add guidance on implications of "addtocomp" values

[iantaylor-NOAA]

"Can we provide some guidance in documentation on implications of how much bigger than >0 they should use?"

Originally posted by @Rick-Methot-NOAA in nmfs-ost/ss3-source-code#555 (review)

The current documentation (section 7.14.0.2 "Added Constant to Proportions") says "Constant added to observed and expected proportions at length and age to make logL calculations more robust. Tail compression occurs before adding this constant. Proportions are renormalized to sum to 1.0 after constant is added."

Here's a potential addition: "The constant should be greater than 0. Commonly used values range from 0.00001 to 0.01. Larger values will cause differences among bins with smaller values to be less influential, leading to greater relative influence of the bins with largest proportions of the compositions."

@Rick-Methot-NOAA, do you have additions or revisions to that change?

Also, I see that section "7.16.3 Age Composition Specification" says "If age data are included in the model, the following set-up is required, similar to the length data section." I think we should change the last part of that sentence to something like "See Length Composition Data Structure [with hyperlink] for details on each of these inputs."

[iantaylor-NOAA]

I should note that I've used 0.001 in the most recent assessments that I've worked on. The example table in the User Manual has 0.0001. Perhaps suggesting a single value (either of these would be fine) would be more helpful than providing a range of commonly used values.

[Rick-Methot-NOAA]

Is there any discussion on this topic in FIMS development?
Would be great to pick a model with noisy comp data and run a profile to demo influence of the add_to.
In addition to age and length, the same applies to generalized size comp.

[iantaylor-NOAA]

This issue has not yet been addressed for FIMS. I agree that a profile over this value would be good to see.