Unexpected spike on start of measure_node_collection

[simonmichau]

Each time measure_node_collection is called there appears a membrane potential spike in the beginning of the measurement without any reason. This likely has to do with the connection of new measurement devices. A side-effect seems to be that the real-time factor increases drastically with each new measurement.

[zbarni]

Is it critical to add new devices after simulation start, or can everything be created and connected in the beginning?

[simonmichau]

Measurement devices are only connected once now. This doesn't solve the problem that the connection of a measurement device causes a spike in the first place.

(Image shows a measurement without noise or pattern input)

[zbarni]

It's not clear whether the connection of the measurement device causes the increased Vm initially or it just captures it.. a very simple way to debug this is to simulate for a short time, and print out the Vm at each step from the C++ code without connecting the recorder. Or connect it a bit later, say at 500 ms, and see if the voltage increases only then.