Computational Integer Programming Lecture 8: Branch and Bound Dr. Ted Ralphs Computational MILP Lecture 8 1 Reading for This Lecture Nemhauser and Wolsey Sections II.3.1, II.3.6, II.4.1, II.4.2, II.5.4 • Wolsey Chapter 7 • 1 Computational MILP Lecture 8 2 Computational Integer Optimization We now turn to the details of how integer optimization problems are • solved in practice. Computationally, the most important aspects of solving integer • optimization problems are { A method for obtaining good bounds on the value of the optimal solution (usually by solving a relaxation or dual; and { A method for generating valid disjunctions violated by a given (infeasible) solution. In this lecture, we will motivate this fact by introducing the branch and • bound algorithm. We will then look at various methods of obtaining bounds. • Later, we will examine branch and bound in more detail. • 2 Computational MILP Lecture 8 3 Integer Optimization and Disjunction As we know, the difficulty in solving an integer optimization problem • arises from the requirement that certain variables take on integer values. Such requirements can be described in terms of logical disjunctions, • constraints of the form [ x Xi 2 1≤i≤k n for Xi R ; i 1; : : : ; k. ⊆ 2 The integer variables in a given formulation may represent logical • conditions that were originally expressed in terms of disjunction. In fact, the MILP Representability Theorem tells us that any MILP can • be re-formulated as an optimization problem whose feasible region is k [ 1 t = i + intcone r ; : : : ; r F P f g i=1 is the disjunctive set defined above, for some appropriately chosen F 1 t n polytopes 1;:::; k and vectors r ; : : : ; r Z . P P 2 3 Computational MILP Lecture 8 4 Two Conceptual Reformulations From what we have seen so far, we have to two conceptual reformulations • of a given integer optimization problem. The first is in terms of disjunction: • ( k !) > [ 1 t max c x x i + intcone r ; : : : ; r (DIS) j 2 P f g i=1 The second is in terms of valid inequalities: • max c>x x conv( ) (CP) j 2 S where is the feasible region. S In principle, if we had a method for generating either of these • reformulations, this would lead to a practical method of solution. Unfortunately, these reformulations are necessarily of exponential size in • general, so there can be no way of generating them efficiently. 4 Computational MILP Lecture 8 5 Valid Disjunctions In practice, we dynamically generate parts of the reformulations (CP) and • (DIS) in order to obtain a proof of optimality for a particular instance. The concept of valid disjunction, arises from a desire to approximate the • feasible region of (DIS). k n Definition 1. Let Xi i=1 be a collection of subset of R . Then if S f g k 1≤i≤k Xi , the disjunction associated with Xi i=1 is said to be valid for an⊇ MILP S with feasible set . f g S k Definition 2. If Xi i=1 is a disjunction valid for and Xi is polyhedral for all i 1; : : : ;f k ,g then we say the disjunction isS linear. 2 f g k Definition 3. If Xi i=1 is a disjunction valid for and Xi Xj = for all i; j 1; : : :f ; k g, we say the disjunction is partitiveS . \ ; 2 f g k Definition 4. If Xi i=1 is a disjunction valid for that is both linear and partitive, we callf itg admissible. S 5 Computational MILP Lecture 8 6 Valid Inequalities Likewise, we can think of the concept of a valid inequality as arising from • our desire to approximate conv( ) (the feasible region of (CP)). S The inequality denoted by (π; π0) is called a valid inequality for if • π>x π x . S ≤ 0 8 2 S n > Note (π; π0) is a valid inequality if and only if x R π x π0 . • S ⊆ f 2 j ≤ g 6 Computational MILP Lecture 8 7 Optimality Conditions p Let us now consider an MILP (A; b; c; p) with feasible set = (Z+ • n−p S P\ × R ), where is the given formulation. + P k Further, let Xi i=1 be a linear disjunction valid for this MILP so that • nf g Xi R is polyhedral. \ P ⊆ > Then maxX \S c x is an MILP for all i 1; : : : ; k. • i 2 For each i = 1; : : : ; k, let i be a polyhedron such that Xi i • P \ S ⊆ P ⊆ Xi. P\ In other words, i is a valid formulation for subproblem i, possibly • strengthened by additionalP valid inequalities. k Note that i is itself a valid linear disjunction. • fP gi=1 We will see why there is a distinction between Xi and i later on. • P Conceptually, we are combining and relaxing the formulations (CP) and • (DIS). 7 Computational MILP Lecture 8 8 Optimality Conditions (cont'd) From the disjunction on the previous slide, we obtain a relaxation of a • general MILP. This relaxation yields a practical set of optimality conditions. • In particular, • > (1) max max n c x zIP ; i21;:::;k x2Pi\R+ ≥ which implies that if we have x∗ such that 2 S > > ∗ (OPT) max max n c x = c x ; i21;:::;k x2Pi\R+ then x∗ must be optimal. 8 Computational MILP Lecture 8 9 More on Optimality Conditions Although it is not obvious, these optimality conditions can be seen as a • generalization of those from LP. They are also the optimality conditions implicitly underlying many • advanced algorithms. There is an associated duality theory that we will see later. • By parameterizing (1), we obtain a \dual function" that is the solution • to a dual that generalizes the LP dual. 9 Computational MILP Lecture 8 10 Branch and Bound Branch and bound is the most commonly-used algorithm for solving • MILPs. It is a recursive, divide-and-conquer approach. • Suppose is the feasible set for an MILP and we wish to compute • S> maxx2S c x. Consider a partition of into subsets ;::: k. Then • S S1 S max c>x = max max c>x x2S f1≤i≤kgfx2Si g . In other words, we can optimize over each subset separately. • Idea: If we can't solve the original problem directly, we might be able to • solve the smaller subproblems recursively. Dividing the original problem into subproblems is called branching. • Taken to the extreme, this scheme is equivalent to complete enumeration. • 10 Computational MILP Lecture 8 11 Branching in Branch and Bound k Branching is achieved by selecting an admissible disjunction Xi and • f gi=1 using it to partition , e.g., i = Xi. S S S\ We only consider linear disjunctions so that the subproblem remain • MILPs after branching. The reason for choosing partitive disjunctions is self-evident. • The way this disjunction is selected is called the branching method and • is a topic we will examine in some depth. ∗ ∗ Generally speaking, we want x 1≤i≤kXi, where x is the (infeasible) • solution produced by solving the62 [bounding problem associated with a given subproblem. A typical disjunction is • ∗ X = xj x ; (2) 1 f ≤ b j cg ∗ X = xj x ; (3) 2 f ≥ d j eg where x∗ argmax c>x. 2 x2P 11 Computational MILP Lecture 8 12 Bounding in Branch and Bound The bounding problem is a problem solved to obtain a bound on the • > optimal solution value of a subproblem maxSi c x. Typically, the bounding problem is either a relaxation or a dual of the • subproblem (these concepts will be defined formally in Lecture 7). Solving the bounding problem serves two purposes. • { In some cases, the solution x∗ to the relaxation may actually be a feasible solution (x∗ , in which case c>x∗ is a global lower bound l( ). 2 S S { Bounding enables us to inexpensively a bound b( i) on the optimal solution value of subproblem i. S If b( i) l( ), then i can't contain a solution strictly better than the • bestS one≤ foundS so far.S Thus, we may discard or prune subproblem i. • 12 Computational MILP Lecture 8 13 Constructing a Bounding Problem There are many ways to construct a bounding problem and this will be • the topic of later lectures. The easiest of the these is to form the LP relaxation max n , P\R+\Xi • obtained by dropping the integrality constraints. For the rest of the lecture, assume all variables have finite upper and • lower bounds. 13 Computational MILP Lecture 8 14 LP-based Branch and Bound: Initial Subproblem In LP-based branch and bound, we first solve the LP relaxation of the • original problem. The result is one of the following: 1. The LP is infeasible MILP is infeasible. 2. We obtain a feasible) solution for the MILP optimal solution. 3. We obtain an optimal solution to the LP that) is not feasible for the MILP upper bound. ) In the first two cases, we are finished. • In the third case, we must branch and recursively solve the resulting • subproblems. 14 Computational MILP Lecture 8 15 Branching in LP-based Branch and Bound In LP-based branch and bound, the most commonly used disjunctions • are the variable disjunctions, imposed as follows: { Select a variable i whose value x^i is fractional in the LP solution. { Create two subproblems. In one subproblem, impose the constraint xi x^i . ∗ ≤ b c In the other subproblem, impose the constraint xi x^i . ∗ ≥ d e What does it mean in a 0-1 problem? • 15 Computational MILP Lecture 8 16 The Geometry of Branching Figure 1: The original feasible region 16 Computational MILP Lecture 8 17 The Geometry of Branching (cont'd) Figure 2: Branching on disjunction x 2 OR x 3 1 ≤ 1 ≥ 17 Computational MILP Lecture 8 18 Continuing the Algorithm After Branching After branching, we solve each of the subproblems recursively.