Given the following two ways of creating the same constraint
def constraint(x):
return x[0] + x[1] # or potentially something much more complicated
...
x = prog.NewContinuousVariables(2,)
prog.AddConstraint(constraint, ..., vars=[x[0], x[0]]) # same decision variable specified twice
def constraint(x):
return 2*x[0]
x = prog.NewContinuousVariables(2,)
prog.AddConstraint(constraint, ..., vars=[x[0]])
# constraint function written in a way that decision variable only specified once
My question is, are these two cases equivalent, both performance-wise and numerically?
The reason I ask is, I'm iteratively building up a constraint involving a long chain of vars in which there may be duplicates of the same decision variable. I'm wondering if I should spend the effort to refactor the constraint such that each decision variable only appears once, or is drake smart enough to handle that for me.
Drake should handle duplicated variables for you. If you encounter unexpected behaviors with duplicated variables, please file a GitHub issue. Thanks