pihnn.nn

Definition of PIHNN networks.

Attributes

device

Classes

ComplexLinear	Extension of torch.nn.Linear to complex values.
ComplexParameter	Extension of torch.nn.Parameter to complex values.
PIHNN	Main class for the employment of physics-informed holomorphic neural networks (PIHNNs) from Calafà et al. [2024].
DD_PIHNN	Domain-decomposition physics-informed holomorphic neural networks (DD-PIHNNs).
enriched_PIHNN	PIHNN with enrichment for cracks, as introduced in Calafà et al. [2025].
L_PIHNN	Laurent physics-informed holomorphic neural networks (L-PIHNNs) from Calafà et al. [2026].
PIHKANLayer	Single layer for pihnn.nn.PIHKAN.
PIHKAN	Physics-informed holomorphic Kolmogorov-Arnold network (PIHKAN).
DD_PIHKAN	Physics-informed holomorphic Kolmogorov-Arnold networks (PIHKANs) with domain decomposition.
enriched_PIHKAN	Physics-informed holomorphic Kolmogorov-Arnold networks (PIHKANs) with enrichment for cracks.
L_PIHKAN	Laurent physics-informed holomorphic Kolmogorov-Arnold network (L-PIHKAN).

Functions

custom_format_warning(message, category, filename, lineno)	Custom formatting for warnings.
derivative(f, z[, holom, conjugate])	Compute the derivative \(\frac{df}{dz}\) through PyTorch automatic differentiation.

Module Contents

pihnn.nn.device