I believe I found a bug in the nest interface nest.Recorder._recorder function:
def _record(self, variable, new_ids, sampling_interval=None):
"""
Add the cells in new_ids to the set of recorded cells for the given
variable. Since a given node can only be recorded from by one multimeter
(http://www.nest-initiative.org/index.php/Analog_recording_with_multimeter, 14/11/11)
we record all analog variables for all requested cells.
"""
if variable == 'spikes':
self._spike_detector.add_ids(new_ids)
else:
self.sampling_interval = sampling_interval
self._multimeter.add_variable(variable)
self._multimeter.add_ids(new_ids)
The problem is that the variable=='spike' condition doesn't contain the line:
self.sampling_interval = sampling_interval
If one calls the pyNN record function on population, one can specify sampling_interval. If one makes multiple such calls, for example with different variables, pyNN is checking if all of them are using the same sampling interval, by checking against the self.sampling_interval. This works fine, if the first call of record is on variable that is not 'spikes'. On that call the code above set's the self.sampling_interval to a value, and then the next calls have to be with the same value (see assert in the Recorder.record). But if the first call is with a variable 'spikes' this doesn't happen (because of the missing line abobe), and even if all the calls to record are with the sampling_interval of the same value it complains that they are not.
I believe I found a bug in the nest interface nest.Recorder._recorder function:
def _record(self, variable, new_ids, sampling_interval=None): """ Add the cells in
new_idsto the set of recorded cells for the given variable. Since a given node can only be recorded from by one multimeter (http://www.nest-initiative.org/index.php/Analog_recording_with_multimeter, 14/11/11) we record all analog variables for all requested cells. """ if variable == 'spikes': self._spike_detector.add_ids(new_ids) else: self.sampling_interval = sampling_interval self._multimeter.add_variable(variable) self._multimeter.add_ids(new_ids)The problem is that the variable=='spike' condition doesn't contain the line: self.sampling_interval = sampling_interval
If one calls the pyNN record function on population, one can specify sampling_interval. If one makes multiple such calls, for example with different variables, pyNN is checking if all of them are using the same sampling interval, by checking against the self.sampling_interval. This works fine, if the first call of record is on variable that is not 'spikes'. On that call the code above set's the self.sampling_interval to a value, and then the next calls have to be with the same value (see assert in the Recorder.record). But if the first call is with a variable 'spikes' this doesn't happen (because of the missing line abobe), and even if all the calls to record are with the sampling_interval of the same value it complains that they are not.