Thanks Miles. As someone new to the whole domain of optimization, that wasn’t obvious. A question: The lines in v should in fact be normalised to the range -1…1. Is this sufficient constraint to gain benefit from JuMP, or should I still investigate Optim?