DEPPISCH1 |
Markov Chain
Monte Carlo Techniques for Parameter Inference |
Type |
Theoretical |
#students |
1 |
Orientation |
Fitting experimental data to theoretical models is a crucial task in
many physics analyses. If the number of parameters in a model is large or if
the relation between the parameters and the experimental observables is
complicated, this can be difficult and non-trivial. Among other applications,
Monte Carlo techniques based on randomly (but adaptively) exploring the
parameter space have been proven highly successful in this regard. |
What |
In this project, you will learn the basic principles of so called
Markov Chain Monte Carlo algorithms for statistical analysis. This includes a
review of existing types of algorithms and tools. You will implement your own
analysis software framework, and compare it to other solutions, and you will
apply it to a real life physics problem (fitting neutrino parameters in particle
physics). |
Special Knowledge |
Programming knowledge
(C++ or Java preferred), Interest in numerical and statistical analysis |
Supervisor |
Dr. Frank Deppisch (f.deppisch@ucl.ac.uk) |