Framing constraints conditioned on values of optimization variables

[NISHCHAL HOYSAL G]

Hi,

Is there a way to apply constraints over an optimization variable depending on the value of another optimization variable? For example, I want to implement some thing like the following (I use python).

x = opti.variable(1,10)

y = opti.variable(1,10)

# constrain value of elements of x to be less than 0 if the corresponding element (of the same index) in y id in the range [0, 5]#

I tried to use conditional() and if_else(). But I guess they do not support evaluating conditions with symbolic variables.

Any way to do this?

Regards,

Nishchal Hoysal G

[Jonas Hjulstad]

I think you will have to use binary variables, but I haven't implemented it in CasADI before myself. Basically, you create a binary variable B (in dimension [1,10])

B.T @ x <= 0  (10 Inequality constraints, will always be 0 when B_i is 0)

B.T @ y - 5 <= 0 (Will also be fulfilled when B_i is 0, depends on y when B_i = 1)

0 - B.T @ y <= 0 

It requires a solver which handles binary variables, like bonmin.

Regards,

Jonas Hjulstad

[NISHCHAL HOYSAL G]

Thank you Jonas. I have used a different round-about approach to handle the constraints for now using ipopt. I will surely check bonmin.

Regards,

Nishchal Hoysal G