MexicanHat

Contents

MexicanHat#

class braintools.conn.MexicanHat(sigma, max_distance=None, weight=None, delay=None, **kwargs)#

Mexican hat (Laplacian of Gaussian) connectivity pattern.

A special case of DoG with specific amplitude ratios to approximate the Laplacian of Gaussian. Creates strong lateral inhibition patterns.

Parameters:
  • sigma (Array | ndarray | bool | number | bool | int | float | complex | Quantity) – Standard deviation of the Gaussian (surround sigma will be sqrt(2)*sigma).

  • max_distance (Array | ndarray | bool | number | bool | int | float | complex | Quantity | None) – Maximum distance for connections (default: 4*sigma).

  • weight (Initialization | float | int | ndarray | Array | Quantity | None) – Weight initialization.

  • delay (Initialization | float | int | ndarray | Array | Quantity | None) – Delay initialization.

Notes

The weight at distance d is 2 * G(d; sigma) - G(d; sqrt(2) * sigma) with unit-peak Gaussians from braintools.init.GaussianProfile (see DoGKernel). This is not braintools.init.MexicanHatProfile, which is a Ricker wavelet clipped at zero.

Candidate pairs are found with a k-d tree over the positions and handled in chunks of bounded size, so memory grows with the number of edges rather than with pre_size * post_size. The edges, their order and the weights are the same as in version 0.3 (up to rounding). Generation runs on the host with NumPy and cannot be traced by jax.jit, jax.vmap or jax.grad.

Examples

>>> import brainunit as u
>>> import numpy as np
>>> from braintools.conn import MexicanHat
>>> positions = np.random.RandomState(0).uniform(0, 1000, (500, 2)) * u.um
>>> mexican = MexicanHat(
...     sigma=40 * u.um,
...     weight=1.0 * u.nS
... )
>>> result = mexican(
...     pre_size=500, post_size=500,
...     pre_positions=positions, post_positions=positions
... )
>>> result.n_connections
17110