pinnx.problem module

This module defines problem types for physics-informed neural networks, including PDEs, integro-differential equations, and operator learning problems. Problems encapsulate the geometry, differential equations, boundary/initial conditions, and neural network approximators.

Problem Interface

Base class for all problem types.

Problem

Base Problem Class.

Dataset Utilities

Classes for managing training data and function spaces.

DataSet	Fitting Problem set.
Function	Approximate a function via a network.
MfDataSet	Multifidelity function approximation from data set.
MfFunc	Multifidelity function approximation.
TripleDataset	Dataset with each data point as a triple.
TripleCartesianProd	Dataset with each data point as a triple.
QuadrupleDataset	Dataset with each data point as a quadruple.
QuadrupleCartesianProd	Cartesian Product input data format for MIONet architecture.

Differential Equation Problems

Problem classes for various types of differential equations.

PDE & TimePDE: For solving partial differential equations (steady-state and time-dependent).

IDE: For integro-differential equations.

FPDE & TimeFPDE: For fractional PDEs using fractional derivatives.

PDEOperator: For learning PDE operators (e.g., DeepONet).

IDE	IDE solver.
PDE	ODE or time-independent PDE solver.
TimePDE	Time-dependent PDE solver.
FPDE	Fractional PDE solver.
TimeFPDE	Time-dependent fractional PDE solver.
PDEOperator	PDE solution operator.
PDEOperatorCartesianProd	PDE solution operator with problem in the format of Cartesian product.