devices.dynapse.mapper

devices.dynapse.mapper(graph: GraphModule, in_place=False, clustering_method: str = 'greedy', n_cluster: int | None = None, max_neuron_per_cluster: int = 256) → Dict[str, float][source]

mapper maps a computational graph onto Dynap-SE2 architecture. Each neural core shares a parameter group in Dynap-SE2. Therefore neurons inside the same core operates with the same current setting. mapper function cluster parameters in groups and allocates an hardware posititon for neurons.

See also

The tutorials in Quick Start with Dynap-SE2

Parameters:

  • graph (GraphModule) – Any graph (constraints apply) aimed to be deployed to Dynap-SE2

  • clustering_method (str, optional) – the clustering approach. "random" or "greedy", defaults to "greedy"

  • n_cluster (int, optional) – number of clusters. minimum number of clusters possible if None, defaults to None

  • max_neuron_per_cluster (int, optional) – maximum number of neurons allowed per cluster, defaults to NUM_NEURONS

Returns:

a specification object which can be used to create a config object

Return type:

Dict[str, float]