Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Improve handling of initial values for hierarchical distributions #255

Closed
gowerc opened this issue Feb 5, 2024 · 0 comments · Fixed by #254
Closed

Improve handling of initial values for hierarchical distributions #255

gowerc opened this issue Feb 5, 2024 · 0 comments · Fixed by #254
Labels
enhancement New feature or request

Comments

@gowerc
Copy link
Collaborator

gowerc commented Feb 5, 2024

Several parameters have their distributions fixed via the hierarchical distribution for example:

    lm_gsf_psi_bsld ~ lognormal(lm_gsf_mu_bsld[pt_study_index], lm_gsf_omega_bsld);  
    lm_gsf_psi_ks ~ lognormal(lm_gsf_mu_ks[pt_arm_index], lm_gsf_omega_ks);  
    lm_gsf_psi_kg ~ lognormal(lm_gsf_mu_kg[pt_arm_index], lm_gsf_omega_kg);  
    lm_gsf_psi_phi ~ beta(lm_gsf_a_phi[pt_arm_index], lm_gsf_b_phi[pt_arm_index]);  

Currently our framework sets their initial value via prior_none() which then uses Stans default process of drawing an initial value from a Uni(-2, 2) distribution.

prior_none() is required in order to get the current code to create an initial value without inserting an additional prior Stan statement. Potentially as a future improvement it would be good to remove prior_none() entirely and enable an additional argument to suppress inclusion of the Stan code from the existing prior_x() functions so that you can set initial values more easily.

@gowerc gowerc added the enhancement New feature or request label Feb 5, 2024
@gowerc gowerc linked a pull request Feb 6, 2024 that will close this issue
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant