J. Edmonds. Matroids and the Greedy Algorithm. Mathematical Programming, 1:125--136, 1971.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....base of (X; I) It is known that this problem is related to matroids through the following greedy algorithm. More precisely speaking, the following greedy algorithm returns an optimal solution for any non negative weight vector w 2 R if and only if the simplicial complex (X; I) is a matroid [8, 16]. Algorithm: Greedy Algorithm Input: a simplicial complex (X; I) and a non negative weight vector w 2 R ; Step 1: sort X = f1; ng so that w[1] w[2] Delta Delta Delta w[n] Step 2: S ; Step 3: for i = 1 to n do Step 3 1: if S [ fig 2 I then S S [ fig; end of for Step ....

J. Edmonds, Matroids and the greedy algorithm, Mathematical Programming 1, 1971, 127--136. 8

....equal to the value of w(A) and the value of (3.4) is the same as the value of the Held Karp heuristic. It will follow that the optimal value of the asymmetric Subtour LP is equal to the value of the Held Karp heuristic if the polytope Bx d has integer extreme points. By a theorem of Edmonds [7], Bx d has integer extreme points because it represents the intersection of two matroids (the 1 tree matroid and the indegree 1 mattold) The choice of which constraints from the asymmetric Subtour LP to assign to Ax b and which to assign to Bx d was somewhat arbitrary, so it turns out ....

....1 constraints for Bx d. Theorem 3.1.4 follows from picking the subtour elimination constraints plus the additional redundant constraint i,j Xij : n for Bx d, while Theorem 3.1.5 comes from picking the indegree I and outdegree 1 constraints for Bx d. In each case, the results of Edmonds [7] guarantee that Bx d has integer extreme points. Theorem 3.1.5 is of special interest, since some non constructive approximation algorithms for the asymmetric TSP are based on the assignment problem. Several researchers have used the assignment problem with no edge weighting as a lower bound ....

J. Edmonds (1971). Matroids and the greedy algorithm. Math. Programming 1, 127-136.

....by Lawler [8] or Papadimitriou and Steiglitz [9] A natural question, precisely posed below, is the following. For which optimization problems does the greedy algorithm give the correct solution. In a sense this question is answered by a classical theorem in matroid theory due to Rado and Edmonds [4]. In the matroid case, the greedy algorithm always solves the optimization problem. That is, the greedy algorithm solves the optimization problem for every linear objective function. There are situations, however, for which the greedy algorithm works for many but not all linear objective ....

J. Edmonds, Matroids and the greedy algorithm, Mathematical Programming, 1 (1971), 127-136.

....we associate with the objective function c by using the lexicographic order in order to break ties. The convex hull of all incidence vectors of independent sets in the matroid is described by the non negativity inequalities and the set of all rank inequalities of the type x(F ) rank(F ) for F E [Ed71]. For the rank inequality x(F ) rank(F ) a basis of fx 2 IR n : x(F ) 0g is given by the vectors e i , i 2 E n F and by vectors of the type e i Gamma e j for i; j 2 F . For an inequality x i 0, a basis of fx 2 IR n : x i = 0g is given by the vectors e j , j 2 E n fig. This fact is ....

J. Edmonds, Matroids and the greedy algorithm, Mathematical Programming, 1, 1971, 127--136. 9

....conclusions will be more accurate after the data switch has been done because each filtered data item is a representative for some well known subtables . Our algorithm in an ordinary form dates back to Mullat (1971) At first glance, the ordinary form seems similar to the greedy heuristic (Edmonds 1971), but this is not the case. The starting point for the ordinary version of the algorithm is the entire table from which the elements are removed. Instead, the greedy heuristic starts with the empty set, and the elements are added until some criterion for stopping is fulfilled. However, the ....

J. Edmonds, Matroids and the Greedy Algorithm, Math. Progr., No. 1 (1971) 127-136.

....should be combined into one. Then, based on the set of center nodes in each subset of documents, potential representatives can be identified. The set of potential representatives is the minimal set of center nodes that points to the largest number of documents in the subset. Greedy algorithm [6]isemployed that the center node with the largest number of out links to the documents in its subset is selected. If all the nodes that a center node points to are being pointed by the set of nodes in the potential representatives, the node is ignored from the set of potential representaitives. 3 ....

Jack Edmonds. Matroids and the greedy algorithm. In Mathematical Programming, 1:126-136, 1971.

....process, but for such problems there does not seem to be any advantage in doing so. Thirdly, there may be other ways to express the generation of solutions without using a limit operator. Suitable frameworks include relational catamorphisms (folds) and anamorphisms (unfolds) 9] matroids [3], and greedoids [6, 7, 8] and there is much well developed theory about these structures to assist the programmer. Such theory can be of great assistance in helping the programmer to develop an algorithm once the problem at hand has been fitted into the structure. 4.2 Advantages of lim One of ....

Jack Edmonds. Matroids and the greedy algorithm. Mathematical Programming, 1:126--136, 1971.

....of new lower bounds and the discovery (and verification) of new combinatorial search strategies. 1 Introduction Many formalisms have been proposed over the years to capture combinatorial optimization algorithms such as dynamic programming, branch and bound, and greedy (see, for example, [6, 13, 16, 17, 18, 19]) These models tend be relatively narrow in that each captures one specific class of solution, and hence provides neither conceptual unity nor a common framework in which the techniques can be studied. In 1989 Helman [9] presented a common formalism that captures dynamic programming and branch ....

Edmonds, J., "Matroids and the greedy algorithm", Math. Programming 1, 1971, 127--136.

....optimizes all bottleneck functions, structures which are less constrained than matroid embeddings. 1 Introduction Obtaining an exact characterization of the class of problems for which the greedy algorithm returns an optimal solution has been an open problem. Rado [9] Gale [3] and Edmonds [1] independently showed that matroids characterize a subclass of problems on which the greedy algorithm always optimizes linear objectives; their results are limited by the assumption that the greedy algorithm operates on a hereditary set system, whereas most common greedy algorithms operate on set ....

....axiom is often phrased more generally: exchange axiom) If X;Y 2 C and jY j jX j, then 9x 2 X Gamma Y such that Y [ fxg 2 C. In the presence of the trivial axiom, the exchange axiom is equivalent to the combination of the accessibility and augmentation axioms. Rado [9] Gale [3] and Edmonds [1] independently proved that the best in greedy algorithm optimizes all linear objective functions over a hereditary set system (S; C) if and only if (S; C) is a matroid. Korte and Lovasz [6; 7] defined greedoids and proved that the best in greedy algorithm optimizes all linear objective functions ....

[Article contains additional citation context not shown here]

J. Edmonds, "Matroids and the greedy algorithm," Math. Programming 1 (1971), pp. 127--136.

....Fulkerson, and Johnson [6] for the traveling salesman problem) Notice that if y s = 1, for all s # S, then (2) reduces to a SEC. The set of feasible solutions for (5) corresponds to the set of all trees of G. The above formulation can be seen as a generalization of the spanning tree polytope [8], by noting that if the vertices that appear in an optimal (minimum weight) tree are given, then the PCSPG reduces to finding a minimum spanning tree (MST) of the subgraph of G induced by those vertices. 3. Solving the linear programming relaxation A linear programming (LP) relaxation for (5) ....

J. Edmonds. Matroids and the greedy algorithm. Mathematical Programming, 1:127--136, 1971.

....the elements of N so that w i w i 1 ; i = 1; 2; Delta Delta Delta ; n Gamma 1. Let T = OE; i = 1. 1. If w i 0 or i n, stop T is optimal, i.e. x j = 1 for j 2 T and x j = 0 for j = 2 T. If w i 0 and T [ fig 2 F, add element i to T. 2. Increment i by 1 and return to step 1. Edmonds [40, 41] derived a complete description of the matroid polytope, the convex hull of the characteristic vectors of independent sets of a matroid. While this description has a large (exponential) number of constraints, it permits the treatment of linear optimization problems on independent sets of matroids ....

J.Edmonds, Matroids and the greedy algorithm, Mathematical Programming, (1971) 127-136.

....theorem with any basis of U , i.e. a subset fu i 1 ; Delta Delta Delta ; u i d g of U which is a basis of L, an upper bound ffi(P ) P d j=1 k(P; u i j ) is obtained. Appealing to the well known validity of the so called greedy algorithm for finding a minimal weight basis of a matroid ( 5] [2]) we find that the basis of U yielding the best upper bound is the lexicographically first one, i.e. the basis fu i 1 ; Delta Delta Delta ; u i d g, where i 1 Delta Delta Delta i d and, for j = 1; Delta Delta Delta ; d, the index i j is the smallest such that u i j is nonzero and ....

J. Edmonds, "Matroids and the greedy algorithm", Math. Prog. 1 (1971), 127-136.

....has been given in the literature to greedy structures. One mathematical structure which can model several greedy algorithms is that of a matroid. A matroid is a hereditary set system with an exchange property (the matroid property) These were first thought of in 1935 by Whitney [99] Edmonds in [29] first linked matroids to greedy algorithms. However matroids do not include all greedy structures, and not every matroid is a greedy structure, and for the specific purpose of getting closer to characterizing greedy structures, greedoids were introduced by Korte and Lov asz [55, 57] These are a ....

Jack Edmonds. Matroids and the greedy algorithm. Mathematical Programming, 1:126-- 136, 1971.

....f(S) f(T ) for S T Nn , ii) f is submodular, i.e. f(S [ T ) f(S T ) f(S) f(T ) for all S; T Nn , and (iii) f is nonnegative. From (4) we see that P (q; n) fx 2 R n j x(S) f(S) for all S Nng, so this is the polymatroid associated with the submodular function f. Edmonds (see [3]) showed that maximizing a nonnegative linear objective function over a polymatroid can be done using the greedy algorithm. Thus, in the case k = n, we can solve the problem (3) by the following greedy algorithm (assume for simplicity that c1 : cn 0) x1 = f(f1g) q1 , x2 = f(f1; 2g) ....

Edmonds, J., Matroids and the greedy algorithm, Mathematical Programming 1 (1971) 127-136.

....n is in P (q; k) if and only if v 1:j q 1:j for all 1 j k. Each n majorization polyhedron may be viewed as a polymatroid (see e.g. 4] 5] associated with the set function f(S) P r j=1 q j for each S N n where r : jSj. Trivially, this function is monotone and submodular. Thus (see [3]) 3) may be solved by the greedy algorithm and the optimal solution (when c 1 : c n 0) is x = q. This result also follows easily from (1) Some further properties in the case k = n are discussed in [2] For k n, however, P (q; k) may not be a polymatroid and therefore the greedy ....

Edmonds, J., Matroids and the greedy algorithm, Mathematical Programming 1 (1971) 127-136.

....in P (q; k) if and only if v 1:j # q 1:j for all 1 # j # k. Each n majorization polyhedron may be viewed as a polymatroid (see e.g. 4] 5] associated with the set function f(S) P r j=1 q j for each S # N n where r : S . Trivially, this function is monotone and submodular. Thus (see [3]) 3) may be solved by the greedy algorithm and the optimal solution (when c 1 # . # c n # 0) is x = q. This result also follows easily from (1) Some further properties in the case k = n are discussed in [2] For k n, however, P (q; k) may not be a polymatroid and therefore the greedy ....

Edmonds, J., Matroids and the greedy algorithm, Mathematical Programming 1 (1971) 127-136.

No context found.

J. Edmonds. Matroids and the Greedy Algorithm. Mathematical Programming, 1:125--136, 1971.

No context found.

J. Edmonds, "Matroids and the greedy algorithm". Mathematical Programming, Vol. 1, 1971, pp. 127-136. Current approach Lower bound 38,30 38,34 0,0 38,32 GA Current approach Lower bound Average compaction,% 50,0 50,5 51,0 Lower bound Current approach GA 0,0 43,5 44,0 GATTO tests HITEC tests SYMBAT tests

No context found.

Edmonds, J.: Matroids and the greedy algorithm. Math. Program. 1, 127--136 (1971)

No context found.

Jack Edmonds. Matroids and the greedy algorithm. Mathematical Programming, 1:127--136, 1971.

No context found.

J. Edmonds (1971) Matroids and the greedy algorithm. Mathematical Programming 1: 127136.

No context found.

J. Edmonds, Matroids and the greedy algorithm, Mathematical Programming 1, 1971, 127--136.

No context found.

J. Edmonds, Matroids and the greedy algorithm, Mathematical Programming 1, 1971, 127--136.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC