The (amazing) stm package contains several functions for selecting the optimal number of topics in a topic model (k). Two of these functions are searchK() and manyTopics().
Their descriptions in the documentation make them sound very similar: https://cran.r-project.org/web/packages/stm/stm.pdf
searchK(): "With user-specified initialization, this function runs selectModel for different user-specified topic numbers and computes diagnostic properties for the returned model. These include exclusivity, semantic coherence, heldout likelihood, bound, lbound, and residual dispersion."
manyTopics(): "Works the same as selectModel [for which searchK() is a wrapper], except user specifies a range of numbers of topics that they want the model fitted for. For example, models with 5, 10, and 15 topics. Then, for each number of topics, selectModel is run multiple times."
From what I can tell, one difference might be that manyTopics() automatically picks the Pareto dominant model for each choice of k, whereas searchK() supposedly allows the user to pick between them given k. (However, I haven't seen any examples on how to actually do this, or why one would want to.)
Is there actually a difference, and if so, for what purpose should each function be used?