Hi there, (to mod: not sure where to post this, please move if I've got it wrong) I have a grid of values with 41x161 nodes describing some parameter space. Each node has an associated value, λ, which represents the uncertainty of the parameter choice at that node. I want to make/find an algorithm that samples my grid randomly, because I want to explore the parameter space. However I want the algorithm to sample the 'best' space more often, i.e. the nodes with low λ. Therefore I am looking for an algorithm that is random, but also is more likely to select the best nodes. I have two ideas: 1) Create an array with all nodes, but over represent the best nodes (i.e. put them into the array multiple times). The amount of times a node is put into the array is a function of its 'goodness'. Then randomly sample this array. Good nodes are more likely to be sample simply because they occur most often. 2) Randomly sample a node and then randomly decide whether or not to use that node based on a 'gate keeper' that has a randomly assigned threshold tolerance. Nodes that are good are more likely to pass whether or not the gate keeper has a high or low tolerance, whereas bad nodes will only pass if the gate keeper is feeling generous. Are there any well established, good ways to do this. Does what I'm doing even make sense?