Suppose I'm solving ~1000 linear programs with ~10 variables and ~100 constraints each. Is it faster to solve these sequentially (creating a new model each time) or to create a single large linear program with components that don't interact (i.e. the components corresponding to each original linear program) and solve it once? Does the latter benefit from any sort of parallelism? Is there a point (in terms of the size of the problem) where one becomes faster than the other?
Please sign in to leave a comment.