dirichlet multinomial parameters must be labeled consecutively

[k-doering-NOAA]

It looks like Dirichlet multinomial parameters must be labeled consecutively from 1 to the last parameter in defining them in the data file in order for SS to read them correctly. For example, say a user has a model with DM parameters labelled 1 to 4 in the data file. If a user decides to drop a fleet, say, the one associated with DM parameter 1, then there will only be DM parameters for 2 to 4 (3 total parameters). SS will stil expect to read 4 parameter lines in the control file, though, and will exit on error when it cannot figure out which fleet parameter 1 (which is not longer being used) belongs with.

I think it would be helpful to either 1) exit on error and provide an informative msg when users have not used consecutive parameter numbers for DM in the data file or 2) revise the DM parameter option so that users do not need consecutive parameter numbers. The first is the simpler of the options, but I could see 2) being more useful.

[Rick-Methot-NOAA]

I will take a look at this. Please send me an example setup in which I can invoke the problem.

[k-doering-NOAA]

Thanks, just forwarded you an email with an example.

[Rick-Methot-NOAA]

The numbering of the requested D-M parameters must be consecutive and cannot exceed Nfleet.
I created a fatal warning if user tries to assign a D-M index > Nfleet.
I have not created an algorithm to figure out if any number gets skipped in the sequence.

So, this is legit and will take 6 parameters:
#_mintailcomp addtocomp combine_M_F CompressBins CompError ParmSelect minsamplesize
-0.001 1e-07 0 10 1 1 0.01 #_f2_volpal
-0.001 1e-07 0 10 1 2 0.01 #_f3_trawlers
-0.001 1e-07 0 10 1 3 0.01 #_f4_art
-0.001 1e-07 0 10 1 4 0.01 #_f5_cdtrw
-0.001 1e-07 0 10 1 4 0.01 #_SpSurv
-0.001 1e-07 0 10 1 5 0.01 #_CdSurv
-0.001 1e-07 0 10 1 5 0.01 #_PtSurv
-0.001 1e-07 0 10 1 5 0.01 #_SpCPUE
-0.001 1e-07 0 10 1 6 0.01 #_PtCPUE

but this is not legit because one number is skipped #_mintailcomp addtocomp combine_M_F CompressBins CompError ParmSelect minsamplesize
-0.001 1e-07 0 10 1 1 0.01 #_f2_volpal
-0.001 1e-07 0 10 1 2 0.01 #_f3_trawlers
-0.001 1e-07 0 10 1 4 0.01 #_f4_art
-0.001 1e-07 0 10 1 4 0.01 #_f5_cdtrw
-0.001 1e-07 0 10 1 4 0.01 #_SpSurv
-0.001 1e-07 0 10 1 5 0.01 #_CdSurv
-0.001 1e-07 0 10 1 5 0.01 #_PtSurv
-0.001 1e-07 0 10 1 5 0.01 #_SpCPUE
-0.001 1e-07 0 10 1 6 0.01 #_PtCPUE

[k-doering-NOAA]

Ok, thanks for confirming @Rick-Methot-NOAA .

I added a warning in the r4ss fxn to read datafiles that does check that there aren't any skipped values. But I could see how this algorithm to do the check would be more lines of code (and hence more time to create) in c++ code...

Perhaps the one other mitigation (besides adding the warning in SS) that could be done would be to add to the user manual?

[Rick-Methot-NOAA]

Yes! Please do add to the user manual.

[k-doering-NOAA]

Great, I added a sentence to the manual clarifying that parameter numbers cannot be skipped. I think this can be closed if there won't be any more changes to the ss source code. Thanks @Rick-Methot-NOAA !