For hierarchical clustering, we need a distance function between clusters, considering individual instances to be the smallest clusters. This is sometimes called the linkage function. So, for example, the linkage function could be “the Euclidean distance between the closest points in each of the clusters,” which would apply to any two clusters.
