search

suny-downstate-medical-center / netpyne

A Python package to facilitate the development, parallel simulation, optimization and analysis of multiscale biological neuronal networks in NEURON.
http://www.netpyne.org
MIT License
142 stars 134 forks source link

[New feature] Extend declarative language to allow for specification of conn matrix #540

Open salvadord opened 3 years ago

salvadord commented 3 years ago

Although conn rules allow for more flexibility, simpler conn matrices are widely used -- these could be implemented as a special case to define connectivity, or there could be a method that converts these to conn rules.

salvadord commented 10 months ago

rquest from samneno: issue #110 "Convert tables/matrices to netpyne format (eg. pmat or wmat)" "Old-style simulations we've been using on neurosim include easy-to-read tables (wmat for weights, pmat for probability of connections, ...). These tables are typically indexed by cell type. Eg wmat[type1][type2] = 2 - means weight from type1 -> type2 is 2.0. NetPyne's syntax is more flexible/powerful but perhaps these tables are easier to read at a glance, with fast lookup of desired values via indexing. So - feature request is to have conversion to these types of tables (this is already possible given drawing of connection weights between types).

some commonly used tables: synloc - synaptic location (perhaps not general enough) delm (delays) wmatex - external weights wmat - weights pmat - probability of connections numc - number of cells of atype ix,ixe - starting/ending IDs of cells (indexed by cell type) allcells, icells, ecells - scalars representing # of cells in network and I and E cells

some tables could have added flexibility by combining with dictionaries..."

salvadord commented 10 months ago

An example of converting from pmat, wmat to netpyne conn rules is available in the M1 model