Purpose

Intended to give good interpolation results when input data is non-uniform, anisotropic and potentially sparse. Will also give good results for regular grids (though more efficient methods generally exist for these cases).

Method

This code uses Natural Neighbour Interpolation. This requires that a Delaunay Mesh of the data is created, (a simple flipping algorithm is used with robust predicates provided by (link to come). The Voronoi Cells are then extracted, and the weighting function is computed based on the volumes stolen by the newly created Voronoi Cells. I wrote a short report explaining various interpolation functions and their implementations which this code was part of, this report can also be accessed from the download link.

Creator

Created by Ross Hemsley for the Neutron Simulation package Mcstas.

Use

Code has been written not to require any external dependencies. Should simply compile and run. Read documentation in code for usage notes. If you wish to use this code, please drop me an e-mail to let me know what you are using it for.

FAQ

Q: I can't seem to compile the code

A: The attached makefile will only work for the unit testing inside the main files, you will need to enable the testing by uncommenting the relevent lines in the file to use this. If you wish to use example.c you need to compile it yourself, try

$ gcc example.c natural.c delauny.c utils.c -03