GUI: Network configuration different after GUI initialization

[gtdang]

The GUI does not reconstruct the same network connectivity and drives after reconstruction from widget values.

Test:

1. Load test network configuration hnn_core/tests/assets/jones2009_3x3_drives.json. This config has all drive types (evoked, bursty, poisson) and tonic bias added.
2. Run a simulation
3. Compare the Network used to run the simulation (derived from GUI widget values) to a Network created by the API read_network_connectivity function.

Problems:

1. Number of connections is not the same. There are more bursty and poisson connections.
   • The GUI passes weights and constants dictionaries by cell types to the add_x_drive functions. Because there is an input box for each cell type, even weights with 0 are passed into the drive construction. This creates additional unnecessary null-weight connections.
   • The GUI doesn't have an input widget for the n_drive_cells parameter. It uses the defaults 'n_cells' for evoked and poisson, and 1 for bursty. The synchronous inputs checkbox does send a n_drive_cells value to drive construction but otherwise the parameter is hidden from user input. But with the default bursty drives are always synchronous.
2. The order of the drives is not the same. They are sorted alphabetically when read into the GUI.
   • gid_ranges and pos_dict are not the same as a consequence of 1 and 2
3. Tonic bias is not in the GUI-derived Network
   • This is because there isn't a way to load tonics into the GUI from file yet.
4. Cell parameters are not the same. This is being addressed in [MRG] make L5/L2 pyr have different geometry options in GUI #848

To Do:

• Fix drive order when loading drives [MRG] GUI: fix drive sorting #851
• Add tonic loading from file [MRG] feat: GUI tonic loading #852
• Filter weights and constants for only cells with non-zero weights during drive construction. [MRG] GUI update init network #853
• Add n_drive_cells input widget [MRG] GUI n_drive_cells widget #861

[gtdang]

@ntolley
I'm working on the n_drive_cells and sync_input rework today. I just want to make sure I understand the API correctly. Is this understanding correct?

1.n_drive_cells=1 and cell_specific=False - Synchronous input: 1 drive cell is connected to all network cells
2.n_drive_cells=3 and cell_specific=False - 3 drive cells are connected to all network cells
3.n_drive_cells='n_cells' and cell_specific=True - a unique drive cell gets assigned to each available simulated cell in the network with 1-to-1 connectivity
4.n_drive_cells='n_cells' and cell_specific=False - This is invalid
5.n_drive_cells=3 and cell_specific=True - This is invalid

Is 2 also "synchronous input" but with 3 spikes?

[ntolley]

This is all correct!