Given two formulations P1 and P2 for the same integer programming (IP) problem, where Pi denotes the formulation for a problem and the corresponding polytope. We say that P1 is a stronger formulation than P2 if P1 ⊂ P2.
- Why do we prefer stronger formulations for IPs over weaker ones?
- Why is the convex hull of the feasible integer solutions the best formulation for an IP we can hope for?
- Why do we think we can solve an IP efficiently when we are given its convex hull representation?
- Why will it generally not be helpful to search for the convex hull representation of a problem?
http://am121.seas.harvard.edu/site/wp-content/uploads/2019/11/sec7-Student.pdf