Edoardo Balzani
Edoardo Balzani
Gamma glm ready to review. I edited a couple of things in the main GLM class. Mainly, now you can pass a function that aggregates the likelihood of different samples....
Implement a class that wraps `jaxopt` solvers and performs batching. If users passes an array on the GPU, keep it there, otherwise transfer. If the user passes an hdf5 lazy-loaded?...
If you add more than one basis, or if you have a basis and an intercept, there is a family of solutions that are equivalent. How do we constraint the...
Use data in nwb from `Sinaps` probes (>700 units), fit a fully coupled model to test the scalability of the batched GLM. Add some calcium imaging data once the gamma...
Add a flag to allow custom solvers to be passed
Add a gamma observation model for calcium imaging analysis
Notes are out-of-sync after we changed the API for GLM to single neuron. Both `base_regressor` and `GLM` needs to be updated, GLMRecurrent has been removed.
create a function that creates the coupling matrix, with a few option, as in the magic_designer method, see #65 for which argument should be passed
- Added `FourierBasis` to the basis module - tests and docs modified accordingly
Make it clear why neuroscientist should use this package compared with alternatives: include examples like: predict connectivity, grid cell map...