Haikun Xu

Hi Jim, I tried what you suggested here "fit = fit_model( ..., build_model=FALSE )" to change parameter bounds. But I did not find fit$tmb_list$Lower; fit$tmb_list only includes Map, Data, Parameters,...

Hi Kelli, I really appreciate your effort to dig into the code to make sure that Mohn's rho is computed correctly for "quarters-as-years" models.

Yes, I have sent @kellijohnson-NOAA the cpp code for estimating the two autocorrelation parameters for selectivity deviations. It would make the usage of semi-parametric selectivity much more convenient if r4ss...

Hi Ian and Rick, Sorry for the late reply. I have not tested a case with multiple sigma values before. I don't see any bug in that part of the...

Thank you @iantaylor-NOAA. I understand why rho has no impact. When using age-specific signamsel, the likelihood is computed based on rho = 0 regardless of rho input (see how rho...

For the near term, the most we can do is testing the age-specific option with no rhos (we can specify several identical sigma_sels and compare the likelihood with that from...

Based on Ian's hake example, I tested the age-specific sigma_sel option with several identical sigma_sels and found that it provides the same results and likelihoods as using the single sigma_sel...

@Rick-Methot-NOAA Sure. The model in which multiple sigma_sels are estimated converged with a positive definite Hessian, so no sign it is not working correctly,