plot was showing, and it's correct as is.