Fig. 6.
Bayes factor estimates obtained with Warp-III sampling for the B-model vs. Bres-model example as a function of the number of posterior samples (collapsed across all chains) used for computing the marginal likelihood for each model. Crosses indicate the median Bayes factor estimates and bars indicate the range of the estimates across the ten repetitions. The data set was generated from the B-model with n = 20 and k = 200 and is identical to the one used in the left panel of Fig. 5. The Bayes factor is expressed in favor of the data-generating model. Available at https://tinyurl.com/y599st45 under CC license https://creativecommons.org/licenses/by/2.0/