Skip to content

[Bug]: Richards growth curve no longer converging in 3.30.25 #772

Description

@zandjyo

Describe the bug

Anyone else having issues with fitting a Richards growth curve in Version 3.30.25? For me it throws a NaN when attempting to fit a Richards growth curve. It will fit the Von Bert fine and when the Richards parameter is fixed at 1.0 will converge on the same values as the Von Bert. However, when attempting to fit the Richards parameter the model becomes very unstable either settling at a local minima or NaNing out. This same model runs fine in all previous Stock Synthesis versions. Fixing the other growth parameters allow the model to fit a Richards parameter value, but the results are highly unstable and often settle at some local minimum that is not the MLE.

To Reproduce

Attempt to fit a Richards Growth Curve in Stock Synthesis V3.30.25 with an uniformed prior on the Richards parameter. For the EBS Pacific cod model, Stock Synthesis files can be found here: https://afsc-assessments.github.io/EBS_PCOD/2024_ASSESSMENT/NOVEMBER_MODELS/APPENDICES/APPENDIX_2.3_2024_MODELS.zip

Expected behavior

The model should converge on values similar to where it converged in prior versions of Stock Synthesis with an improved fit over the Von Bertalanffy.

Screenshots

No response

Which OS are you seeing the problem on?

Windows

Which version of SS3 are you seeing the problem on?

V3.30.25

Additional Context

No response

Metadata

Metadata

Labels

change loguse for issues that should appear in change login progressThis is being worked on in a branch

Type

Projects

Status
In Progress
Status
In Progress

Relationships

None yet

Development

No branches or pull requests

Issue actions