search

kyoungseoun-chung / pyapes

Python Awesome Partial differential Equation Solver (pyapes, a general purpose finite difference PDE solver): WIP
2 stars 0 forks source link

Periodic and Neumann/Symmetry BSc are wrong #2

Closed kyoungseoun-chung closed 1 year ago

kyoungseoun-chung commented 1 year ago

Current spatial discretization scheme cannot handle periodic and symmetry BCs since padding copies adjacent inner nodes. (so far I only tested with Dirichlet BC so couldn't find properly)

Either change implementation or need to find a way.

Previously we employed periodic by default using roll function but discarded that idea due to difficulty in Div operator. Might need to revive this approach.

kyoungseoun-chung commented 1 year ago

Constructing A matrix is required. Current implementation was only for Dirichlet and Symmetric BC. Need to revise whole process.

kyoungseoun-chung commented 1 year ago

Constructing A matrix is required. Current implementation was only for Dirichlet and Symmetric BC. Need to revise whole process.

Need separate Aops. Like Am, Ac, and Ap.

kyoungseoun-chung commented 1 year ago

Decided to deprecate fdc module.

No. I should keep this. fdc and fdm are meant to be different. (fdc returns Tensor while fdm returns class).