Clean add generic submodel#143

Open

lisa-gm wants to merge 78 commits into

devfrom

clean_add_generic_submodel

Commits on Feb 16, 2026

first steps in the direction of ar1

lisa-gm
authored and
vincent-maillou
committed
ar1 model runs through. not sure if it computes the correct the thing or not

lisa-gm
authored and
vincent-maillou
committed
gaussian ar1 seems to be working. still needs a bit of clean up

lisa-gm
authored and
vincent-maillou
committed
added verbosity level to not always show timer, added initial draft for rescaling function for hyperparameters within prior hyperparameter class and from where it needs to be called

lisa-gm
authored and
vincent-maillou
committed
updated selected inversion to be dense inverse. probably due to natural ordering and no pivoting numerically quite unstable. only seems to work for very small expamples

lisa-gm
authored and
vincent-maillou
committed
added doc string and normalizing constant for both gaussian priors

lisa-gm
authored and
vincent-maillou
committed
first real progress with the reparametrization issue. seems to be working now for minimize() call for gr with gamma prior

lisa-gm
authored and
vincent-maillou
committed
work in progress, lots of quadrature. lots of test. nothing in core yet

lisa-gm
authored and
vincent-maillou
committed
incorporating new functions into dalia

lisa-gm
authored and
vincent-maillou
committed
just updated run script

lisa-gm
authored and
vincent-maillou
committed
fixed it again ...

lisa-gm
authored and
vincent-maillou
committed
no print in broadcasting function

lisa-gm
authored and
vincent-maillou
committed
removed extra print statements

lisa-gm
authored and
vincent-maillou
committed
added print gs small example

lisa-gm
authored and
vincent-maillou
committed
Update theta handling in check_vector_consistency

lisa-gm
authored and
vincent-maillou
committed
Check backend flags for cupy availability

lisa-gm
authored and
vincent-maillou
committed
incorporated copilot suggestions

lisa-gm
authored and
vincent-maillou
committed
did it resolve now ...

lisa-gm
authored and
vincent-maillou
committed
first version seems to be working for some priors

lisa-gm
authored and
vincent-maillou
committed
first version seems to be working ...

lisa-gm
authored and
vincent-maillou
committed
added first calls in run scripts. seems to look ok at first glance

lisa-gm
authored and
vincent-maillou
committed
added forgotten file

lisa-gm
authored and
vincent-maillou
committed
edited tests gaussian quadrature

lisa-gm
authored and
vincent-maillou
committed
updated test reparametrization

lisa-gm
authored and
vincent-maillou
committed
added directions beta prior

lisa-gm
authored and
vincent-maillou
committed
fixed which parameter scale is called, added missing link fucntion, added plotting, modified dalia calls

lisa-gm
authored and
vincent-maillou
committed
gr seems to work

lisa-gm
authored and
vincent-maillou
committed
now working for gaussian regression

lisa-gm
authored and
vincent-maillou
committed
bivariate quadrature + correlation work in progress ...

lisa-gm
authored and
vincent-maillou
committed
ar1 seems to work

lisa-gm
authored and
vincent-maillou
committed
started adding inverse gamma prior

lisa-gm
authored and
vincent-maillou
committed
updated priors in AR1 examples. seemed to have fixed the not spd issue which makes sense. before we could have negative variance / precision ...

lisa-gm
authored and
vincent-maillou
committed
just updates run.py files

lisa-gm
authored and
vincent-maillou
committed
fixed problem with scipy triangular solve that was making scipy solver compute incorrect residual

lisa-gm
authored and
vincent-maillou
committed
fixed return type declarations of functions and removed main in scipy solver

lisa-gm
authored and
vincent-maillou
committed
found another incorrect return type declaration

lisa-gm
authored and
vincent-maillou
committed
EXPL: applied stashed changes, mainly small updates in the run scripts that were necessary after updating dalia

lisa-gm
authored and
vincent-maillou
committed
ENH: incorporated changes from stashed version for gpu, still some errors but need to migrate to alex

Lisa Gaedke-Merzhaeuser
authored and
vincent-maillou
committed
added gpu compatible version

lisa-gm
authored and
vincent-maillou
committed
ENH: added automatic folder generation for generate_data.py scripts in g_ar1 and p_ar1

lisa-gm
authored and
vincent-maillou
committed
WIP: Fixing Brainaic on new_stats_capabilities
vincent-maillou
committed
WIP: modified the flow of the marginal computation to accommodate the GaussianMVC prior (it's a hack for now)
vincent-maillou
committed
WIP: brainiac marginal computation working
vincent-maillou
committed
WIP: formating
vincent-maillou
committed
WIP: I honestly don't know what short is appropriate but I'm putting one. Applied minor stashed changes from my brainiac model work.

lisa-gm
authored and
vincent-maillou
committed
WIP: updated run and plotting script brainiac

lisa-gm
authored and
vincent-maillou
committed
EXPL: renamed cov_theta to cov_theta-internal in overlooked example run.py files

lisa-gm
authored and
vincent-maillou
committed
WIP: Added examples to run scripts
vincent-maillou
committed

Commits on Mar 6, 2026

WIP: Fixed typos and solved GPU running for some examples
vincent-maillou
committed

Commits on Mar 18, 2026

Commits on Apr 9, 2026

Commits on Apr 14, 2026

Commits on Apr 22, 2026

STATS: added generic model
lisa-gm
committed

Commits on May 13, 2026

Commits on May 20, 2026

ENH: synthetic bee genomics like model seems to be working
lisa-gm
committed

Commits on Jun 9, 2026

first version of run script that exports sample variance
lisa-gm
committed

Commits on Jun 10, 2026

added sample functionality for latent parameters
lisa-gm
committed