Hybrid forward sampling is a simple modification of the Probabilistic Logic Sampling algorithm that works in hybrid Bayesian networks, i.e., networks including both discrete and continuous variables. The algorithm draws samples in the forward direction, i.e., first it samples from parents-less nodes and then, once samples have been drawn from all parents of a child node, from the child node. Samples are drawn from both continuous and discrete nodes, according to the nodes' definitions. The results of the hybrid forward sampling algorithm are shown as histograms of the samples. Here is a hybrid Bayesian network (included among the example networks as Heat Equations Autodiscretized Hybrid.xdsl) updated by the hybrid forward sampling algorithm:
Please note that the hybrid forward sampling algorithm derives the marginal probability distributions over the variables in a hybrid network according to their types. In both cases, probability distribution are histograms of samples drawn during the sampling.