More information needed on dev_se and dev_autocorr parameters

[iantaylor-NOAA]

I've been working with @charliehinchliffe on understanding the impact of Q deviations as implemented by SSMSE. SSMSE inputs time-varying LnQ_base_..._DEVadd parameters in the .par file but these were getting rescaled by the LnQ_base_..._dev_se parameter, causing confusion.

The role of these two parameter is confusing for a few reasons. I'm happy to help revise this section, but any input on how to do that would be welcome:

• The terms "dev_se" and "dev_autocorr" used in the parameter labels can't be found in the user manual because different notation is used (see text below). Adding those parameter label suffixes alongside the math notation will help find this line within the 14.1.3 Specification of Time-Varying Parameters: Short Parameter Lines
  
• The description of the deviation link 14.1.2 Specification of Time-Varying Parameters: Long Parameter Lines doesn't make clear where the parameters are specified (some kind of link to the section below would be helpful here)
  
• If there's an easy explanation it would be helpful to explain the reasons for this difference: "SS3 expects the estimated deviations to be normal in distribution and the deviation values are multiplied by the standard error parameter as they are used. This differs from recruitment deviations and from the approach in SS3 v.3.24."

[Rick-Methot-NOAA]

I have shared a draft google doc with SS3 team so we can work on completing that 5 ear old essay on this topic.

[e-perl-NOAA]

Rick, are you thinking that the essay could be included as an additional vignette or that it would be incorporated into the user manual?

[Rick-Methot-NOAA]

Elizabeth, I think both. Devs in SS3-ADMB are not nearly as sophisticated as the random effects in TMB, so their use needs sidebars. That is intention of the essay, which will need close scrutiny given the time since it was drafted. Then we need to be sure the manual is not inconsistent with it and includes a link to the essay.