BayesicFitting

Model Fitting and Evidence Calculation

View project on GitHub

 

Notes

Here we present a number of notes related to BayesicFitting.

To quote John Skilling on one of the Maxent conferences: "All these things are probably known to people who know these things."

Well.. It aint me, babe.

Data Quality.

The quality note discusses the merits of defining data quality in terms of accuracy versus weights.

Dimensions.

High dimensional space is a confusing place to be. With increasingly more space at the outskirts for higher dimensionality. How this influences Nested Sampling is inverstigated in Dimensions.

Prior Sampling.

Sampling from the Prior can lead in some cases to problems.

BoundingBox

A look at bounding boxes in higher dimensions.

Splines.

The splines note presents details on the algorithms for splines construction in

  • SplinesModel: simple, fast and dense
  • BSplinesModel: recursive de Boor algoritme, slow
  • BasicSplineModel: non-recursive de Boor, faster
  • SplinesDynamicModel: a BasicSplinesModel which is
    • Dynamic in the number of knots
    • Modifiable in the position of the knots