neurical

neurical

This module provides solutions using the scientific machine learning approach (i.e. mostly based on neural networks of deep learning) where a specific loss function based on PDE with specific boundary conditions are used to force the laws of physics within the domain of interest. In contrast to numerical and analytical solutions, solution classes based on neurical solution usually requires a training (i.e. model.fit()) after compiling (i.e. model.compile()).

Hint

This module might be very interesting for you if you are learning about scientific machine learning.

List of neurical solutions:

• PINN: Physics-Informed-Neural-Network

• DeepONet: Deep-Operator-Network (not available)

Attention

More efficient and state-of-the-art solutions will be added in the future based on the research progress in the field of scientific machine learning.

Information:

• design pattern: inheritance, abstraction

• base class: Solution

• base class type: ABS (abstract)

Classes

PINN(**kwargs)	PINN solution class.
DeepONet(**kwargs)	DeepONet solution class.

Comments 💬

Feel free to make a comment, ask a question, or share your opinion about this specific content. Please keep in mind the Commenting Guidelines ⚖.