PDEDerivatives<T>
Models & Types · AiDotNet.PhysicsInformed.Interfaces
Holds the derivatives needed for PDE computation.
How It Works
The point-wise APIs (ComputeResidual, boundary residuals) read the
shapes listed on each property. The batched, tape-differentiable residual
(PDESpecificationBase.ComputeTapeResidual) reads the same layout with a leading batch axis:
[batch, output_dim, input_dim] for the gradient, [batch, output_dim, input_dim, input_dim] for the Hessian.
Constructors
| Constructor | Summary |
|---|---|
PDEDerivatives() |
Properties
| Property | Summary |
|---|---|
Tensor<T> FirstDerivatives { get; set; } |
First-order derivatives (gradient) of the output with respect to each input dimension. |
Tensor<T> HigherDerivatives { get; set; } |
Higher-order derivatives (4th order and above) if needed for the specific PDE. |
Tensor<T> SecondDerivatives { get; set; } |
Second-order derivatives (Hessian) of the output with respect to input dimensions. |
Tensor<T> ThirdDerivatives { get; set; } |
Third-order derivatives of the output with respect to input dimensions. |