When I read the 1.7 Gaussian process in the scikit-learn user guide (https://scikit-learn.org/stable/modules/gaussian_process.html#gaussian-process-regression-gpr), I found that the hyperparameters of the Gaussian process can be optimized not only using the log-marginal-likelihood method but also Markov chain Monte Carlo(MCMC). But I don't know how to implement this method, does anyone know, thanks for anyone's answer.