In previous versions of NEST, each stimulation device stored an internal representation of the data it needed for generating the stimuli for connected nodes.
In NEST 3.0, the interface for stimulation devices has been extended by a means to get this data from an external source instead of setting it directly from your PyNEST script. Such an external source could be another simulator, or a generic signal generator toolkit.
Technically, this new feature is implemented through stimulation backends that can be based on NEST has been extended by an architecture to select of Nest has been modified to include a backend for stimulation devices. This modification is inspired by the backend for recording devices. (recording from simulations)
With NEST 3.0, we change the terminology of input device to stimulation device. NEST 3.0 supports one stimulation backend, MPI communication. This backend is available only when the NEST is compiled with MPI support. This is useful in case of co-simulation and allows a closed loop simulation if it’s coupled with the MPI recording backend.
We added the new parameter
stimulus_source, which can be used to select
a backend for the stimulation device. By default, all stimulation
devices generate their own stimulation signals based on their
parametrization. If a backend, such as
mpi is selected, the
values for producing the stimuli are received from the backend’s input
All details about the new infrastructure can be found in the guide on stimulating the network.