snnib.blender.network.Network

class snnib.blender.network.Network(network_container: bpy.types.Object, template_neuron: bpy.types.Object, network_file: str = None)

container representing a spiking neural network

Attributes

• network_container – container defining the boundaries of the SNNIB network inside [blender](https://www.blender.org/)

• template_neuron – object serving as template for all neurons in the network

• network_file – SNNIB compatible file representing a network output from a simulation

Inferred Attributes

• axon_length

  • float

  • length of the axon root

  • axon only starts to branch out after that point

  • obtained from [blender](https://www.blender.org/) UI

• axon_objects

  • List[bpy.types.Object]

  • objects of all axons that are part of the network

• n_neurons

  • int

  • number of neurons contained in the network

  • obtained from [blender](https://www.blender.org/) UI

  • only relevant for randomly generated network

• neuron_objects

  • List[bpy.types.Object]

  • objects of all neurons that are part of the network

• p_synapses

  • float

  • probability of a synapse forming between any two neurons

  • obtained from [blender](https://www.blender.org/) UI

  • only relevant for randomly generated network

• p_spike

  • float

  • probability of a neuron spiking at any time

    • i.e., on for every neuron for every frame p_spike a spike is emitted with probability p_spike

  • obtained from [blender](https://www.blender.org/) UI

  • only relevant for randomly generated network

• Rng

  • np.random.Generator

  • random number generator to use for network generation

• seed

  • int

  • random seed

  • obtained from [blender](https://www.blender.org/) UI

  • only relevant for randomly generated network

Methods

• _get_mean_outconnection() – returns mean direction of outgoing connections of some neuron

• generate_network() – generates a random network

• read_network() – loads a network from a file

• setup_container() – sets up the network container

• draw_neurons() – creates neurons and adds them to the scene

• draw_synapses() – creates outgoing connections and ads them to the scene

Dependencies

• bpy

• bmesh

• json

• logging

• numpy

• typing

__init__(network_container: bpy.types.Object, template_neuron: bpy.types.Object, network_file: str = None)

constructor

• generates a random network if no network_file was found

• otherwise reads the network from network_file

Parameters

• network_container

  • bpy.types.Object

  • object representing the network container in the scene

  • bounding box of network_container defines the region the network will occupy

    • neuron coordinates will be remapped to lie within the bounding box

• template_neuron

  • bpy.types.Object

  • object representing a template to use for every neuron contained in the network

    • neurons in the network are instanced from template_neuron

  • you only need to modify template_neuron in order to procedurally apply the changes to all neurons in the network

• network_file

  • str, optional

  • file containing a save network to load into SNNIB

  • the default is None

    • use random generation instead of reading from a file

Raises

Returns

Methods

__init__(network_container, template_neuron)	constructor
draw_neurons()	draws neurons into the scene
draw_synapses()	draws synapses into the scene
generate_network()	generates random network based on user input
read_network()	reads a network from self.network_file
setup_container()	sets up the network container

draw_neurons()

draws neurons into the scene

• will

  • instantiate neurons based on self.template_neuron

  • initialize axons (axon roots)

  • apply geo nodes inputs

  • adjust geo nodes mappings (to roughly match actual simulation)

Parameters

Raises

Returns

draw_synapses()

draws synapses into the scene

• will

  • add spline to respective axon object for every existing synapse

Parameters

Raises

Returns

generate_network()

generates random network based on user input

• will

  • generate random coordinates within `self.network_container`s bounding box

  • generate random spiketrains for every neuron

    • for every frame there is a probability self.p_spike to emit a spike

  • generate random connections between neurons

    • for every pair of neurons a connections exists with probability self.p_synapses

  • generate metadata corresponding to the render settings

Parameters

Raises

Returns

read_network()

reads a network from self.network_file

• will

  • read the network elements and parameters

  • map the contained neuron coordinates to `self.network_container`s bounding box

Parameters

Raises

Returns

setup_container()

sets up the network container

• will

  • clear all children of self.container

  • add the respective geo nodes node tree

Parameters

Raises

Returns