DirichletBoundaryCondition<T>

Models & Types · AiDotNet.PhysicsInformed.PDEs

Pins the solution to a fixed value on the boundary: u = value.

Implements: IBoundaryCondition<T>

For Beginners

A differential equation on its own has infinitely many solutions — it says how things change, not where they start or end. A Dirichlet condition supplies one of the missing facts: “at the edge of the domain, the value is this”. Holding a metal rod’s end at 0°C is a Dirichlet condition.

How It Works

Five PINN types take IBoundaryCondition<T>[] in their constructors and the library implemented the interface zero times, so every user had to write this before they could solve anything (#2099). A physics-informed network without boundary conditions is not solving a boundary-value problem, so this is the common case rather than an edge one.

Example

var architecture = new NeuralNetworkArchitecture<double>(
    InputType.OneDimensional, NeuralNetworkTaskType.Regression, inputSize: 2, outputSize: 1);
            
// u = 0 wherever the first input coordinate is at 0 or 1
var bc = new DirichletBoundaryCondition<double>(boundaryValue: 0.0, lowerBound: 0.0, upperBound: 1.0);
            
var pinn = new PhysicsInformedNeuralNetwork<double>(
    architecture, new HeatEquation<double>(), new IBoundaryCondition<double>[] { bc });

Constructors

Constructor Summary
DirichletBoundaryCondition(double boundaryValue, double lowerBound, double upperBound, int coordinate, double tolerance) Initializes a Dirichlet condition on one input coordinate.

Properties

Property Summary
string Name { get; }

Methods

Method Summary
T ComputeBoundaryResidual(Vector<T> inputs, Vector<T> outputs, PDEDerivatives<T> derivatives)
bool IsOnBoundary(Vector<T> inputs)