Worst-case analyses, linear programming and the bin-packing problem

In this paper we consider the familiar bin-packing problem and its associated set-partitioning formulation. We show that the optimal solution to the bin-packing problem can be no larger than 4/3 ⌈Z _LP⌉, whereZ _LP is the optimal solution value of the linear programming relaxation of the set-partitioning formulation. An example is provided to show that the bound is tight. A by-product of our analysis is a new worst-case bound on the performance of the well studied First Fit Decreasing and Best Fit Decreasing heuristics. This research was supported in part by ONR Contracts N00014-90-J-1649 and N00014-95-1-0232, and NSF Contracts DDM-8922712 and DDM-9322828.     

Analysis of Classes of Heuristics for Scheduling a Two-Stage Flow Shop with Parallel Machines at One Stage

This paper addresses two-stage flow shop scheduling with parallel machines at one stage. For finding a minimum makespan schedule, which is strongly NP-hard, some efficient heuristics have been proposed in the literature. In this paper we enrich the set of heuristics by introducing a few classes of heuristics, and show that the existing heuristics can be put into this classification scheme. Furthermore, we give a complete theoretical analysis of the worst-case performance of the classes. Some empirical evaluations and comparisons for the average-case performance of a few typical heuristics in the classes are also performed.     

Average-case complexity and decision problems in group theory

We investigate the average-case complexity of decision problems for finitely generated groups, in particular, the word and membership problems. Using our recent results on “generic-case complexity”, we show that if a finitely generated group G has word problem solvable in subexponential time and has a subgroup of finite index which possesses a non-elementary word-hyperbolic quotient group, then the average-case complexity of the word problem of G is linear time, uniformly with respect to the collection of all length-invariant measures on G. This results applies to many of the groups usually studied in geometric group theory: for example, all braid groups B_n, all groups of hyperbolic knots, many Coxeter groups and all Artin groups of extra-large type.     

Worst case analysis of Max-Regret,Greedy and other heuristics for Multidimensional Assignment and Traveling Salesman Problems

Optimization heuristics are often compared with each other to determine which one performs best by means of worst-case performance ratio reflecting the quality of returned solution in the worst case. The domination number is a complement parameter indicating the quality of the heuristic in hand by determining how many feasible solutions are dominated by the heuristic solution. We prove that the Max-Regret heuristic introduced by Balas and Saltzman (Oper. Res. 39:150–161, 1991) finds the unique worst possible solution for some instances of the s-dimensional (s≥3) assignment and asymmetric traveling salesman problems of each possible size. We show that the Triple Interchange heuristic (for s=3) also introduced by Balas and Saltzman and two new heuristics (Part and Recursive Opt Matching) have factorial domination numbers for the s-dimensional (s≥3) assignment problem.     

Worst-case analysis of the WSPT and MWSPT rules for single machine scheduling with one planned setup period

In this article, we consider the single machine scheduling problem with one planned setup period, with the aim of minimizing the weighted sum of the completion times. We study the WSPT and MWSPT heuristics and we show that the worst-case performance ratio is 3 for the two heuristics in some cases and it is unbounded otherwise. We also show that these worst-case performance ratios are tight.     

Quadri-tilings of the Plane

We introduce quadri-tilings and show that they are in bijection with dimer models on a family of graphs R ^* arising from rhombus tilings. Using two height functions, we interpret a sub-family of all quadri-tilings, called triangular quadri-tilings, as an interface model in dimension 2+2. Assigning “critical" weights to edges of R ^*, we prove an explicit expression, only depending on the local geometry of the graph R ^*, for the minimal free energy per fundamental domain Gibbs measure; this solves a conjecture of Kenyon (Invent Math 150:409–439, 2002). We also show that when edges of R ^* are asymptotically far apart, the probability of their occurrence only depends on this set of edges. Finally, we give an expression for a Gibbs measure on the set of all triangular quadri-tilings whose marginals are the above Gibbs measures, and conjecture it to be that of minimal free energy per fundamental domain.     

Flood search under the California Split rule

We consider flood search on a line and show that no algorithm can achieve an average-case competitive ratio of less than 4 when compared to the optimal off-line algorithm. We also demonstrate that the optimal scanning sequences are described by simple recursive relationships that yield surprisingly complex behavior related to Hamiltonian chaos.     

Scaling limit of isoradial dimer models and the case of triangular quadri-tilings

We consider dimer models on planar graphs which are bipartite, periodic and satisfy a geometric condition called isoradiality, defined in [R. Kenyon, The Laplacian and Dirac operators on critical planar graphs, Invent. Math. 150 (2) (2002) 409–439]. We show that the scaling limit of the height function of any such dimer model is a Gaussian free field. Triangular quadri-tilings were introduced in [B. de Tilière, Quadri-tilings of the plane, math.PR/0403324, Probab. Theory Related Fields, in press]; they are dimer models on a family of isoradial graphs arising from rhombus tilings. By means of two height functions, they can be interpreted as random interfaces in dimension 2+2. We show that the scaling limit of each of the two height functions is a Gaussian free field, and that the two Gaussian free fields are independent.     

Maximum stable set formulations and heuristics based on continuous optimization

 The stability number α(G) for a given graph G is the size of a maximum stable set in G. The Lovász theta number provides an upper bound on α(G) and can be computed in polynomial time as the optimal value of the Lovász semidefinite program. In this paper, we show that restricting the matrix variable in the Lovász semidefinite program to be rank-one and rank-two, respectively, yields a pair of continuous, nonlinear optimization problems each having the global optimal value α(G). We propose heuristics for obtaining large stable sets in G based on these new formulations and present computational results indicating the effectiveness of the heuristics. Received: December 13, 2000 / Accepted: September 3, 2002 Published online: December 19, 2002 RID="★" ID="★" Computational results reported in this paper were obtained on an SGI Origin2000 computer at Rice University acquired in part with support from NSF Grant DMS-9872009. Key Words. maximum stable set – maximum clique – minimum vertex cover – semidefinite program – semidefinite relaxation – continuous optimization heuristics – nonlinear programming Mathematics Subject Classification (2000): 90C06, 90C27, 90C30     

The projected newton method for solving inverse eigenvalue problems - the case of multiple eigenvaues

This paper deals with the optimal solution of ill-posed linear problems, i.e..linear problems for which the solution operator is unbounded. We consider worst-case ar,and averagecase settings. Our main result is that algorithms having finite error (for a given setting) exist if and only if the solution operator is bounded (in that setting). In the worst-case setting, this means that there is no algorithm for solving ill-posed problems having finite error. In the average-case setting, this means that algorithms having finite error exist if and only lf the solution operator is bounded on the average. If the solution operator is bounded on the average, we find average-case optimal information of cardinality n and optimal algorithms using this information, and show that the average error of these algorithms tends to zero as n→∞. These results are then used to determine the [euro]-complexity, i.e., the minimal costof finding an [euro]-accurate approximation. In the worst-case setting, the [euro]comp1exity of an illposed problem is infinite for all [euro]>0; that is, we cannot find an approximation having finite error and finite cost. In the average-case setting, the [euro]-complexity of an ill-posed problem is infinite for all [euro]>0 iff the solution operator is not bounded on the average, moreover, if the the solutionoperator is bounded on the average, then the [euro]-complexity is finite for all [euro]>0.     

Heuristic optimisation in financial modelling

There is a large number of optimisation problems in theoretical and applied finance that are difficult to solve as they exhibit multiple local optima or are not ‘well-behaved’ in other ways (e.g., discontinuities in the objective function). One way to deal with such problems is to adjust and to simplify them, for instance by dropping constraints, until they can be solved with standard numerical methods. We argue that an alternative approach is the application of optimisation heuristics like Simulated Annealing or Genetic Algorithms. These methods have been shown to be capable of handling non-convex optimisation problems with all kinds of constraints. To motivate the use of such techniques in finance, we present several actual problems where classical methods fail. Next, several well-known heuristic techniques that may be deployed in such cases are described. Since such presentations are quite general, we then describe in some detail how a particular problem, portfolio selection, can be tackled by a particular heuristic method, Threshold Accepting. Finally, the stochastics of the solutions obtained from heuristics are discussed. We show, again for the example from portfolio selection, how this random character of the solutions can be exploited to inform the distribution of computations.     

Crossing minimization in weighted bipartite graphs

Given a bipartite graph G=(L₀,L₁,E) and a fixed ordering of the nodes in L₀, the problem of finding an ordering of the nodes in L₁ that minimizes the number of crossings has received much attention in literature. The problem is NP-complete in general and several practically efficient heuristics and polynomial-time algorithms with a constant approximation ratio have been suggested. We generalize the problem and consider the version where the edges have nonnegative weights. Although this problem is more general and finds specific applications in automatic graph layout problems similar to those of the unweighted case, it has not received as much attention. We provide a new technique that efficiently approximates a solution to this more general problem within a constant approximation ratio of 3. In addition we provide appropriate generalizations of some common heuristics usually employed for the unweighted case and compare their performances.     

Inequalities for the first‐fit chromatic number

The First‐Fit (or Grundy) chromatic number of G, written as χ_FF(G), is defined as the maximum number of classes in an ordered partition of V(G) into independent sets so that each vertex has a neighbor in each set earlier than its own. The well‐known Nordhaus‐‐Gaddum inequality states that the sum of the ordinary chromatic numbers of an n‐vertex graph and its complement is at most n + 1. Zaker suggested finding the analogous inequality for the First‐Fit chromatic number. We show for n ≥ 10 that ?(5n + 2)/4? is an upper bound, and this is sharp. We extend the problem for multicolorings as well and prove asymptotic results for infinitely many cases. We also show that the smallest order of C₄‐free bipartite graphs with χ_FF(G) = k is asymptotically 2k² (the upper bound answers a problem of Zaker [9]). © 2008 Wiley Periodicals, Inc. J Graph Theory 59: 75–88, 2008     

The average-case analysis of some on-line algorithms for bin packing

In this paper we give tighter bounds than were previously known for the performance of the bin packing algorithms Best Fit and First Fit when the inputs are uniformly distributed on [0, 1]. We also give a general lower bound for the performance of any on-line bin packing algorithm on this distribution. We prove these results by analyzing optimal matchings on points randomly distributed in a unit square. We give a new lower bound for the up-right matching problem. A preliminary version of this paper appeared inProc. 25th IEEE Symposium on Foundations of Computer Science, 193–200. This research was done while the author was at the Massachusetts Institute of Technology Dept. of Mathematics and was supported by a NSF Graduate Fellowship and by Air Force grant OSR-82-0326.     

Spreadsheet Implementable Inventory Control for a Distribution Center

This paper develops and tests a simple procedure for establishing stocking rules for a multi-component distribution center that supplies spare parts for an equipment maintenance operation. Our basic formulation seeks to minimize inventory investment subject to constraints on average service level and replenishment frequency. We simplify this formulation by classifying parts according to a new ABC methodology and applying heuristics to the classical (Q,r) model that lead to closed-form expressions for the stocking parameters. Our numerical results show that: (1) the proposed ABC scheme does not introduce large errors provided that it is done in a manner that reflects the key parameters in the model, and (2) any of a number of simple reorder point heuristics can provide the basis for an effective spreadsheet implementable system for controlling inventory in a complex multi-component environment as long as the service level is checked against the exact formula.     

Greedy heuristics for the diameter-constrained minimum spanning tree problem

The diameter-constrained minimum spanning tree problem is an NP-hard combinatorial optimization problem that seeks a minimum cost spanning tree with a limit D imposed upon the length of any path in the tree. We begin by presenting four constructive greedy heuristics and, secondly, we present some second-order heuristics, performing some improvements on feasible solutions, hopefully leading to better objective function values. We present a heuristic with an edge exchange mechanism, another that transforms a feasible spanning tree solution into a feasible diameter-constrained spanning tree solution, and finally another with a repetitive mechanism. Computational results show that repetitive heuristics can improve considerably over the results of the greedy constructive heuristics, but using a huge amount of computation time. To obtain computational results, we use instances of the problem corresponding to complete graphs with a number of nodes between 20 and 60 and with the value of D varying between 4 and 9.     

Searching for a counterfeit coin with b-balance

We consider the classical problem of searching for a heavier coin in a set of n coins, n-1 of which have the same weight. The weighing device is b-balance which is the generalization of two-arms balance. The minimum numbers of weighings are determined exactly for worst-case sequential algorithm, average-case sequential algorithm, worst-case predetermined algorithm, average-case predetermined algorithm.We also investigate the above search model with additional constraint: each weighing is only allowed to use the coins that are still in doubt. We present a worst-case optimal sequential algorithm and an average-case optimal sequential algorithm requiring the minimum numbers of weighings.     

Semidefinite programming relaxations for graph coloring and maximal clique problems

The semidefinite programming formulation of the Lovász theta number does not only give one of the best polynomial simultaneous bounds on the chromatic number χ(G) or the clique number ω(G) of a graph, but also leads to heuristics for graph coloring and extracting large cliques. This semidefinite programming formulation can be tightened toward either χ(G) or ω(G) by adding several types of cutting planes. We explore several such strengthenings, and show that some of them can be computed with the same effort as the theta number. We also investigate computational simplifications for graphs with rich automorphism groups. Partial support by the EU project Algorithmic Discrete Optimization (ADONET), MRTN-CT-2003-504438, is gratefully acknowledged.     

The probabilistic analysis of a greedy satisfiability algorithm

On input a random 3‐CNF formula of clauses‐to‐variables ratio r₃ applies repeatedly the following simple heuristic: Set to True a literal that appears in the maximum number of clauses, irrespective of their size and the number of occurrences of the negation of the literal (ties are broken randomly; 1‐clauses when they appear get priority). We prove that for r₃ < 3.42 this heuristic succeeds with probability asymptotically bounded away from zero. Previously, heuristics of increasing sophistication were shown to succeed for r₃ < 3.26. We improve up to r₃ < 3.52 by further exploiting the degree of the negation of the evaluated to True literal. © 2005 Wiley Periodicals, Inc. Random Struct. Alg., 2006