A few IPython notebooks to show examples of inference of exposure times and erosion rates from measured cosmogenic nuclides concentrations vs. depth.
It compares both Bayesian and Frequentist (Maximum Likelihood Estimation) approachs with different optimization and/or sampling algorithms (grid search, non-linear fitting, MCMC...).
- General notes about the statistical inference methods
- A Model of nuclides concentration vs. depth
- Datasets
Maximum Likelihood Estimation w/ Grid Search:
Maximum Likelihood Estimation w/ non-linear fitting:
Bayesian approach w/ MCMC:
Bayesian approach w/ MCMC (more robust, based on emcee:
Author: B. Bovy
License: MIT (see LICENSE)