sli_neuron - neuron with SLI callback
The sli_neuron is a model whose state, update and calibration can be
defined in SLI.
The state of the neuron is a SLI dictionary which can be retrieved
with GetStatus. The state should contain two procedures:
1. /update
2. /calibrate
/calibrate is called before the simulation starts. It is used to
pre-compute constants of the dynamics and to scale
parameters to the temporal resolution. A minimal
implementation of calibrate is:
/calibrate { GetResolution /h Set } def
/update is called during simulation and must propagate the node's
state by one integration step h. If /update decides that
the node should spike, it must set the variable /spike to
true.
Both /calibrate and /update are called with the node's status
dictionary on top of the dictionary stack. This means that node
variables override names in the systemdict or in the userdict.
Moreover, all definitions are done in the node's statusdict and
persist throughout the simulation.
Errors.
If an error occurs during the evaluation of /calibrate or /update, the
errorneous neuron is skipped and update proceeds to the next node
until the time-slice is finished. After the time-slice is finished,
simulation terminates with an additional error message that issues the
global id of the errorneous node.
Errors are handled by the SLI standard errorhandler. In addition, the
contents of the error dictionary is copied into the node's status
dictionary, to allow debugging of the node.
SpikeEvent, CurrentEvent, DataLoggingRequest
SpikeEvent
Diesmann, Plesser, Gewaltig
January 2009
/var/www/debian/nest/nest-simulator-2.20.0/nest/sli_neuron.h