gridhopping
Tests how fast is gridhopping when compared to the basic O(N^3) method.
Methods for generating signed distance estimators
We use an estimator based on random fourier features when the mapping is learned by ordinary least squares. This code is contained in the rff-experiments folder.
Also, we use an ordinary feed forward neural netowrk to approximate the distance bound: nn-experiments.
To generate these estimators for a specified 3D mesh, there is a learning phase. The 3D meshes used in our experiments are described in the next section.
3D models
You can download the models using the provided script: models/dl.sh.
