Skip to content

Latest commit

 

History

History
9 lines (5 loc) · 722 Bytes

README.md

File metadata and controls

9 lines (5 loc) · 722 Bytes

Subspace least-squares multidimensional scaling

https://arxiv.org/abs/1709.03484

© Amit Boyarski, Adi Weinberger, 2020

The code solves a stress based MDS problem where the input is a triangular mesh. It uses the SMACOF algorithm with a subspace parametrization based on the eigedecomposition of the Laplace-Beltrami operator. This parametrization allows solving huge MDS problems in a fraction of the time compared to the standard SMACOF algorithm.

In order to run the project, you should put a 'shape_name.off' file, and a pre-computed matrix of pairwise (geodeisc) distances 'D_shape_name.mat' in the 'input' folder, and update the paths in the main.py file. Then run 'main.py' and it will generate the embedding.