I'm working on optimizing the parameters of a mathematical model to fit experimental data, using an existing formula for the likelihood of observing the data given a set of parameter values. At the moment I use an algorithm (bound optimization by quadratic approximation, BOBYQA) to find the values that give the maximum likelihood, but I'd like to try using MCMC methods to integrate over the likelihood function, but haven't used them before. Can anyone recommend a good approach and/or existing library for implementing MCMC methods in C/C++?

  • $\begingroup$ It's not C/C++ but you might look into pymc from the NumPy family of packages (for Python) or try a linear program solver like lp_solve which handles integer variables (and is written in ANSI C). A survey (and proposed exact sampling method) using ILP is here. $\endgroup$
    – hardmath
    Oct 30, 2017 at 15:34


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.