search

pacific-hake / hake-assessment

:zap: :fish: Build the assessment document using latex and knitr
MIT License
13 stars 6 forks source link

B_t / B_0 calculation always done as median(B_t) / median(B_0) ? #19

Closed andrew-edwards closed 8 years ago

andrew-edwards commented 8 years ago

It seems that B_t / B_0 is calculated from MCMC output as

median(B_t) / median(B_0) [median is calculated across MCMC values]

rather than

median( B_t / B_0) [calculate the ratio for each MCMC then take the median].

In our assessments using Awatea we did the latter, so I just want to check that the former is always done throughout the SS code. And we should mention this somewhere for clarity, and clarify the text in the one-page summary (and probably elsewhere), which last year implied that the latter definition was used.

So someone should confirm that the former is always done (I expect it is for consistency).

And the text should be clearer about this.

As an example, in the phase plot in Figure 34, some of the code is:

make.phase.plot(...) { ... sb0 <- model$mcmccalcs$sinit[2] [the [2] extracts the median ] ... sb <- smed[yrs %in% c(start.yr:(end.yr-1))]/sb0 [smed is the median spawning biomass for each year]

so it is calculating median / median.

andrew-edwards commented 8 years ago

Doing as median of ratios - keep a look out in the code for any wrong ones. Chris has done it once early on.