when I check the execution time of the statements between the cvx_begin and cvx_end the most time consuming parts are “constraint definition”; but the final method which is minimize(objectiveFunction) is really fast. Can i say the solving time of my program is low while the problem transformation is high?

If the answer is yes, then is it possible to change the value of constant parameters after the problem definition phase, and just run the final method without spending time for recalculation of problem definition phase?

Best regards.

The model transformation process is *everything* that happens from the `cvx_begin`

statement, until the solver actually starts. Even `cvx_end`

does a certain amount of problem transformation.

No, there is no way to avoid this cost when solving a second or third problem. Every model is constructed fresh.

1 Like