enerzyme.models.layers.rbf.GaussianRBFLayer#
- class enerzyme.models.layers.rbf.GaussianRBFLayer(num_rbf: int, cutoff_sr: float, cutoff_fn: Literal['polynomial', 'bump'] = 'bump')[source]#
Bases:
BaseRBFRadial basis functions based on Gaussian functions:
\[g_i(x) = \exp(-\mathtt{width}\cdot(x-\mathtt{center}_i)^2)\]Here, \(i\) takes values from \(0\) to \(\mathtt{num\_rbf}-1\). The center is chosen to optimally span the range \(x\in (0,\mathtt{cutoff}]\) and the \(\mathtt{width}\) parameter is selected to give optimal overlap between adjacent Gaussian functions.
- Arguments:
- num_basis_functions (int):
Number of radial basis functions.
- cutoff (float):
Cutoff radius.