simrun ❭ modular_reduced_model_inference ❭ strategy ❭ Strategy_spatiotemporalRaisedCosine ❭ normalize

Strategy_spatiotemporalRaisedCosine.normalize¶

simrun.modular_reduced_model_inference.strategy.Strategy_spatiotemporalRaisedCosine.normalize(x, flipkey=None)¶

Normalize the kernel basis functions such that sum of all absolute values of all kernels is 1.

Attention

These are the same basis functions as in RaisedCosineBasis. These are thus not multiplied with the synapse activation data, as is the case with compute_basis()

Parameters:¶

x (array) – The learnable weights \(\mathbf{x}\) and \(\mathbf{y}\) as a 1D array. These are converted to spatial and temporal weights per group with convert_x.

Returns:¶

The normalized learnable weights \(\mathbf{x}\).

Return type:¶

array