iChaker
commented
2 years ago Hello I am back,
yes, the grid is defined in real parameter space. In simulated_fields, I write latent parameters directly in PRF structure, this is because all the PRF structure parameters should be latent (check new comments)
I have updated the example scripts to work on your end, with precise comments at every step run order: calculate_population_response > U simulated_fields > sim specify_PRFs > PRF estimate_GLMs
@ollie-hulme
Best, Iyadh
Hi @iChaker,
with the last commit, all the example scripts don't work anymore, as they are still relying on tau and eps.
I think right now I am also having an issue with regards to setting up the grids, the latent parameter space and the real parameter space (also adding @ollie-hulme).
Do I understand it correctly, that the grid (i.e. for alpha / eta) is in real parameters space - so that alpha can be in the range [0, 1] or eta in the range [-1.5, 1.5], so the pre-computation is in real-parameter space?
For simulations (like in
simulate_fields.m), however, we need to define the latent parameters. So if we would want to use the real parameters for simulations, we would need to transfer the parameters from real to latent?The priors again are in latent space right?
All the best, Simon