University
Bayesian statistics
-
Project in Bayesian (nonparametric) clustering: introductory slides and code
It compares the behaviour of two different loss functions (Binder and variation of information) for partition estimation, when the parameters of the prior are misspecified.
(Probabilistic) Measure theory
- The Borel-Kolmogorov paradox
, reason for which you need to define the limiting operation when you condition on a
0 probability event; measure theory “solves” this issue of conditional probability reasoning in terms
of sub-σ-algebras, introducing the concept of regular (version of a) conditional probability: conditional
probability is defined as the conditional expectation of an indicator function.
To clarify concepts more, navigate to this Wikipedia page
A (not so) simple exercise in probability
- I think it makes a good effort at showing the power of abstraction in Mathematics
Numerical Analysis
- My GitHub repository for a C++ interface for fixed-point iterations. In particular, it implements the Anderson accelerator strategy and applies it to Finite Elements problems. In the documents folder you can also find a complete report.