Paola Pizzichetti: Warm-starting in convex MINLP
Time: Fri 2022-11-11 14.15 - 15.00
Video link: Zoom meeting ID: 686 7101 5535
There is a wide range of applications in which we must repeatedly solve mixed integer nonlinear programs. If the instances are only slightly different, can we do better than starting from scratch?
In principle, warm-starting techniques refer to the use of information about a known solution or information learned from solving a closely related instance to reduce the computational effort needed to solve the problem significantly.
Here, we discuss three stages of warm-starting with different degrees of adaptation of the lower-bounds approximation from one instance to the next. Finally, this will let us reduce the number of optimization iterations that need to be solved.