CompartmentSpecific#

class braintools.conn.CompartmentSpecific(compartment_mapping, connection_prob=0.1, weight=None, delay=None, morphology_info=None, allow_self_connections=False, **kwargs)#

General compartment-specific connectivity pattern.

This is the base class for targeting specific compartments in multi-compartment neurons. It provides flexible mapping from source compartments to target compartments with customizable connection rules.

Parameters:
  • compartment_mapping (Mapping[Any, int | str | Sequence[int | str]]) – Mapping from source compartment types to target compartment types. Keys and values can be compartment indices (int) or names (str).

  • connection_prob (float | Dict) – Connection probability. Can be global or per-compartment-pair.

  • weight (Initialization | float | int | ndarray | Array | Quantity | None) – Weight initializer (e.g. braintools.init.Normal).

  • delay (Initialization | float | int | ndarray | Array | Quantity | None) – Delay initializer (e.g. braintools.init.Uniform).

  • morphology_info (Dict | None) – Information about neuron morphology structure.

  • allow_self_connections (bool) – Whether a neuron may connect to itself (autapse). When False and the pre/post populations have the same size, connections with pre_idx == post_idx are dropped regardless of compartment. Default False.

Notes

Generation runs on the host with NumPy and cannot be traced by jax.jit, jax.vmap or jax.grad.

Changed in version 0.4.0: Connected pairs are sampled by drawing a binomial edge count and then that many distinct pairs, without a dense pre_size x post_size matrix, so time and memory grow with the number of edges. The edges for a given seed differ from 0.3 (same distribution). This also applies to the subclasses that delegate to it (SomaToDendrite, AxonToSoma, BranchSpecific, AllToAllCompartments and the other targeting shortcuts).

Examples

Axon-to-soma connections:

>>> import brainunit as u
>>> import numpy as np
>>> from braintools.conn import (
...     CompartmentSpecific, AXON, SOMA, BASAL_DENDRITE, APICAL_DENDRITE
... )
>>> from braintools.init import Normal
>>> axon_soma = CompartmentSpecific(
...     compartment_mapping={AXON: SOMA},
...     connection_prob=0.1,
...     weight=Normal(mean=2.0 * u.nS, std=0.5 * u.nS),
...     seed=0,
... )
>>> result = axon_soma(pre_size=100, post_size=100)
>>> result.n_connections
993
>>> bool(np.all(result.post_compartments == SOMA))
True

Complex multi-compartment targeting:

>>> # Dendrites to soma, axon to dendrites
>>> multi_target = CompartmentSpecific(
...     compartment_mapping={
...         BASAL_DENDRITE: SOMA,
...         APICAL_DENDRITE: SOMA,
...         AXON: [BASAL_DENDRITE, APICAL_DENDRITE]
...     },
...     connection_prob={
...         (BASAL_DENDRITE, SOMA): 0.3,
...         (APICAL_DENDRITE, SOMA): 0.2,
...         (AXON, BASAL_DENDRITE): 0.05,
...         (AXON, APICAL_DENDRITE): 0.08
...     }
... )

Named compartment mapping:

>>> named_mapping = CompartmentSpecific(
...     compartment_mapping={
...         'axon': ['basal_dendrite', 'apical_dendrite'],
...         'basal_dendrite': 'soma'
...     },
...     connection_prob=0.1
... )
generate(pre_size, post_size, pre_positions=None, post_positions=None, **kwargs)[source]#

Generate compartment-specific connections.

Return type:

ConnectionResult