About box constraint for u

[Mesywang]

Hi Gianluca, thanks for maintaining this project!
I want to add box contraints for x and u in a ocp-qp problem, and the dim of state(x) and input(u) is 6 and 2 in my problem.
I set idxbx = {0, 1, 2, 3, 4, 5}, idxbu = {0, 1}, and set "lbx, ubx, lbu, ubu" with reasonable size and value, the result is that: all of state(x) variable is correctly limited by lbx/ubx, and the first dimension of u is correctly limited by lbu/ubu, but the second dimension of u is not bounded, it violated the constraints we set.
So, is there some problems with my usage? or can you provide a u box constraint example for me? thanks for your help!

[giaf]

Hi, the procedure you describe is correct.

Is the solver converging? I.e. what is the solver status
https://github.com/giaf/hpipm/blob/master/examples/c/example_d_ocp_qp.c#L212
or if you print the infinity norm of the residuals as here
https://github.com/giaf/hpipm/blob/master/examples/c/example_d_ocp_qp.c#L326
or the full statistics per IPM iteration as here
https://github.com/giaf/hpipm/blob/master/examples/c/example_d_ocp_qp.c#L330
do you see that convergence has been achieved?
If the solver could not converge, then the solution you get can indeed violate the constraints.

If the solver did however converge, then there should be some issue somewhere else.
In this case, if you can share a stand-alone C testing script reproducing the issue, I can have a look into it.