How to setup a problem where some constraints depend on a single variable_set and others depend on a series of variable sets

[Levi-Armstrong]

I have noticed a significant performance hit after switching over to leverage IFOPT and looking for guidance. I am using IFOPT for structuring a robot manipulator motion planning problem where each state in the trajectory is a variable set. The current challenge I have is that some constraints/cost depend on a variable_set but I have others which depend on a series of variable_sets. The constraints/cost which depend on multiple variable sets have seen a performance hit because it now iteratively fills small section of the jacobian at a time instead of performing a single fill for the series of variable sets. @awinkler Do you have an thoughts on how best to approach this is type of problem using IFOPT?