Final states / default state values

[kirkw-1]

When producing a final_states output using:

x_o, final_states = jitted_sim_gt()

the returned output contains 94 excess rows/components (approx 15% of my valid jaxley computed components total). The excess rows have (default?) state values of -1, with some 0. The first n rows accord with my model number of components, values are approx as expected, and model is working roughly as expected, so not sure what could be causing this output, and how it may affect computation.

Attached an extract, last few valid rows, then first few excess rows, further excess rows are identical.
extra segments example.ods

[michaeldeistler]

Hi Kirk,

yes, there are excess rows. You can get rid of them with something like the following:

final_states["eCa"][cell._internal_node_inds]

The reason for the excess rows is that, internally, Jaxley also tracks voltage at branchpoints (but no other states).

I think I agree we should remove these states by default (when running jx.integrate(..., return_states=True), so I will convert this discussion to an issue. Thanks for reporting!

Michael