Automated generation of basis-function approximation of the synaptic functions information
The issue
Currently, a user needs to provide the information of the tau_syn_basis and associated synaptic weights for each (double) alpha function, and point the file in config files. These properties are computed externally, but they can also be derived from the (double) alpha parameters.
For better usability, it can also be computed when a TF simulation runs on a SONATA network for the first time.
Resources
The alpha-basis and weight calculation is demonstrated in this notebook. (The latter half uses variable tau_syn_basis, which gives better approximations.)
https://github.com/JavierGalvan9/V1_GLIF_model/blob/master/synaptic_data/alpha_basis_calculation.ipynb
Goal
Ideally, we would like to implement automatic calculation of these basis function expansion, and store a single file that contains both tau_syn_basis and weight coefficients. (maybe in a single h5 file?)
It should be done only once per network. (A config file specifies the file location; if found, load it; otherwise compute and store in a file.)
Potential timeline
This is not an urgent problem, but should be implemented before we have a wide user base who want to run their custom models. This can be after the workshop.
Automated generation of basis-function approximation of the synaptic functions information
The issue
Currently, a user needs to provide the information of the tau_syn_basis and associated synaptic weights for each (double) alpha function, and point the file in config files. These properties are computed externally, but they can also be derived from the (double) alpha parameters.
For better usability, it can also be computed when a TF simulation runs on a SONATA network for the first time.
Resources
The alpha-basis and weight calculation is demonstrated in this notebook. (The latter half uses variable tau_syn_basis, which gives better approximations.)
https://github.com/JavierGalvan9/V1_GLIF_model/blob/master/synaptic_data/alpha_basis_calculation.ipynb
Goal
Ideally, we would like to implement automatic calculation of these basis function expansion, and store a single file that contains both tau_syn_basis and weight coefficients. (maybe in a single h5 file?)
It should be done only once per network. (A config file specifies the file location; if found, load it; otherwise compute and store in a file.)
Potential timeline
This is not an urgent problem, but should be implemented before we have a wide user base who want to run their custom models. This can be after the workshop.