skip to content

A Bayesian approach to parameter identification in Turing systems

Presented by: 
E Campillo-Funollet University of Sussex
Thursday 26th November 2015 - 11:00 to 12:30
INI Seminar Room 2
We present a methodology to identify parameters in Turing systems from noisy data. The Bayesian framework provides a rigorous interpretation of the prior knowledge and the noise, resulting in an approximation of the full probability distribution for the parameters, given the data. Although the numerical approximation of the full probability distribution is computationally expensive, parallelised algorithms produce good approximations in a few hours. With the probability distribution at hand, it is straightforward to compute credible regions for the parameters. The methodology is applied to a well-known Turing system: the Schnakenberg system.
University of Cambridge Research Councils UK
    Clay Mathematics Institute London Mathematical Society NM Rothschild and Sons