arpastrana / jax_fdm Goto Github PK

It is often the case that a constrained form found structure will meet the required goals and constraints, but its appearances is jagged or rugged.
A NetworkXYZLaplacian goal can help smoothen the shape of a constrained form found solution.

Some works in the literature have illustrated the potential of the laplacian energy of the xyz coordinates of a network as an optimization regularizer, like Mark Tams's and Kenryo Takahashi's.
See here for a COMPAS implementation of the graph laplacian.

The laplacian energy of the XYZ coordinates of a mesh is

x^TLx

Where x are the xyz coordinates of the mesh and L is the Laplacian of the network.
This equation is general, and x can be replaced by any other nodal quantity of interest.