Fig. 3From: Comprehensive benchmarking of Markov chain Monte Carlo methods for dynamical systemsAnalysis pipeline for the quantitative comparison of sampling methods. a Multiple MCMC runs per scenario. b Diversity of raw results. c Removal of burn-in. d Similarity grouping of runs across scenarios of the same benchmark problem. Each frame color belongs to similar chains. e Identification of groups with good exploration quality by comparing all groupsBack to article page