Two-Commodity Multiroute Maximum Flow Problem

Size: px
Start display at page:

Download "Two-Commodity Multiroute Maximum Flow Problem"

Transcription

1 Two-Commodity Multiroute Maximum Flow roblem Donglei Du R. Chandrasekaran August, 005 Abstract We consider a two-commodity multiroute maximum flow problem in an undirected network a generalization of the standard two-commodity maximum flow problem. An efficient combinatorial algorithm, which always guarantees a quarter-integer solution when the capacities are integers, is devised to solve a special case based on a novel extension of the augmentingpath technique. artial results are reported and the difficulties in applying the augmenting-path technique to the general case are explained. Keywords: augmenting-path, multiroute flow, two-commodity Corresponding Author: Faculty of Administration, University of New Brunswick,.O. Box 00, Fredericton, NB, Canada, E3B 5A3, ddu@unb.ca. Research supported in part by NSERC grant Department of Computer Sciences, University of Texas at Dallas, Richardson, TX 75083, U.S.A., chandra@utdallas.edu.

2 Introduction The roblem roviding high-level survivability is critical for computer, telecommunication, transportation and other networks. Traditional network flow models are vulnerable to arc failures because flows are routed along single paths. There has been recent interest in extending the traditional network flow model to a multiroute version that can survive arc failures up to a certain level [, 3,, 6, 7, 3, ]. Consider a network with specified source-destination pairs. In the traditional flow model, flow is sent along single paths between given source-destination pairs, and lost whenever arcs on the path fail. This routing method is vulnerable to even one arc failure. The multiroute flow model differs from the traditional one in that flow must to be routed along a multiroute channel that is, a channel that consists of m arc-disjoint paths rather than a single path from the source to the destination, where m is any given positive integer. Multiroute channels naturally provide protection against arc failures up to m. Existing research has focused on the one-commodity multiroute maximum flow problem [, 3,, 6, 7, 3, ]. The major concern of this research, however, is a two-commodity multiroute maximum flow problem, to be formally defined after some notations. For any finite set S, let S denote the cardinality of S, and let R S + denote the set of non-negative real column vectors with dimension S. Definition. Two-commodity multiroute maximum flow problem (m,m )-MF: Suppose G = (V,E) is an undirected network with vertex set V, edge set E, and edge capacity vector u R E +. There are two pairs of source-destination vertices {s,t } and {s,t } in V, and two positive integers m,m Z +. () An elementary m q -route (q =, ) from s q to t q consists of a set of m q arc-disjoint s q -t q paths in G. An edge (i,j) E is contained in an elementary m q -route if (i,j) is contained in some s q -t q path therein. () An m q -route flow (q =, ) from s q to t q is an assignment of non-negative weights to elementary m q -routes. The value of the an m q -route flow is the sum of the weights assigned.

3 (3) An (m,m )-route flow consists of an m - and an m -route flow, where the sum of weights assigned to the elementary m - and m -routes containing edge (i,j) is no more than its capacity u ij. The value of an (m,m )-route flow is the sum of the values of the m - and m -route flows. The problem is to find an (m,m )-route flow of maximum value. Let q (q =, ) denote the set of all elementary m q -routes from s q to t q. Let q ij (q =, ) denote the set of all elementary m q -routes from s q and t q containing edge (i,j) E. If weight x q p (q =, ) is assigned to each p q, then (m,m )-MF can be formulated as a linear program: max p x p + p x p () p ij x p + p ij x u ij, (i,j) E; () x p 0, p ; (3) x p 0, p. () The above linear program is solvable in polynomial time based on the Ellipsoid Algorithm [0] although not an efficient one. The idea is to provide a polynomial-time subroutine that solves the separation problem of the linear program that, given a solution, either proves the solution is feasible or finds a violated constraint. Herein, we solve the dual program of ()-(). The separation problem of the dual program is the minimum cost m q -route problem (q =, ), where the cost of each arc is the dual variable associated with that arc. We thus need verify only whether the minimum cost m q -route between s q and t q is at least. The problem is solvable in both weakly and strongly polynomial time; it can be viewed as a unit-capacity minimum cost flow problem, for which many efficient algorithms are available (Chapters 9- of []). In ()-() above, we define flows on the elementary m - and m -routes using the route-flow form. In traditional network flow models, flows can also be defined using the edge-flow form. An advantage of edge-flow over route-flow form is that its specification involves only a polynomial

4 number of variables, thus most existing algorithms are based on the edge-flow form. The same is true for the multiroute version that we have proposed. Definition. Consider q {, }. () An m q -route edge-flow of value F q /m q is given by vector f q R E +, satisfying flow conservation constraints (5), capacity constraints (6), and extra capacity constraints (7): f q ij = j V F q, i = s; 0, i s, t; F q, i = t. (5) 0 f q ij u ij (i,j) E; (6) 0 f q ij F q m q, (i,j) E. (7) () An (m,m )-route edge-flow (f,f ) of value F /m + F /m is a pair of m -route edgeflow f of value F /m and m -route edge-flow f of value F /m. Here all flows are algebraic meaning positive value for f q ij implies the q commodity flow in the direction from i to j and negative flow in the reverse direction. Given any m q -route flow x q R q + (q =, ), an m q -route edge-flow f q R E + with the same value is obtained by taking f q ij = p q x q p. The other direction and proof, first established ij by Kishimoto [3], is much more involved: Theorem.3 (Kishimoto [3]) Given any m q -route edge-flow f q R E + of value F q /m q (q =, ), we can construct an m q -route path-flow with the same value. An alternative proof of the above result is given by Aggarwal and Orlin []. The equivalence of edge-flow and route-flow implies that (m,m )-MF can also be modeled 3

5 as a linear program in the edge-flow form: max F + F (8) m m f q ij = j V F q, i = s; 0, i s, t; q =, ; F q, i = t. f ij + f ij u ij (i,j) E; (0) f q ij F q m q (i,j) E and q =,. () The above linear program is of polynomial size and the coefficients of the constraint matrix are bounded by m and m, therefore it is solvable in strongly polynomial time using the technique of Tardos [0]. The algorithm is inefficient, however, because of its poor worst-case complexity. We have devised an efficient, strongly polynomial, combinatorial algorithm detailed below. Related Work Given the lack of research on multiroute flow problems of more than one commodity, we review results for one-commodity cases. The one-commodity multiroute maximum flow problem is introduced by Kishimoto and Takeuchi [5], and Kishimoto et al [7], who have provided a strongly polynomial combinatorial algorithm that solves a sequence of the ordinary maximum flow problems and implies a max-flow/min-cut relationship. Aggarwal and Orlin [], who have given an improved strongly polynomial combinatorial primal-dual algorithm for the same problem, simplified and extended Kishimoto and Takeuchi s theory. They also propose a weakly polynomial algorithm based on binary search, which results in a number of O(log( V U max )), where U max = max (i,j) E u ij. Both Kishimoto s and Aggarwal and Orlin s algorithms are based on reducing the multiroute maximum flow problem to a sequence of ordinary maximum flow problems by parameterization. This is not coincidental: Du and Chandrasekaran [6] show that both algorithms can be unified into a common framework based on Newton s method. They [6, 7] also devise an augmenting-path algorithm by exploiting the combinatorial properties of the problem, the foundation for the algorithm presented here. We note that (m,m )-MF is also related to the standard two-commodity maximum flow problem (,)-MF using our notation. Hu [] first proposes an augmenting-path algorithm, obtaining the following results: () Extremal roperty: there always exists an optimal solution (9)

6 such that one of the two commodities is maximized; () Half-integer roperty: if the capacities are integers, there always exists a half-integer optimal solution; and (3) Max-flow/min-cut roperty: maximum flow equals minimum cut. The main idea of Hu s algorithm is to fix the optimal value F of the first commodity so as to improve the value F of the second commodity without changing F. The edge flows may change, of course, when the first commodity is recirculated. Itai [] presents a strongly polynomial implementation of Hu s algorithm by using Dinic s layer network idea [5]. Sakarovitch [8] has proved that the problem is also solvable by applying transforming variables to the standard single-commodity maximum flow problem. New Contribution The general (m,m )-MF is difficult to attack for the purpose of efficient combinatorial algorithms. In this research, we focus on the special (, )-MF (m =,m = ). An immediate idea is to use the parameterization technique of the single-commodity multiroute maximum flow problem [, 6, 3]. This idea, however, fails here, because it leads to solving a twocommodity flow problem with both sum (0) and individual capacity constraints (), for which no efficient combinatorial algorithm exists. Instead we extend the augmenting-path algorithms of Hu [] and Du and Chandrasekaran [6, 7] to solve (, )-MF. Our algorithm always results in a quarter-integer solution when the capacity function is an integer, but the analysis of our algorithm is challenging because of the lack of an explicit max-flow/min-cut relationship. Finally, we give partial results on (, )-MF (m = m = ), a direct extension of (, )-MF, and explain why the augmenting-path technique fails for this problem. The rest of our discussion is organized as follows. In Section, we solve (, )-MF by presenting an augmenting-path algorithm. In Section 3, we deal with (, )-MF and point out the difficulties in extending the augmenting-path technique to the problem herein. We conclude in Section. 5

7 (, )-MF In this section, a new augmenting-path algorithm is developed to solve (, )-MF in an undirected network. Our strategy generalizes the augmenting-path algorithms for the standard twocommodity (, )-MF [] and for the single-commodity multiroute maximum flow problem [6, 7]. The balance of discussion focuses on edge-flows, so from now on we refer to any multiroute edge-flow as m -route flow, m -route flow, or (m,m )-route flow for any positive integers m and m, which should not in context pose confusion. In Section., we first prove that Hu s extremal property holds for (, )-MF, paving the way for our augmenting-path type algorithm. We describe and analyze our algorithm in Section.; we illustrate the algorithm by example in Section.3.. Extremal roperty Lemma. There exists an optimal (, )-route flow (f,f ) with values ( F, F /) for (, )- MF such that F = F, where F is the maximum value of the first flow when the second is ignored. roof: If not, suppose F < F, and that after ignoring the second commodity, there is an augmenting-path from s to t. Along this path, we can increase a certain amount of -route flow, but cannot locate a path with ( F, F /) because of the presence of a second commodity on some arcs of the path. Let e = (a,b) and e = (c,d) be the first and last such arcs respectively. Case. There is one s -t path passing through both e and e carrying second commodity. Because each unit flow of the second commodity is sent along -arc-disjoint-path, there exists an arc-disjoint s -t path which carries the same amount of -route flow; see Figure (a) for an illustration, where : s a e b c e d t is the dotted line; : s t is the dashed line. Reducing appropriate δ units of the -route flow on each of and allows sending δ units 6

8 s s s a e e b c d t s ' ' e e a b c d t ' t t (a) Case (b) Case Figure : Extremal property for (, )-MF of -route flow along the following path: s a e b c e d t. Case. There are two s -t paths and passing through e = (a,b) and e = (c,d), respectively. We have, in like manner, arc-disjoint paths and ; see Figure (b) for an illustration, where : s e a b t and : s t are the dotted lines; : s e c d t and : s t are the dashed lines. Notice that and cannot be coupled in every (, )-route flow; otherwise, we can adjust the flow along the following paths to increase the total value of F + F /, contradicting the previously held assumption that ( F, F /) is the optimal value. Therefore reducing an appropriate number of δ units of the -route flow on each of,, and permits sending δ units -route flow along the following paths. s a s e c d t ; s a e b t d t. This finishes the proof. Figure shows that this result is not symmetrical. It is therefore crucial to begin by maximizing the first commodity in Lemma.. As is evident, maximizing the second commodity prevents - route flow from being sent: we get 3/ although the optimal value is. 7

9 s s t t Figure : Asymmetry of the extremal property for (, )-MF. Main Algorithm and Analysis First, some preliminary definitions and notations: Definition. Consider any feasible (, )-route flow f = (f,f ) with values ( F, F /), where f is a -route flow with value F for the first commodity and f is a -route flow with value F / for the second commodity. () A forward (directed) graph G + (f) = (V,A + (f)) is defined as follows: arc (i,j) A + (f) if and only if δ + ij = u ij f ij f ij > 0. A backward (directed) graph G (f) = (V,A (f)) is defined analogously: arc (i,j) A (f) if and only if δ ij = u ij + f ij f ij > 0. () An arc (i,j) is tight if f ij = F /. (3) A -path from s to t consists of two s -t paths, not necessarily arc-disjoint. An arc on the -path is called a double arc if it is a shared arc; a single arc otherwise. The shortest -path from s to t contains the fewest arcs, where each double arc is counted twice. See Figure 3 for an illustration of a typical -path. () A bridge of the forward and backward graphs is an arc, the deletion of which leaves no directed s -t path. (5) For any directed graph G = (V,A), and an arc set H A, denote G\H as the graph obtained from G by deleting all the arcs in H. 8

10 s i t (b) j s i j t (a) s i t j Figure 3: Illustration of -path: (a) is a -path, containing two single paths (b) and (c), where (i,j) is a double arc. (c) The following simple fact for the forward and backward graphs is useful in the flow-updating process at Step of algorithm AUG later. roposition.3 Suppose we want to direct a flow value pattern of (0, δ) that is, send δ of second flow without changing the value of the first. Let a be a rational number within the interval [, ].. Each arc in the forward graph admits a flow value pattern of (aδ,aδ), allowing aδ first and second flows to be sent simultaneously from s to t and s to t respectively.. Each arc in the backward graph admits a flow value pattern ( aδ,aδ), allowing aδ first and second flows to be sent simultaneously from t to s and s to t respectively.. We now state the augmenting-path algorithm AUG. Algorithm AUG Input. An instance of the (, )-MF Output. An optimal (, )-route flow Step. Ignoring the second commodity, find a maximum -route flow f with value F. Use any available algorithm for the standard one-commodity maximum flow problem with capacity vector u R E + (e.g. []). 9

11 A S ( f ) (, ) A D ( f ) s t (, ) (a) Forward -path (, ) A S ( f ) A D ( f ) (, ) A S ( f ) (, ) s t (b) Backward -path (, ) A S ( f ) Figure : Illustration of augmenting scheme for (, )-MF Step. Fix the edge flows of the first commodity, and find a maximal -route flow ˆf with value ˆF /. Use any algorithm for the one-commodity -route maximum flow problem with capacity vector û = {u ij f ij } ij E R E + (e.g. [7]). Let f = ˆf and F = ˆF. Step 3. f = (f,f ) is a (, )-route flow with values ( F, F /). Step 3.. If there is no directed s -t path in either G + (f) or G (f), stop with an optimal (, )-route flow f; otherwise denote A T +(f) and A T (f) to be sets of tight bridges of G + (f) and G (f), respectively. If A T +(f) = A T (f) = (that is, no new tight bridges exist), go to Step 3.; otherwise let G + (f) := G + (f)\a T (f) and G (f) := G + (f)\a T (f) and proceed to Step 3.. Step 3.. If an elementary -route in G + (f) exists, find the shortest one. If, on the other hand, at least one bridge in G + (f) exists, find the shortest -path (not necessarily an arc-disjoint one). This process generates a forward -path in the forward graph. Repeat the same for the backward graph to generate a backward -path see Figure for an illustration. Let A +(f) and A (f) be the set of arcs on the forward and backward -paths; merge A +(f) and A (f) to form the augmenting scheme. Step. In the forward -path, A +(f) = A S +(f) A D +(f), where A S +(f) and A D +(f) are respectively the single and double arcs of A +(f) in the forward -path; analogously A (f) = A S (f) A D (f) in the backward -path. Further denote A S T + (f), A D T + (f), A S T (f), and A D T (f) as the tight arcs in A S +(f), A D +(f), A S (f), and A D (f) respectively see Figure for an illustration. Flows on any arc belonging to A +(f) and A (f) are updated using the 0

12 following formulas: (i,j) A S +(f) : fij := fij + δ, fij := fij + δ; () (i,j) A D +(f) : fij := fij + δ, fij := fij + δ; (3) (i,j) A S (f) : fij := fij δ, fij := fij + δ; () (i,j) A D (f) : fij := fij δ, fij := fij + δ; (5) F := F, F := F + δ. (6) In the above δ = min{δ,δ,δ 3,δ,δ 5,δ 6 } (7) where Go to step. δ = min ( (i,j) A S T + (f) ) A D T ( (f) A S T (f) ) A D T + (f) ( F δ = min (i,j) A S T + (f) A S T (f) f ij ( ) δ 3 = min uij f (i,j) A S + (f) ij fij ; ( ) δ = min uij f (i,j) A D + (f) ij fij ; ( ) δ 5 = min uij + f (i,j) A S (f) ij fij ; ( ) δ 6 = min uij + f (i,j) A D (f) ij fij. ) ; ( ) F f ij ; In the algorithm above, Step is self-explanatory. The optimal value of the -route flow can be fixed in the rest of the algorithm because of the extremal property (Lemma.). Step generally yields a maximal (locally optimal) solution for the second commodity unless the -route flow is altered. Step 3 is designed to locate an augmenting scheme by allowing adjustment of the -route flow. This step, if successful, will result in flow update in preparation for Step. Once the flow is updated, further -route augmenting may not require alteration of -route flow. Return to Step is all that s needed; the process is repeated until an optimal solution is found.

13 Elaboration of Steps 3. and 3. may be necessary. The purpose of Step 3. is locating a pair of forward and backward -paths in the forward and backward graphs; once identified, the -paths can be merged to form an augmenting scheme Figure. However, when the pair of -paths is merged, the same arc may occur in both, and we need to guarantee that a tight arc can occur in both -paths only when it is not a single arc in either of the -paths. Otherwise, there exists a tight arc (i,j) in both -paths and it is a double arc in at least one of the pair of -paths. Let the current flows be (f,f ) with values ( F, F /). To increase the flow value of the second commodity by δ, the flow on arc (i,j) is updated by either fij + 3δ, assuming (i,j) is single in one of the -paths or fij + δ, if in both. Because (i,j) is a tight arc, we have f ij = F /, suggesting that extra capacity constraints (7) cannot be guaranteed for the new flow. To find such a pair of -paths, Step 3. adopts an iterative process. First note that bridges in the forward or backward graph must be on every -path from s to t. With any iteration, we therefore update the forward (backward) graph by deleting all tight bridges of the backward (forward) graph. This updating may result in creation of new tight bridges in the graphs, leading to the next iteration. Iteration is terminated only when () there is no s -t path in either graphs in which case the algorithm stops with no augmenting scheme; or () no new tight bridges are created in which case the algorithm proceeds with Step 3. to find the augmenting scheme. Note that the number of iterations is bounded by E, the limit for the number of tight bridges in G + (f) and G (f). Step 3. s method resembles that of Du and Chandrasekaran [6, 7] for the one-commodity multiroute maximum flow problem. As previously remarked [6, 7], simply choosing the shortest -path in the forward or backward graph will not guarantee strongly polynomial implementation, unlike it does with the standard maximum flow problem [5, 8]. Du and Chandrasekaran [6, 7] suggest choosing the shortest elementary -route whenever possible. The idea also works for the two-commodity case (Theorem.7), as will be shown in the proof.

14 We show how finding the shortest -path in Step 3. is a matter of discovering the shortest elementary -route in a modified graph. All bridges in the forward graph must, obviously, be on every -path from s to t. arallel arcs may be used to duplicate bridges for the purpose of constructing a new forward graph, then Suurballe and Tarjan s algorithm ([9]) applied to find the shortest elementary -route in O( E + V log V ). The shortest -path in the backward graph may be found the same way. The shortest -paths in the original forward and backward graphs may be recovered by re-merging the parallel arcs. rovided that the initial flow is feasible, we begin by establishing that the updated flow at any iteration in algorithm AUG is feasible as well. Lemma. Let (f ij, f ij) be a feasible (, )-route flow with values ( F, F /). Then the (, )- flow ( f ij, f ij ), defined in ()-(5), is a new feasible (, )-flow with values ( F, F / + δ). roof: Evidently the flow conservation constraints (5) hold. Capacity (6) and extra capacity constraints (7) follow by examining the following sixteen mutually exclusive and exhaustive cases. Case. (i,j) A +(f), (i,j), (j,i) / A (f); or (i,j), (j,i) / A +(f) and (i,j) A (f):.. (i,j) A S +(f);.. (i,j) A D +(f);.3. (i,j) A S (f);.. (i,j) A D (f). Case. (i,j) A +(f) A (f):.. (i,j) A S +(f) A S (f);.. (i,j) A S +(f) A D (f);.3. (i,j)a D +(f) A S (f);.. (i,j) A D +(f) A D (f) Case 3. (i,j) A +(f) and (j,i) A (f): 3.. (i,j) A S +(f), and (j,i) A S (f); 3.. (i,j) A S +(f), and (j,i) A D (f); 3.3. (i,j) A S +(f), and (j,i) A S (f); 3.. (i,j) A S +(f), and (j,i) A D (f) Case. (j,i) A +(f) and (i,j) A (f):.. (j,i) A S +(f), and (i,j) A S (f);.. (j,i) A S +(f), and (i,j) A D (f);.3. (j,i) A D +(f), and (i,j) A S (f);.. (j,i) A D +(f), and (i,j) A D (f) 3

15 Only Case.3 is shown in its entirety; the others are similar. We demonstrate the capacity constraints (6) in the following. f ij δ + f ij + 3δ u ij ; f ij δ f ij 3δ u ij ; f ij + δ + f ij + 3δ u ij ; f ij + δ f ij 3δ u ij. From the definition of δ, we have δ δ AD + (f) u ij f ij f ij; therefore the first one follows since the original flow satisfies the capacity constraints. The third one follows similarly, since we have δ δ AS (f) u ij + f ij f ij. The other two are obvious. To demonstrate the extra capacity constraint (6): f ij F + δ, the following statements must first be shown: f ij 3δ F + δ; f ij + 3δ F + δ. The first one is obvious. The definition of δ implies δ δ F f ij. Therefore the second one follows in that the original flow satisfies its extra capacity constraints. Next, the correctness of algorithm AUG must be proven. Theorem.5 If the algorithm AUG terminates, it results in an optimal solution. roof: Suppose algorithm Aug terminates for some (, )-route flow f = (f,f ) with values ( F, F /). As remarked earlier, for a given flow f, Step 3. of Algorithm Aug iteratively updates the forward and backward graphs, say, K times before Aug terminates. Denote the updated forward graphs as G +(f),,g K +(f), and let A T + (f),,a T K + (f) be respectively the tight bridges therein; analogously, G (f),,g K (f) and A T (f),,a T K (f) are defined for the updated backward graphs. By the updating rule in Step 3., for any k =,,K, we have G k+ + (f) = G k +(f)\a T k (f); G k+ (f) = G k (f)\a T k + (f).

16 Assume, on the contrary, f is not an optimal solution; then a flow value pattern of (0, δ) is still allowed that is, send δ of second flow without changing the value of the first. This assumption, however, leads to a contraction, as shown below. The following claim is crucial. Claim.6 If a flow value pattern of (0, δ) is allowed, then at least δ units of -route flow must be increased along any tight bridge in A T k + (f), and decreased along any tight bridge in A T k (f) for any k =,,K. Since there is nothing to prove when a tight bridge set is empty, we therefore assume, without affecting the later argument, that A T k + (f) and A T k (f) for any k =,,k. We prove the claim by induction on k. For the basis step k =, because of symmetry, we show only the above claim for tight bridges in A T + (f) at least δ units of -route flow must be increased along any arc in A T + (f). Note that s and t are -arc connected in G +(f) since A T + (f). Consider any bridge (i,j ) A T + (f). Let S be the set of vertices in G +(f) that can be reached from s without passing (i,j ) in G +(f). Then in the original undirected network G, (S, S ) is an s -t cut, the edges of which consists of three mutually exclusive parts: (i,j ); A + = {(i,j) (S, S ) : δ + ij = 0}, that is, edges absent from G +(f) in the construction; A = {(i,j) ( S,S ) : δ + ji > 0,δ+ ij = 0}, that is, edges in G +(f), but going from S to S. 5

17 Let (δi j,δi j ), (δ,δ ), and (δ,δ ) be the total of the - and -route flows across A + A + A A (i,j ), A +, and A respectively. We then have the following constraints for these quantities: δ i j δ; (8) δ A + δ A δi j + δ A + δi j + δ A + + δ A + + δ A δ A δ A = 0; (9) = 0; (0) = 0; () = δ. () Note that (8) follows because (i,j ) is a tight bridge; that (9) and (0) follows from the definition of sets A + and A ; that () and () follows from the non-optimal assumption. Therefore δ i j = δ A δ A + = δ A + δ A + = δ δ i j δ, where the three equalities follow from (), (9) and (0), and () respectively, and the inequality follows from (3). Suppose the claim were correct for up to k. We will now show that it is also correct for k +. Because of symmetry, we show only the above claim for tight bridges in the A T k+ (f) at least δ units of -route flow must be decreased along any tight bridge in A T k+ (f). Note that s and t are -arc connected in G k+ (f) since A T k+ (f). Consider any bridge (i k+,j k+ ) A T k+ (f). Let S k+ be the set of vertices in G k+ (f) that can be reached from s without passing (i k+,j k+ ) in G k+ (f). Then in the original undirected network G, (S k+, S k+ ) is an s -t cut, the edges of which consists of four mutually exclusive parts: (i k+,j k+ ); A + k+ = {(i,j) (S k+, S k+ ) : δ ij = 0}, that is, edges absent from G (f) in the construction, going from S k+ to S k+ ; A k+ = {(i,j) ( S k+,s k+ ) : δ ji > 0,δ ij S k+ to S k+ ; = 0}, that is, edges in Gk+ (f), but going from 6

18 k l= AT l + (f), that is, tight bridges deleted from G (f) in the previous k iterations. For the last part, note that not all arcs of k l= AT l + (f) contribute to the construction to the cut (S k+, S k+ ); for any (i,j) k l= AT l + (f), we have either i,j S k+, or i,j S k+, or (i,j) (S k+, S k+ ), or (j,i) S k+, S k+ ). Obviously first two cases have no contribution; we prove that the last case cannot happen, which implies only the third case contributes. Suppose on the contrary, we have (j,i) (S k+, S k+ ) for some (i,j) k l= AT l + (f). Then δ + ij > 0 and fij = F / > 0. This implies that u ij + fji fji > 0, and hence (j,i) G k+ (f). So we have i S k+ according to the construction of set S k+, a contradiction to the assumption i / S k+. { ( k (Sk+ Therefore the last part only contains A ( k) = (i,j) l= AT l + (f)), S } k+ ). Let (δi k+ j k+, δi k+ j k+ ), (δa ( k),δa ( k) ), (δ,δ ), and (δ,δ be the total of the A + k+ A + k+ A k+ Ak+) - and -route flows sent along (i k+,j k+ ), A ( k), A + k+ and A k+ respectively. These quantities satisfy the following constraints: δ A + k+ δ A k+ δ i k+ j k+ δ; (3) δ A + k+ = 0; () δ A k+ = 0; (5) δi k+ j k+ + δa ( k) + δ δ = 0; A + k+ A k+ (6) δi k+ j k+ + δa ( k) + δ δ = δ; A + k+ A k+ (7) δa ( k) δ A ( k) ; (8) δa ( k) T δ A ( k), (9) where (3) and (9) follow because of the tightness of arc (i k+,j k+ ) and arcs of A ( K) ; () and (5) follows from the definition of A + k+ and A k+ ; (6) and (7) follow from the non-optimal 7

19 assumption; (8) follows from the induction hypothesis. So δi k+ j k+ = δ δ δ A k+ A + A k+ ( k) = δ δ δ A k+ A + A k+ ( k) = δ + δ i k+ j k+ + δ A ( k) δ A ( k) δ + δ + δ A ( k) δ A ( k) = δ. The three equalities follow from (6), () and (5), and (7) respectively; the inequality follows from (3), (8), and (9). This finishes the claim. Armed with the above claim, we are ready to establish the contradiction. At termination, there is no s -t path in either G K +(f) or G K (f). Because of symmetry, we consider only backward graph G K (f). Let S K be the set of vertices in G K (f) that can be reached from s in G K (f). Then in the original undirected network G, (S K, S K ) is an s -t cut, the edges of which consists of three mutually exclusive parts: A + K = {(i,j) (S K, S K ) : δ ij = 0}, that is, edges absent from G (f) in the construction, going from S K to S K ; A K = {(i,j) ( S K,S K ) : δ ji > 0,δ ij = 0}, that is, edges in GK (f), but going from S K to S K ; K l= AT l + (f), that is, tight bridges deleted from G (f) in the previous K iterations. Similar as before, the cut only consists of edges in A (<K) = (i,j) { ( K (SK l= AT l + (f)), S } K ). The situation is similar to that of the induction step of Claim.6, but there is no arc (i k+,j k+ ). We define analogous notations: (δa (<K),δA (<K) ), (δ,δ ), (δ,δ A + K A + K A K AK). By Claim.6, we know the total -route flow along the arcs of A (<K) increases units of flow by a certain amount: δ A (<K). To maintain -route flow, we must decrease units of -route flow along arcs of A + K and A K. However, by the definitions of A+ K and A K, we must also decrease units of -route flow along them. Because all arcs in A (<K) are tight, we can increase 8

20 at most δ A (<K) units of the -route flow along the A (<K) -arcs, therefore the total -route flow across the cut (S K, S K ) is no more than 0, a contradiction to the non-optimal assumption. Last is analysis of the time bound of algorithm AUG. Theorem.7 Algorithm AUG delivers an optimal solution in T time, where T is the time to solve a one-commodity -route maximum flow problem. The algorithm in [6, 7], for example, has T = O( V A + V A log V ). roof: Consider a flow f generated at the beginning of Step 3 at an arbitrary state. Let G + (f) be the forward graph generated by Step 3.. Let α + (f) denote the number of arcs of the shortest -paths in G + (f) (double arcs being counted twice) and β + (f) be the number of arcs in the union of the shortest -paths in G + (f). Let f be the updated flow generated from f at Step, so that analogous quantities of α + ( f) and β + ( f) in G + ( f) (the forward graph generated at the end of Step 3. for flow f) may be defined. Du and Chandrasekaran [6, 7] have shown the following to be true: Lemma.8 [6, 7] (i) α + ( f) α + (f); (ii) If α + ( f) = α + (f), then β + ( f) < β + (f). The above result is equally applicable to the backward graph, therefore both α + ( f) and α ( f) are non-decreasing and at least one of the two decreases strictly with each update of the flow. The worst-case scenario takes the time of two runs of the one-commodity -route maximum flow problem, the solution s overall time bound..3 An Illustrative Example of Algorithm AUG Figure 5(a) illustrates the process of algorithm AUG. The number on each arc indicates its capacity. Figure 5(b) shows what results at the end of Step of algorithm AUG, where the first and second numbers are, respectively, the flows of the first and second commodities, and the third number 9

21 6 3 s s t t (a) 3 s s 6 5 t t (c) 6 (-,) 5 (-,) (0,) (0,) s (,) (-,-) (,) t (0,) (0,) 3 (-,) (-,) (-,) s (e) 6 s (0,0,) (6,0,6) (,0,) (,0,) t (0,0,) (,0,) 3 (0,0,6) (,0,) (,0,) 5 (0,0,) (,0,) (b) 3 (0,0,) (,0,) (,0,) s (0,0,) s s 6 s (,,3) (-0.5,-0.5,) (,-,) (d) (-0.5,0.5,) (,0,) (0,-,) (5.5,0.5,6) (,0,) (0,,) (,,6) (.5,-0.5,) (.5,0.5,) t (.5,0.5,) (f) (,0,) (,0,) (.5,0.5,) 6 t t s 6 t s 3 s s 3 s t t t t (g) (h) Figure 5: Illustration of algorithm AUG on an example 0

22 represents the capacity. The value of the the two commodities is (6, 0). The - and -route flows remain the same after Step. Current forward and backward graphs are shown in Figures 5(c) and 5(d). The dashed line Figure 5(c) and the dotted line in Figure 5(d) are, respectively, the corresponding forward and backward -paths constructed after Step 3. Figure 5(e) represents the augmenting scheme achieved by merging these two -paths, Figure 5(f) the new flows with value (6, /), obtained after Step. Figures 5(g) and 5(h) show the new forward and backward graphs. At this stage, no backward -path in the backward auxiliary graph exists; the optimal solution is shown in Figure 5(f). 3 (, )-MF We report in this section partial results of our study of (, )-MF, an immediate extension of (, )-MF. As may be noted for even this special case, the previously accepted technique for solving (, )-MF is inadequate. In Sections 3. and 3., examples show that extremal and maxflow/min-cut properties are not generally true for (, )-MF and explain why previous techniques should no longer be used to solve the general problem, (, )-MF. In Section 3.3, we demonstrate that the notion of extremal property remains valid if there is just one source or destination for (, )-MF. 3. No Extremal roperty The extremal property crucial for Hu s and our (, )-MF and (, )-MF algorithms no longer holds for (, )-MF. As is shown in Figure 6, no -route flow can be sent from s to t once the optimal -route flow from s to t of value 3/ is presented. However, the optimal flow has a value of, therefore two units of -route flow may be sent from s to t as follows: 0.5 units of flow along double path [s s t, s t ] and 0.5 units of flow along double path [s t t, s t ]. Two units of the second commodity from s to t may also be sent: 0.5 units of flow along double path [s s t, s t ] and 0.5 units of flow along double path [s t t, s t ].

23 s s t t (b) Figure 6: Violation of Extremal property for (, )-MF 3. No Max-Flow/Min-Cut Here we show that the max-flow/min-cut equality does not hold for (, )-MF, given single source s and two destinations t, t. We also define two distinct cuts. Definition 3. Consider any undirected, one-source network with two separate destinations. () A type- cut is a set of edges whose deletion disconnects s from t and t. The cut capacity of a type- cut may be described as the lesser of two quantities: (a) half the sum of the capacities of edges in the cut, and (b) the sum of the capacities of arcs, excluding the largest capacity in the cut. () A type- cut consists of a set of edges whose deletion leaves no double path from s to t or t. The cut capacity of a type- cut is half the sum of the capacities of arcs in the cut. The cut capacities are, clearly, the upper bounds of the maximum flow. Figure 7 shows that the max-flow/min-cut equality does not hold. The maximum flow is 8; the minimum cut capacity is min{, 0} = Extremal roperty for Single Source (, )-MF For (, )-MF with a single source or single destination (i.e., s = s or t = t in L 5 ), the extremal property may be proven.

24 t 0 s 0 t Figure 7: Violation of Max-flow/min-cut for single source (, )-MF Lemma 3. There exists an optimal solution with values ( F /, F /) for (, )-MF with single sources such that F = F where (assuming the second flow is ignored) F / is the maximum value of the first commodity. roof: If not, suppose F < F. Disregard the second commodity and locate the -path from s to t, which will permit δ of the -route flow to be increased along this path. If a path in ( F /, F /) cannot be found, it must be because of the presence of the second commodity on arcs of the path. Without loss of generality, assume this -path contains arc disjoint s-t paths U and L. Let e = (i k,j k ) and e = (c,d) be the last arcs with positive second commodity on U and L respectively. Case. If an s-t path passes through both e and e, let be the coupled path (Figure 8(a)), making the proper augmenting path easily verifiable: : s a e b U t, : s t c e d L t. Case. Assume two different s-t paths and passing through e and e respectively. Let 3 and be the corresponding coupled paths (Figure 8(b)). The following illustrates an augmenting scheme: : s a e b U t, : s 3 t b U t ; 3

25 U s t t ' L (a) Case a c e e b d s 3 U L t (b) Case a c e e b d t Figure 8: Extremal property for single source (, )-MF and This finishes the proof. 3: s c e d L t, : s t d L t. Concluding Remarks Certain open problems remain. First, there is no result at present for multi-commodity cases. Knowledge about the two-commodity problem is still relatively limited; here, we solve a special case. One two-commodity problem for prompt investigation is whether or not the augmenting-path technique can be extended to solve either (, )-MF or (,m)-mf. We believe the first step to be a matter of solving the special (, )-MF with a single source or destination, for which we have demonstrated the extremal property holds. References [] C. C. Aggarwal, and J. B. Orlin, On Multiroute Maximum Flows in Networks, Networks, 39() (00) 3-5. [] R. K. Ahuja, T. L. Maganti, and J. B. Orlin, Network Flows: Theory, Algorithms, and Applications, rentice Hall, Englewood Cliffs, NJ, 993.

26 [3] Y.. Aneja, R. Chandrasekaran, S. N. Kabadi, and K.. K. Nair, Flows over Edge-Disjoint Mixed Multipaths and Applications, Working aper, Faculty of Administration, University of New Brunswick, 00. [] R. Chandrasekaran, K.. K. Nair, Y.. Aneja, and S. N. Kabadi, Multi-Terminal Multipath Flows: Synthesis, Discrete Applied Mathematics, 3 (00) [5] E. A. Dinic, Algorithm for Solution of a roblem of Maximum Flow in a Network with ower Estimation, Sov. Math. Dokl, (970) [6] D. Du, Multiroute Flow roblem, hd Thesis, The University of Texas at Dallas, 003. [7] D. Du and R. Chandrasekaran, Multiroute Maximum Flow roblem Revisited, submitted to Networks, 00. [8] J. Edmonds and R. M. Karp, Theoretical Improvements in Algorithmic Efficiency for Network Flow roblems Journal of ACM, 9 (97) 8-6. [9] L. R. Ford and D. R. Fulkerson, Flows in Networks, rinceton University ress, 96. [0] M. Grotschel, L. Lovasz, and A. Shrijver, Geometric Algorithms and Combinatorial Optimization, Springer-Verlag, Berlin, 988. [] T. C. Hu, Multicommodity Network Flows, Operations Research, (963) [] A. Itai, Two-commodity Flow, Journal of ACM, 5() (978) [3] W. Kishimoto, A Method for Obtaining the Maximum Multiroute Flows in a Network, Networks, 7() (996) [] S. N. Kabadi, R. Chandrasekaran, K.. K. Nair, and Y.. Aneja, Integer Version of Multipath Flow Network Synthesis roblem, Working aper, Faculty of Administration, University of New Brunswick,

27 [5] W. Kishimoto and M. Takeuchi, On Two-route Flows in an Undirected Network, IEICE Technical Report, CAS90-9, DS-90-3, (990) (in Japanese). [6] W. Kishimoto and M. Takeuchi, On M-route Flows in a Network, IEICE Trans J-76-A, (993) (in Japanese). [7] W. Kishimoto, M. Takeuchi, and G. Kishi, Two-route Flows in an Undirected Flow Network, IEICE Trans J-75-A, (99) (in Japanese). [8] M. Sakarovitch, Two-commodity Network Flows and L, Mathematical rogramming, (973) -0. [9] J. W. Suurballe and R. E. Tarjan, A Quick Method for Finding Shortest air of Disjoint aths, Networks, (98) [0] E. Tardos, A Strongly olynomial Algorithm to Solve Combinatorial Linear rograms, Operations Research, 3 (986)

The Multiroute Maximum Flow Problem Revisited

The Multiroute Maximum Flow Problem Revisited The Multiroute Maximum Flow Problem Revisited Donglei Du R. Chandrasekaran January 20, 2005 Abstract We are given a directed network G = (V, A, u) with vertex set V, arc set A, a source vertex s V, a destination

More information

FRACTIONAL PACKING OF T-JOINS. 1. Introduction

FRACTIONAL PACKING OF T-JOINS. 1. Introduction FRACTIONAL PACKING OF T-JOINS FRANCISCO BARAHONA Abstract Given a graph with nonnegative capacities on its edges, it is well known that the capacity of a minimum T -cut is equal to the value of a maximum

More information

Strongly Polynomial Algorithm for a Class of Minimum-Cost Flow Problems with Separable Convex Objectives

Strongly Polynomial Algorithm for a Class of Minimum-Cost Flow Problems with Separable Convex Objectives Strongly Polynomial Algorithm for a Class of Minimum-Cost Flow Problems with Separable Convex Objectives László A. Végh April 12, 2013 Abstract A well-studied nonlinear extension of the minimum-cost flow

More information

Algorithms and Theory of Computation. Lecture 11: Network Flow

Algorithms and Theory of Computation. Lecture 11: Network Flow Algorithms and Theory of Computation Lecture 11: Network Flow Xiaohui Bei MAS 714 September 18, 2018 Nanyang Technological University MAS 714 September 18, 2018 1 / 26 Flow Network A flow network is a

More information

Maximum flow problem

Maximum flow problem Maximum flow problem 7000 Network flows Network Directed graph G = (V, E) Source node s V, sink node t V Edge capacities: cap : E R 0 Flow: f : E R 0 satisfying 1. Flow conservation constraints e:target(e)=v

More information

Preliminaries. Graphs. E : set of edges (arcs) (Undirected) Graph : (i, j) = (j, i) (edges) V = {1, 2, 3, 4, 5}, E = {(1, 3), (3, 2), (2, 4)}

Preliminaries. Graphs. E : set of edges (arcs) (Undirected) Graph : (i, j) = (j, i) (edges) V = {1, 2, 3, 4, 5}, E = {(1, 3), (3, 2), (2, 4)} Preliminaries Graphs G = (V, E), V : set of vertices E : set of edges (arcs) (Undirected) Graph : (i, j) = (j, i) (edges) 1 2 3 5 4 V = {1, 2, 3, 4, 5}, E = {(1, 3), (3, 2), (2, 4)} 1 Directed Graph (Digraph)

More information

arxiv: v1 [cs.dm] 12 Jun 2016

arxiv: v1 [cs.dm] 12 Jun 2016 A Simple Extension of Dirac s Theorem on Hamiltonicity Yasemin Büyükçolak a,, Didem Gözüpek b, Sibel Özkana, Mordechai Shalom c,d,1 a Department of Mathematics, Gebze Technical University, Kocaeli, Turkey

More information

Approximate Binary Search Algorithms for Mean Cuts and Cycles

Approximate Binary Search Algorithms for Mean Cuts and Cycles Approximate Binary Search Algorithms for Mean Cuts and Cycles S. Thomas McCormick Faculty of Commerce and Business Administration University of British Columbia Vancouver, BC V6T 1Z2 Canada June 1992,

More information

WEAKLY AND STRONGLY POLYNOMIAL ALGORITHMS FOR COMPUTING THE MAXIMUM DECREASE IN UNIFORM ARC CAPACITIES

WEAKLY AND STRONGLY POLYNOMIAL ALGORITHMS FOR COMPUTING THE MAXIMUM DECREASE IN UNIFORM ARC CAPACITIES Yugoslav Journal of Operations Research 6 (016), Number, 159 171 DOI: 10.98/YJOR14117015G WEAKLY AND STRONGLY POLYNOMIAL ALGORITHMS FOR COMPUTING THE MAXIMUM DECREASE IN UNIFORM ARC CAPACITIES Mehdi GHIYASVAND

More information

Fast algorithms for even/odd minimum cuts and generalizations

Fast algorithms for even/odd minimum cuts and generalizations Fast algorithms for even/odd minimum cuts and generalizations András A. Benczúr 1 and Ottilia Fülöp 2 {benczur,otti}@cs.elte.hu 1 Computer and Automation Institute, Hungarian Academy of Sciences, and Department

More information

Lecture 21 November 11, 2014

Lecture 21 November 11, 2014 CS 224: Advanced Algorithms Fall 2-14 Prof. Jelani Nelson Lecture 21 November 11, 2014 Scribe: Nithin Tumma 1 Overview In the previous lecture we finished covering the multiplicative weights method and

More information

Multicommodity Flows and Column Generation

Multicommodity Flows and Column Generation Lecture Notes Multicommodity Flows and Column Generation Marc Pfetsch Zuse Institute Berlin pfetsch@zib.de last change: 2/8/2006 Technische Universität Berlin Fakultät II, Institut für Mathematik WS 2006/07

More information

The min cost flow problem Course notes for Search and Optimization Spring 2004

The min cost flow problem Course notes for Search and Optimization Spring 2004 The min cost flow problem Course notes for Search and Optimization Spring 2004 Peter Bro Miltersen February 20, 2004 Version 1.3 1 Definition of the min cost flow problem We shall consider a generalization

More information

The Maximum Flow Problem with Disjunctive Constraints

The Maximum Flow Problem with Disjunctive Constraints The Maximum Flow Problem with Disjunctive Constraints Ulrich Pferschy Joachim Schauer Abstract We study the maximum flow problem subject to binary disjunctive constraints in a directed graph: A negative

More information

Robust Network Codes for Unicast Connections: A Case Study

Robust Network Codes for Unicast Connections: A Case Study Robust Network Codes for Unicast Connections: A Case Study Salim Y. El Rouayheb, Alex Sprintson, and Costas Georghiades Department of Electrical and Computer Engineering Texas A&M University College Station,

More information

An algorithm to increase the node-connectivity of a digraph by one

An algorithm to increase the node-connectivity of a digraph by one Discrete Optimization 5 (2008) 677 684 Contents lists available at ScienceDirect Discrete Optimization journal homepage: www.elsevier.com/locate/disopt An algorithm to increase the node-connectivity of

More information

7. Lecture notes on the ellipsoid algorithm

7. Lecture notes on the ellipsoid algorithm Massachusetts Institute of Technology Michel X. Goemans 18.433: Combinatorial Optimization 7. Lecture notes on the ellipsoid algorithm The simplex algorithm was the first algorithm proposed for linear

More information

Lecture notes on the ellipsoid algorithm

Lecture notes on the ellipsoid algorithm Massachusetts Institute of Technology Handout 1 18.433: Combinatorial Optimization May 14th, 007 Michel X. Goemans Lecture notes on the ellipsoid algorithm The simplex algorithm was the first algorithm

More information

Lecture 1. 1 Overview. 2 Maximum Flow. COMPSCI 532: Design and Analysis of Algorithms August 26, 2015

Lecture 1. 1 Overview. 2 Maximum Flow. COMPSCI 532: Design and Analysis of Algorithms August 26, 2015 COMPSCI 532: Design and Analysis of Algorithms August 26, 205 Lecture Lecturer: Debmalya Panigrahi Scribe: Allen Xiao Oeriew In this lecture, we will define the maximum flow problem and discuss two algorithms

More information

Maximum flow problem (part I)

Maximum flow problem (part I) Maximum flow problem (part I) Combinatorial Optimization Giovanni Righini Università degli Studi di Milano Definitions A flow network is a digraph D = (N,A) with two particular nodes s and t acting as

More information

Lecture 2: Network Flows 1

Lecture 2: Network Flows 1 Comp 260: Advanced Algorithms Tufts University, Spring 2011 Lecture by: Prof. Cowen Scribe: Saeed Majidi Lecture 2: Network Flows 1 A wide variety of problems, including the matching problems discussed

More information

Maximum Flow Problem (Ford and Fulkerson, 1956)

Maximum Flow Problem (Ford and Fulkerson, 1956) Maximum Flow Problem (Ford and Fulkerson, 196) In this problem we find the maximum flow possible in a directed connected network with arc capacities. There is unlimited quantity available in the given

More information

The max flow problem. Ford-Fulkerson method. A cut. Lemma Corollary Max Flow Min Cut Theorem. Max Flow Min Cut Theorem

The max flow problem. Ford-Fulkerson method. A cut. Lemma Corollary Max Flow Min Cut Theorem. Max Flow Min Cut Theorem The max flow problem Ford-Fulkerson method 7 11 Ford-Fulkerson(G) f = 0 while( simple path p from s to t in G f ) 10-2 2 1 f := f + f p output f 4 9 1 2 A cut Lemma 26. + Corollary 26.6 Let f be a flow

More information

Lecture 3. 1 Polynomial-time algorithms for the maximum flow problem

Lecture 3. 1 Polynomial-time algorithms for the maximum flow problem ORIE 633 Network Flows August 30, 2007 Lecturer: David P. Williamson Lecture 3 Scribe: Gema Plaza-Martínez 1 Polynomial-time algorithms for the maximum flow problem 1.1 Introduction Let s turn now to considering

More information

The min cost flow problem Course notes for Optimization Spring 2007

The min cost flow problem Course notes for Optimization Spring 2007 The min cost flow problem Course notes for Optimization Spring 2007 Peter Bro Miltersen February 7, 2007 Version 3.0 1 Definition of the min cost flow problem We shall consider a generalization of the

More information

Lecture 1 : Probabilistic Method

Lecture 1 : Probabilistic Method IITM-CS6845: Theory Jan 04, 01 Lecturer: N.S.Narayanaswamy Lecture 1 : Probabilistic Method Scribe: R.Krithika The probabilistic method is a technique to deal with combinatorial problems by introducing

More information

Lecture 8 Network Optimization Algorithms

Lecture 8 Network Optimization Algorithms Advanced Algorithms Floriano Zini Free University of Bozen-Bolzano Faculty of Computer Science Academic Year 2013-2014 Lecture 8 Network Optimization Algorithms 1 21/01/14 Introduction Network models have

More information

We say that a flow is feasible for G (or just feasible if the graph and capacity function in question are obvious) if

We say that a flow is feasible for G (or just feasible if the graph and capacity function in question are obvious) if CS787: Advanced Algorithms Lecture 4: Network Flow We devote this lecture to the network flow problem and the max-flow min-cut theorem. A number of other problems can be cast into a maximum flow or minimum

More information

CS261: Problem Set #3

CS261: Problem Set #3 CS261: Problem Set #3 Due by 11:59 PM on Tuesday, February 23, 2016 Instructions: (1) Form a group of 1-3 students. You should turn in only one write-up for your entire group. (2) Submission instructions:

More information

Week 4. (1) 0 f ij u ij.

Week 4. (1) 0 f ij u ij. Week 4 1 Network Flow Chapter 7 of the book is about optimisation problems on networks. Section 7.1 gives a quick introduction to the definitions of graph theory. In fact I hope these are already known

More information

Duality of LPs and Applications

Duality of LPs and Applications Lecture 6 Duality of LPs and Applications Last lecture we introduced duality of linear programs. We saw how to form duals, and proved both the weak and strong duality theorems. In this lecture we will

More information

Network Flows. CTU FEE Department of control engineering. March 28, 2017

Network Flows. CTU FEE Department of control engineering. March 28, 2017 Network Flows Zdeněk Hanzálek, Přemysl Šůcha hanzalek@fel.cvut.cz CTU FEE Department of control engineering March 28, 2017 Z. Hanzálek (CTU FEE) Network Flows March 28, 2017 1 / 44 Table of contents 1

More information

Price of Stability in Survivable Network Design

Price of Stability in Survivable Network Design Noname manuscript No. (will be inserted by the editor) Price of Stability in Survivable Network Design Elliot Anshelevich Bugra Caskurlu Received: January 2010 / Accepted: Abstract We study the survivable

More information

Colored Bin Packing: Online Algorithms and Lower Bounds

Colored Bin Packing: Online Algorithms and Lower Bounds Noname manuscript No. (will be inserted by the editor) Colored Bin Packing: Online Algorithms and Lower Bounds Martin Böhm György Dósa Leah Epstein Jiří Sgall Pavel Veselý Received: date / Accepted: date

More information

Finding k disjoint paths in a directed planar graph

Finding k disjoint paths in a directed planar graph Finding k disjoint paths in a directed planar graph Alexander Schrijver CWI Kruislaan 413 1098 SJ Amsterdam The Netherlands and Department of Mathematics University of Amsterdam Plantage Muidergracht 24

More information

Alternating paths revisited II: restricted b-matchings in bipartite graphs

Alternating paths revisited II: restricted b-matchings in bipartite graphs Egerváry Research Group on Combinatorial Optimization Technical reports TR-2005-13. Published by the Egerváry Research Group, Pázmány P. sétány 1/C, H 1117, Budapest, Hungary. Web site: www.cs.elte.hu/egres.

More information

The maximum flow problem

The maximum flow problem The maximum flow problem A. Agnetis 1 Basic properties Given a network G = (N, A) (having N = n nodes and A = m arcs), and two nodes s (source) and t (sink), the maximum flow problem consists in finding

More information

Mathematics for Decision Making: An Introduction. Lecture 13

Mathematics for Decision Making: An Introduction. Lecture 13 Mathematics for Decision Making: An Introduction Lecture 13 Matthias Köppe UC Davis, Mathematics February 17, 2009 13 1 Reminder: Flows in networks General structure: Flows in networks In general, consider

More information

GRAPH ALGORITHMS Week 7 (13 Nov - 18 Nov 2017)

GRAPH ALGORITHMS Week 7 (13 Nov - 18 Nov 2017) GRAPH ALGORITHMS Week 7 (13 Nov - 18 Nov 2017) C. Croitoru croitoru@info.uaic.ro FII November 12, 2017 1 / 33 OUTLINE Matchings Analytical Formulation of the Maximum Matching Problem Perfect Matchings

More information

CS 6820 Fall 2014 Lectures, October 3-20, 2014

CS 6820 Fall 2014 Lectures, October 3-20, 2014 Analysis of Algorithms Linear Programming Notes CS 6820 Fall 2014 Lectures, October 3-20, 2014 1 Linear programming The linear programming (LP) problem is the following optimization problem. We are given

More information

Combinatorial Optimization

Combinatorial Optimization Combinatorial Optimization 2017-2018 1 Maximum matching on bipartite graphs Given a graph G = (V, E), find a maximum cardinal matching. 1.1 Direct algorithms Theorem 1.1 (Petersen, 1891) A matching M is

More information

An Algebraic View of the Relation between Largest Common Subtrees and Smallest Common Supertrees

An Algebraic View of the Relation between Largest Common Subtrees and Smallest Common Supertrees An Algebraic View of the Relation between Largest Common Subtrees and Smallest Common Supertrees Francesc Rosselló 1, Gabriel Valiente 2 1 Department of Mathematics and Computer Science, Research Institute

More information

Contents. Introduction

Contents. Introduction Introduction Contents Chapter 1. Network Flows 1 1.1. Graphs 1 1.2. Shortest Paths 8 1.3. Maximum Flow 17 1.4. Min Cost Flows 22 List of frequently used Symbols 31 Bibliography 33 Index 37 iii i Introduction

More information

Flows. Chapter Circulations

Flows. Chapter Circulations Chapter 4 Flows For a directed graph D = (V,A), we define δ + (U) := {(u,v) A : u U,v / U} as the arcs leaving U and δ (U) := {(u,v) A u / U,v U} as the arcs entering U. 4. Circulations In a directed graph

More information

Almost Cross-Intersecting and Almost Cross-Sperner Pairs offamilies of Sets

Almost Cross-Intersecting and Almost Cross-Sperner Pairs offamilies of Sets Almost Cross-Intersecting and Almost Cross-Sperner Pairs offamilies of Sets Dániel Gerbner a Nathan Lemons b Cory Palmer a Balázs Patkós a, Vajk Szécsi b a Hungarian Academy of Sciences, Alfréd Rényi Institute

More information

Agenda. Soviet Rail Network, We ve done Greedy Method Divide and Conquer Dynamic Programming

Agenda. Soviet Rail Network, We ve done Greedy Method Divide and Conquer Dynamic Programming Agenda We ve done Greedy Method Divide and Conquer Dynamic Programming Now Flow Networks, Max-flow Min-cut and Applications c Hung Q. Ngo (SUNY at Buffalo) CSE 531 Algorithm Analysis and Design 1 / 52

More information

An example of LP problem: Political Elections

An example of LP problem: Political Elections Linear Programming An example of LP problem: Political Elections Suppose that you are a politician trying to win an election. Your district has three different types of areas: urban, suburban, and rural.

More information

CMPSCI 611: Advanced Algorithms

CMPSCI 611: Advanced Algorithms CMPSCI 611: Advanced Algorithms Lecture 12: Network Flow Part II Andrew McGregor Last Compiled: December 14, 2017 1/26 Definitions Input: Directed Graph G = (V, E) Capacities C(u, v) > 0 for (u, v) E and

More information

Some results on max-min fair routing

Some results on max-min fair routing Some results on max-min fair routing Dritan Nace, Linh Nhat Doan University of Technology of Compiegne. Laboratory Heudiasyc UMR CNRS 6599, 60205 Compiègne Cedex, France. Phone: 00 33 344234302, fax: 00

More information

4-coloring P 6 -free graphs with no induced 5-cycles

4-coloring P 6 -free graphs with no induced 5-cycles 4-coloring P 6 -free graphs with no induced 5-cycles Maria Chudnovsky Department of Mathematics, Princeton University 68 Washington Rd, Princeton NJ 08544, USA mchudnov@math.princeton.edu Peter Maceli,

More information

The partial inverse minimum cut problem with L 1 -norm is strongly NP-hard. Elisabeth Gassner

The partial inverse minimum cut problem with L 1 -norm is strongly NP-hard. Elisabeth Gassner FoSP Algorithmen & mathematische Modellierung FoSP Forschungsschwerpunkt Algorithmen und mathematische Modellierung The partial inverse minimum cut problem with L 1 -norm is strongly NP-hard Elisabeth

More information

4 CONNECTED PROJECTIVE-PLANAR GRAPHS ARE HAMILTONIAN. Robin Thomas* Xingxing Yu**

4 CONNECTED PROJECTIVE-PLANAR GRAPHS ARE HAMILTONIAN. Robin Thomas* Xingxing Yu** 4 CONNECTED PROJECTIVE-PLANAR GRAPHS ARE HAMILTONIAN Robin Thomas* Xingxing Yu** School of Mathematics Georgia Institute of Technology Atlanta, Georgia 30332, USA May 1991, revised 23 October 1993. Published

More information

Standard Diraphs the (unique) digraph with no vertices or edges. (modulo n) for every 1 i n A digraph whose underlying graph is a complete graph.

Standard Diraphs the (unique) digraph with no vertices or edges. (modulo n) for every 1 i n A digraph whose underlying graph is a complete graph. 5 Directed Graphs What is a directed graph? Directed Graph: A directed graph, or digraph, D, consists of a set of vertices V (D), a set of edges E(D), and a function which assigns each edge e an ordered

More information

CS675: Convex and Combinatorial Optimization Fall 2014 Combinatorial Problems as Linear Programs. Instructor: Shaddin Dughmi

CS675: Convex and Combinatorial Optimization Fall 2014 Combinatorial Problems as Linear Programs. Instructor: Shaddin Dughmi CS675: Convex and Combinatorial Optimization Fall 2014 Combinatorial Problems as Linear Programs Instructor: Shaddin Dughmi Outline 1 Introduction 2 Shortest Path 3 Algorithms for Single-Source Shortest

More information

Flow Network. The following figure shows an example of a flow network:

Flow Network. The following figure shows an example of a flow network: Maximum Flow 1 Flow Network The following figure shows an example of a flow network: 16 V 1 12 V 3 20 s 10 4 9 7 t 13 4 V 2 V 4 14 A flow network G = (V,E) is a directed graph. Each edge (u, v) E has a

More information

Internet Routing Example

Internet Routing Example Internet Routing Example Acme Routing Company wants to route traffic over the internet from San Fransisco to New York. It owns some wires that go between San Francisco, Houston, Chicago and New York. The

More information

Part V. Matchings. Matching. 19 Augmenting Paths for Matchings. 18 Bipartite Matching via Flows

Part V. Matchings. Matching. 19 Augmenting Paths for Matchings. 18 Bipartite Matching via Flows Matching Input: undirected graph G = (V, E). M E is a matching if each node appears in at most one Part V edge in M. Maximum Matching: find a matching of maximum cardinality Matchings Ernst Mayr, Harald

More information

A Faster Strongly Polynomial Time Algorithm for Submodular Function Minimization

A Faster Strongly Polynomial Time Algorithm for Submodular Function Minimization A Faster Strongly Polynomial Time Algorithm for Submodular Function Minimization James B. Orlin Sloan School of Management, MIT Cambridge, MA 02139 jorlin@mit.edu Abstract. We consider the problem of minimizing

More information

Deciding Emptiness of the Gomory-Chvátal Closure is NP-Complete, Even for a Rational Polyhedron Containing No Integer Point

Deciding Emptiness of the Gomory-Chvátal Closure is NP-Complete, Even for a Rational Polyhedron Containing No Integer Point Deciding Emptiness of the Gomory-Chvátal Closure is NP-Complete, Even for a Rational Polyhedron Containing No Integer Point Gérard Cornuéjols 1 and Yanjun Li 2 1 Tepper School of Business, Carnegie Mellon

More information

Maximum Integer Flows in Directed Planar Graphs with Multiple Sources and Sinks and Vertex Capacities

Maximum Integer Flows in Directed Planar Graphs with Multiple Sources and Sinks and Vertex Capacities Maximum Integer Flows in Directed Planar Graphs with Multiple Sources and Sinks and Vertex Capacities Yipu Wang University of Illinois at Urbana-Champaign ywang298@illinois.edu July 12, 2018 Abstract We

More information

INVERSE SPANNING TREE PROBLEMS: FORMULATIONS AND ALGORITHMS

INVERSE SPANNING TREE PROBLEMS: FORMULATIONS AND ALGORITHMS INVERSE SPANNING TREE PROBLEMS: FORMULATIONS AND ALGORITHMS P. T. Sokkalingam Department of Mathematics Indian Institute of Technology, Kanpur-208 016, INDIA Ravindra K. Ahuja Dept. of Industrial & Management

More information

Discrete Optimization 2010 Lecture 3 Maximum Flows

Discrete Optimization 2010 Lecture 3 Maximum Flows Remainder: Shortest Paths Maximum Flows Discrete Optimization 2010 Lecture 3 Maximum Flows Marc Uetz University of Twente m.uetz@utwente.nl Lecture 3: sheet 1 / 29 Marc Uetz Discrete Optimization Outline

More information

The Steiner Network Problem

The Steiner Network Problem The Steiner Network Problem Pekka Orponen T-79.7001 Postgraduate Course on Theoretical Computer Science 7.4.2008 Outline 1. The Steiner Network Problem Linear programming formulation LP relaxation 2. The

More information

On shredders and vertex connectivity augmentation

On shredders and vertex connectivity augmentation On shredders and vertex connectivity augmentation Gilad Liberman The Open University of Israel giladliberman@gmail.com Zeev Nutov The Open University of Israel nutov@openu.ac.il Abstract We consider the

More information

Optimal Fractal Coding is NP-Hard 1

Optimal Fractal Coding is NP-Hard 1 Optimal Fractal Coding is NP-Hard 1 (Extended Abstract) Matthias Ruhl, Hannes Hartenstein Institut für Informatik, Universität Freiburg Am Flughafen 17, 79110 Freiburg, Germany ruhl,hartenst@informatik.uni-freiburg.de

More information

ALGORITHMS AND COMPLETE FORMULATIONS FOR THE NETWORK DESIGN PROBLEM Trilochan Sastry Indian Institute of Management, Ahmedabad November 1997 Abstract

ALGORITHMS AND COMPLETE FORMULATIONS FOR THE NETWORK DESIGN PROBLEM Trilochan Sastry Indian Institute of Management, Ahmedabad November 1997 Abstract ALGORITHMS AND COMPLETE FORMULATIONS FOR THE NETWORK DESIGN PROBLEM Trilochan Sastry Indian Institute of Management, Ahmedabad November 1997 Abstract We study the multi commodity uncapacitated network

More information

Generalized Pigeonhole Properties of Graphs and Oriented Graphs

Generalized Pigeonhole Properties of Graphs and Oriented Graphs Europ. J. Combinatorics (2002) 23, 257 274 doi:10.1006/eujc.2002.0574 Available online at http://www.idealibrary.com on Generalized Pigeonhole Properties of Graphs and Oriented Graphs ANTHONY BONATO, PETER

More information

COMPLETION OF PARTIAL LATIN SQUARES

COMPLETION OF PARTIAL LATIN SQUARES COMPLETION OF PARTIAL LATIN SQUARES Benjamin Andrew Burton Honours Thesis Department of Mathematics The University of Queensland Supervisor: Dr Diane Donovan Submitted in 1996 Author s archive version

More information

A combinatorial algorithm minimizing submodular functions in strongly polynomial time

A combinatorial algorithm minimizing submodular functions in strongly polynomial time A combinatorial algorithm minimizing submodular functions in strongly polynomial time Alexander Schrijver 1 Abstract We give a strongly polynomial-time algorithm minimizing a submodular function f given

More information

The maximum edge-disjoint paths problem in complete graphs

The maximum edge-disjoint paths problem in complete graphs Theoretical Computer Science 399 (2008) 128 140 www.elsevier.com/locate/tcs The maximum edge-disjoint paths problem in complete graphs Adrian Kosowski Department of Algorithms and System Modeling, Gdańsk

More information

10 Max-Flow Min-Cut Flows and Capacitated Graphs 10 MAX-FLOW MIN-CUT

10 Max-Flow Min-Cut Flows and Capacitated Graphs 10 MAX-FLOW MIN-CUT 10 Max-Flow Min-Cut 10.1 Flows and Capacitated Graphs Previously, we considered weighted graphs. In this setting, it was natural to thinking about minimizing the weight of a given path. In fact, we considered

More information

Tight Approximation Ratio of a General Greedy Splitting Algorithm for the Minimum k-way Cut Problem

Tight Approximation Ratio of a General Greedy Splitting Algorithm for the Minimum k-way Cut Problem Algorithmica (2011 59: 510 520 DOI 10.1007/s00453-009-9316-1 Tight Approximation Ratio of a General Greedy Splitting Algorithm for the Minimum k-way Cut Problem Mingyu Xiao Leizhen Cai Andrew Chi-Chih

More information

1 Perfect Matching and Matching Polytopes

1 Perfect Matching and Matching Polytopes CS 598CSC: Combinatorial Optimization Lecture date: /16/009 Instructor: Chandra Chekuri Scribe: Vivek Srikumar 1 Perfect Matching and Matching Polytopes Let G = (V, E be a graph. For a set E E, let χ E

More information

Nowhere-zero flows in signed series-parallel graphs arxiv: v1 [math.co] 6 Nov 2014

Nowhere-zero flows in signed series-parallel graphs arxiv: v1 [math.co] 6 Nov 2014 Nowhere-zero flows in signed series-parallel graphs arxiv:1411.1788v1 [math.co] 6 Nov 2014 Tomáš Kaiser 1,2 Edita Rollová 1,3 Abstract Bouchet conjectured in 1983 that each signed graph that admits a nowhere-zero

More information

Sergey Norin Department of Mathematics and Statistics McGill University Montreal, Quebec H3A 2K6, Canada. and

Sergey Norin Department of Mathematics and Statistics McGill University Montreal, Quebec H3A 2K6, Canada. and NON-PLANAR EXTENSIONS OF SUBDIVISIONS OF PLANAR GRAPHS Sergey Norin Department of Mathematics and Statistics McGill University Montreal, Quebec H3A 2K6, Canada and Robin Thomas 1 School of Mathematics

More information

Sequences of height 1 primes in Z[X]

Sequences of height 1 primes in Z[X] Sequences of height 1 primes in Z[X] Stephen McAdam Department of Mathematics University of Texas Austin TX 78712 mcadam@math.utexas.edu Abstract: For each partition J K of {1, 2,, n} (n 2) with J 2, let

More information

Discrete Optimization 2010 Lecture 2 Matroids & Shortest Paths

Discrete Optimization 2010 Lecture 2 Matroids & Shortest Paths Matroids Shortest Paths Discrete Optimization 2010 Lecture 2 Matroids & Shortest Paths Marc Uetz University of Twente m.uetz@utwente.nl Lecture 2: sheet 1 / 25 Marc Uetz Discrete Optimization Matroids

More information

The Cutting Plane Method is Polynomial for Perfect Matchings

The Cutting Plane Method is Polynomial for Perfect Matchings The Cutting Plane Method is Polynomial for Perfect Matchings Karthekeyan Chandrasekaran 1, László A. Végh 2, and Santosh S. Vempala 1 1 College of Computing, Georgia Institute of Technology 2 Department

More information

The shortest path tour problem: problem definition, modeling, and optimization

The shortest path tour problem: problem definition, modeling, and optimization The shortest path tour problem: problem definition, modeling, and optimization Paola Festa Department of Mathematics and Applications, University of apoli FEDERICO II Compl. MSA, Via Cintia, 86 apoli,

More information

Notes on Dantzig-Wolfe decomposition and column generation

Notes on Dantzig-Wolfe decomposition and column generation Notes on Dantzig-Wolfe decomposition and column generation Mette Gamst November 11, 2010 1 Introduction This note introduces an exact solution method for mathematical programming problems. The method is

More information

MINIMALLY NON-PFAFFIAN GRAPHS

MINIMALLY NON-PFAFFIAN GRAPHS MINIMALLY NON-PFAFFIAN GRAPHS SERGUEI NORINE AND ROBIN THOMAS Abstract. We consider the question of characterizing Pfaffian graphs. We exhibit an infinite family of non-pfaffian graphs minimal with respect

More information

Polynomiality of Linear Programming

Polynomiality of Linear Programming Chapter 10 Polynomiality of Linear Programming In the previous section, we presented the Simplex Method. This method turns out to be very efficient for solving linear programmes in practice. While it is

More information

Lecture 5. Shearer s Lemma

Lecture 5. Shearer s Lemma Stanford University Spring 2016 Math 233: Non-constructive methods in combinatorics Instructor: Jan Vondrák Lecture date: April 6, 2016 Scribe: László Miklós Lovász Lecture 5. Shearer s Lemma 5.1 Introduction

More information

Reachability-based matroid-restricted packing of arborescences

Reachability-based matroid-restricted packing of arborescences Egerváry Research Group on Combinatorial Optimization Technical reports TR-2016-19. Published by the Egerváry Research Group, Pázmány P. sétány 1/C, H 1117, Budapest, Hungary. Web site: www.cs.elte.hu/egres.

More information

Introduction to Graph Theory

Introduction to Graph Theory Introduction to Graph Theory George Voutsadakis 1 1 Mathematics and Computer Science Lake Superior State University LSSU Math 351 George Voutsadakis (LSSU) Introduction to Graph Theory August 2018 1 /

More information

CS675: Convex and Combinatorial Optimization Fall 2016 Combinatorial Problems as Linear and Convex Programs. Instructor: Shaddin Dughmi

CS675: Convex and Combinatorial Optimization Fall 2016 Combinatorial Problems as Linear and Convex Programs. Instructor: Shaddin Dughmi CS675: Convex and Combinatorial Optimization Fall 2016 Combinatorial Problems as Linear and Convex Programs Instructor: Shaddin Dughmi Outline 1 Introduction 2 Shortest Path 3 Algorithms for Single-Source

More information

Graphs with few total dominating sets

Graphs with few total dominating sets Graphs with few total dominating sets Marcin Krzywkowski marcin.krzywkowski@gmail.com Stephan Wagner swagner@sun.ac.za Abstract We give a lower bound for the number of total dominating sets of a graph

More information

ACO Comprehensive Exam March 17 and 18, Computability, Complexity and Algorithms

ACO Comprehensive Exam March 17 and 18, Computability, Complexity and Algorithms 1. Computability, Complexity and Algorithms (a) Let G(V, E) be an undirected unweighted graph. Let C V be a vertex cover of G. Argue that V \ C is an independent set of G. (b) Minimum cardinality vertex

More information

The Singapore Copyright Act applies to the use of this document.

The Singapore Copyright Act applies to the use of this document. Title On graphs whose low polynomials have real roots only Author(s) Fengming Dong Source Electronic Journal of Combinatorics, 25(3): P3.26 Published by Electronic Journal of Combinatorics This document

More information

An 0.5-Approximation Algorithm for MAX DICUT with Given Sizes of Parts

An 0.5-Approximation Algorithm for MAX DICUT with Given Sizes of Parts An 0.5-Approximation Algorithm for MAX DICUT with Given Sizes of Parts Alexander Ageev Refael Hassin Maxim Sviridenko Abstract Given a directed graph G and an edge weight function w : E(G) R +, themaximumdirectedcutproblem(max

More information

Advanced Combinatorial Optimization Updated April 29, Lecture 20. Lecturer: Michel X. Goemans Scribe: Claudio Telha (Nov 17, 2009)

Advanced Combinatorial Optimization Updated April 29, Lecture 20. Lecturer: Michel X. Goemans Scribe: Claudio Telha (Nov 17, 2009) 18.438 Advanced Combinatorial Optimization Updated April 29, 2012 Lecture 20 Lecturer: Michel X. Goemans Scribe: Claudio Telha (Nov 17, 2009) Given a finite set V with n elements, a function f : 2 V Z

More information

On Two Class-Constrained Versions of the Multiple Knapsack Problem

On Two Class-Constrained Versions of the Multiple Knapsack Problem On Two Class-Constrained Versions of the Multiple Knapsack Problem Hadas Shachnai Tami Tamir Department of Computer Science The Technion, Haifa 32000, Israel Abstract We study two variants of the classic

More information

Some hard families of parameterised counting problems

Some hard families of parameterised counting problems Some hard families of parameterised counting problems Mark Jerrum and Kitty Meeks School of Mathematical Sciences, Queen Mary University of London {m.jerrum,k.meeks}@qmul.ac.uk September 2014 Abstract

More information

On a Conjecture of Thomassen

On a Conjecture of Thomassen On a Conjecture of Thomassen Michelle Delcourt Department of Mathematics University of Illinois Urbana, Illinois 61801, U.S.A. delcour2@illinois.edu Asaf Ferber Department of Mathematics Yale University,

More information

Lecture 24: April 12

Lecture 24: April 12 CS271 Randomness & Computation Spring 2018 Instructor: Alistair Sinclair Lecture 24: April 12 Disclaimer: These notes have not been subjected to the usual scrutiny accorded to formal publications. They

More information

Running Time. Assumption. All capacities are integers between 1 and C.

Running Time. Assumption. All capacities are integers between 1 and C. Running Time Assumption. All capacities are integers between and. Invariant. Every flow value f(e) and every residual capacities c f (e) remains an integer throughout the algorithm. Theorem. The algorithm

More information

Parameterized Algorithms for the H-Packing with t-overlap Problem

Parameterized Algorithms for the H-Packing with t-overlap Problem Journal of Graph Algorithms and Applications http://jgaa.info/ vol. 18, no. 4, pp. 515 538 (2014) DOI: 10.7155/jgaa.00335 Parameterized Algorithms for the H-Packing with t-overlap Problem Alejandro López-Ortiz

More information

Flows and Cuts. 1 Concepts. CS 787: Advanced Algorithms. Instructor: Dieter van Melkebeek

Flows and Cuts. 1 Concepts. CS 787: Advanced Algorithms. Instructor: Dieter van Melkebeek CS 787: Advanced Algorithms Flows and Cuts Instructor: Dieter van Melkebeek This lecture covers the construction of optimal flows and cuts in networks, their relationship, and some applications. It paves

More information

3.8 Strong valid inequalities

3.8 Strong valid inequalities 3.8 Strong valid inequalities By studying the problem structure, we can derive strong valid inequalities which lead to better approximations of the ideal formulation conv(x ) and hence to tighter bounds.

More information

Computing Maximum Flow with Augmenting Electrical Flows

Computing Maximum Flow with Augmenting Electrical Flows Computing Maximum Flow with Augmenting Electrical Flows Aleksander Mądry MIT madry@mit.edu Abstract ( ) We present an Õ m 0 7 U 7 -time algorithm for the maximum s-t flow problem (and the minimum s-t cut

More information