Branch-and-Cut Algorithms for the Vehicle Routing Problem with Trailers and Transshipments

Size: px
Start display at page:

Download "Branch-and-Cut Algorithms for the Vehicle Routing Problem with Trailers and Transshipments"

Transcription

1 Branch-and-Cut Agorithms for the Vece Routing Probem with Traiers and Transspments Technica Report LM Michae Dre Chair of Logistics Management, Gutenberg Schoo of Management and Economics, Johannes Gutenberg University, Mainz and Fraunhofer Centre for Appied Research on Suppy Chain Services SCS, Nuremberg E-mai: 25th September 2012 Abstract Ts paper studies the vece routing probem with traiers and transspments (VRPTT), a practicay reevant, but chaenging, generaization of the cassica vece routing probem. The paper maes three contributions: (i) Buiding on a non-trivia networ representation, two mied-integer programming formuations for the VRPTT are proposed. (ii) Based on these formuations, five different branch-and-cut agorithms are deveoped and impemented. (iii) The computationa behaviour of the agorithms is anayzed in an etensive computationa study, using a arge number of test instances designed to resembe rea-word VRPTTs. Keywords: Branch-and-cut; Vece Routing; Synchronization; Traier; Transspment 1 Introduction The vece routing probem with traiers and transspments (VRPTT) was introduced by Dre [11] in the contet of raw mi coection at farmyards. It is a generaization of the cassica vece routing probem (VRP, Dantzig and Ramser [9], Toth and Vigo [23], Goden et a. [14]). The version of the VRPTT studied in ts paper can be described as foows. There is a set of customers with a given suppy. To coect the suppy, a imited feet of heterogeneous veces (orries and traiers) stationed at a centra depot is avaiabe. Besides unequa fied and variabe costs and capacities, the veces differ in that orries are autonomous and abe to move in time and space on their own, whereas traiers are non-autonomous and can move in time on their own (that is, wait), but must be pued by a orry to move in space. In addition to the depot and the customer ocations, there are so-caed transspment ocations (TLs), where traiers can be pared and where oad transfers from orries to traiers can be performed. Some customers can ony be visited by a orry without a traier (a singe orry) and are hence caed orry customers. The other customers can be visited by a orry with or without a traier and are caed traier customers. There are time windows at the customers as we as at the TLs. A veces start and end their routes at the depot. There is no fied assignment of a traier to a orry. Any traier may be pued, on the whoe or on a part of its itinerary, by any compatibe orry. What is more, any orry may transfer part or a of its oad to any traier at any TL. The time a transspment taes depends on the amount of oad transferred. Lorries need not bring bac a traier, neither one they may have pued when eaving the depot, nor any other. The probem is to determine routes for orries and routes for traiers so that tota costs are minimized, the compete suppy of a customers is deivered to the depot, and oading capacities and time windows are maintained. Moreover, it must be ensured that the routes are synchronized 1

2 with respect to space, time and oad, so that traiers move in space ony when accompanied by compatibe orries and that the veces invoved in a oad transfer operation visit the pertinent ocation at the right time and transfer and receive the right amount of oad. An eampe route pan is depicted in Figure 1. Depot Lorry customer Traier customer Transspment ocation Lorry 1 Lorry 2 Lorry 3 Traier Figure 1: VRPTT eampe route pan In the eampe, orry 1 pus the traier from the depot to a TL and decoupes the traier there. Lorry 1 then visits two orry customers, returns to the traier, transfers oad, eaves the traier at the TL and returns to the depot via two orry and two traier customers. Lorry 2 starts at the depot and visits two orry customers before couping the traier, after orry 1 has performed its oad transfer. Lorry 2 then visits a traier customer, decoupes the traier at another TL, performs a oad transfer, visits some orry customers, returns to the traier, re-coupes it and pus it bac to the depot via a traier customer. Lorry 3 starts at the depot, visits some orry customers, and transfers some oad to the traier we orry 2 is visiting the three rightmost orry customers. After that, orry 3 returns to the depot via another orry customer. The two TLs in the centre of the figure are not used. The justification for studying the VRPTT is that the probem, apart from being interesting and chaenging in its own right, has numerous direct practica appications and, in addition, constitutes an archetypa, generic representative of the cass of vece routing probems with mutipe synchronization constraints (VRPMSs). Contrary to cassica VRPs, where a synchronization between veces is necessary ony with respect to wch vece visits wch customer, VRPMSs ebit additiona synchronization requirements with regard to spacia, tempora, and oad aspects. Ts is discussed in more detai in Section 2. The contribution of the present paper is three-fod: (i) Buiding on a non-trivia networ representation, two mied-integer programming (MIP) formuations for the VRPTT are proposed. (ii) Based on these formuations, five different branch-and-cut agorithms are deveoped and impemented. (iii) The computationa behaviour of the agorithms is anayzed in an etensive computationa study, using a arge number of test instances designed to resembe rea-word VRPTTs. The rest of the paper is structured as foows. The net section provides an overview of reated wor, incuding other types of VRPMSs. Section 3 presents a networ representation for the VRPTT as a basis for the two arc-variabe based formuations deveoped in Section 4. Section 5 describes the soution approaches that were impemented to sove these formuations, and Section 6 presents computationa eperiments performed with the impementations. The fina Section 7 concudes the paper with a research outoo 2 Literature review The avaiabe iterature on the VRPTT and reated VRPMSs is rather imited. As stated, the VRPTT was introduced in Dre [11], where formuations based on arc, path, and so-caed turn variabes were deveoped and a branch-and-cut agorithm for an arc variabe formuation 2

3 was presented. Ts author is not aware of other papers containing a soution approach for the VRPTT. A quite we-studied probem deaing with traiers in vece routing is the truc-and-traier routing probem (TTRP) (see, for eampe, Semet and Taiard [22], Chao [4], Scheuerer [21]). The TTRP is a specia case of the VRPTT with a fied orry-traier assignment, that is, each traier can be pued by ony one orry, and ony ts orry can transfer oad into the traier. Ts maes the probem much easier. In fact, the TTRP is not a VRPMS; there are no synchronization requirements other than customer covering. A recent survey of synchronization in vece routing is given in Dre [12]. In ts paper, five different types of synchronization are identified: (i) tas synchronization, referring to the decision wch vece(s) fufi(s) wch tas(s); (ii) operation synchronization, wch decides about time and ocation of some interaction between veces; (iii) movement synchronization, that is, the decision wch veces join to move in space; (iv) oad synchronization, wch determines the amounts of oad to be coected, deivered, or transspped; and (v) resource synchronization, wch decides about wch vece(s) use(s) a scarce resource at what time. The most important casses of VRPMSs identified in ts survey are muti-echeon vece and ocation-routing probems with tempora synchronization of veces of different echeons (Crainic et a. [8]), simutaneous vece and driver routing probems (as opposed to integrated vece and crew scheduing probems, where the tass to perform usuay have a given fied schedue) (Hois et a. [15]), and picup-and-deivery probems with transspments and simutaneous oad transfers (Cortés et a. [7]). Wors that study probems where different types of autonomous and non-autonomous objects that may join and separate en route (and not ony at a centra depot) are used to fufi tass are Bürcert et a. [3], Hois et a. [15], Cheung et a. [5], Kim et a. [16], and Dre et a. [13]. A of these papers deveop heuristic soution approaches: Bürcert et a. [3] consider a dynamic picup-and-deivery probem using orries, traiers, and swap-body patforms in the contet of ong-hau transport and sove their probem with a hoonic muti-agent system. Both Hois et a. [15] and Dre et a. [13] sove a simutaneous vece and driver routing probem (with appication in posta ogistics in Austraia and ong-hau road transport in Europe respectivey) with two-stage agorithms. Hois et a. [15] first compute abstract vece routes by soving a rich picup-and-deivery probem by heuristic coumn generation. Then, concrete veces and drivers are assigned by taing an integrated vece and crew scheduing approach, again done by soving an MIP by heuristic coumn generation. Dre et a. [13] determine routes for concrete veces in the first stage, and then compute routes for drivers, based on the vece routes from the first stage. They use the same arge neighbourhood agorithm in both stages. Cheung et a. [5] aso describe a picup-and-deivery probem, but for short-distance container transport, using drivers, tractors, and semi-traiers. Their probem is soved by an attribute-decision mode. Finay, Kim et a. [16] deveop a simpe but very effective heuristic for synchronizing service teams that are transported by veces between tas ocations: The teams, the veces, and the net tass are stored in three ists, and in each iteration, a tripet (team, vece, tas) is seected from the ists, using a best-fit criterion. Then the ists are updated to refect the situation when the seected vece transports the seected team to the ocation of the seected tas. Note that a of these wors assume mandatory synchronization of objects. Ts means that, to fufi tass, it is aways required to use a or severa object types (veces, drivers, equipment). In the VRPTT, by contrast, using traiers and performing transspments are optiona operations. Ts adds an additiona degree of freedom to the probem. Moreover, none of the above papers considers a probem where a five types of synchronization described above are reevant, as is the case for the VRPTT. 3

4 3 A networ representation In contrast to standard vece routing probems, for wch an adequate networ representation is straightforward, ts is not the case for the VRPTT, where the foowing critica modeing questions must be addressed: How to ensure that a traier is accompanied by a compatibe orry on an arc, that is, how to synchronize the movements of veces? How to synchronize the visiting times of veces at transspment ocations? How to baance the oad transfer amounts of veces echanging oad? A decouping, transspment or recouping operation is defined by (i) the ocation where the operation taes pace, (ii) the point in time when the operation begins, (iii) the passive vece (traier), wch provides capacity, that is, may receive oad, and/or may be decouped from or (re)couped to a orry, (iv) the active vece (orry), wch requests capacity, that is, may transfer oad, and/or may decoupe or (re)coupe a traier, and (v) the amount of oad transferred into the passive vece, wch is zero if ony decouping or recouping is performed. To answer the above modeing questions and hande the ogic of transspment operations, a space-timevece cass-operation networ D = (V, A) with an associated set F of veces (the feet), is proposed. The feet F is partitioned into two sets F L F T of casses of veces. F L is the set of orry casses, and F T is the set of traier casses. F denotes the number of veces of cass. For a F, q is the capacity of a vece of cass. Henceforth, is used to denote orry casses, and is used to denote traier casses. For a orry cass F L, C() is the set of compatibe casses of traiers, that is, the set of casses of traiers that a orry of cass can pu; for a traier cass F T, it is the set of compatibe casses of orries, that is, the set of casses of orries that can pu a traier of cass. τ is the oad transfer time per unit of oad, wch is assumed to be the same for a vece casses. It is assumed for simpicity that any orry can visit any customer, and that any traier can visit any traier customer. Each verte in V corresponds to a ocation in space, an absoute and/or reative period of time, a type of operation, and a cass of passive vece. L = {Depot} L C L I is the set of reevant rea-word ocations. L C = L CL L CLT is the set of customer ocations, wch is partitioned into L CL, the set of orry customers, and L CLT, the set of traier customers. L I is the set of pure transspment ocations. For each i V, L(i) L denotes the ocation corresponding to i. s i is the suppy of verte i, wch is zero for depot and transspment vertices. For S V, s S := i S s i. There is one start depot verte o and one end depot verte e. For each customer, there is one customer verte. Let V C = V CL V CLT be the set of customer vertices, where V CL is the set of orry customers, and where V CLT is the set of traier customers. Foowing Chao [4] and Scheuerer [21], it is assumed that the ocations corresponding to traier customers can aso be used for paring and transspment operations. Each verte i V has a time window [a i, b i ], where 0 a i b i T, and T is the ength of the panning horizon. o and e both have a time window of [0, T ]. Usuay, in the iterature, a i and b i respectivey denote the eariest and atest point in time when the service to be performed at i can start. In the VRPTT, a i has the same meaning, but, because of the oad-dependent oad transfer times, b i denotes the atest point in time when i must be eft, that is, the service at i must be finished no ater than b i. To provide a unified treatment, ts convention is adopted aso for the depot and customer vertices. For each combination of physica transspment ocation (pure transspment ocation or traier customer ocation) and traier cass, there are F verte tupes representing the operations decouping, transspment, and recouping. To be precise, for each transspment ocation and each traier cass, there are F tupes v d p, vt p, vr p, p = 1,..., F. V d is the set of decouping vertices, V t is the set of transspment vertices, and V r is the set of recouping vertices. V I = V d V t V r. The idea bend ts separation of transspment ocations and processes is the foowing: A verte v d p V d can ony be reached by a orry puing a traier of 4

5 the corresponding cass. The orry then eaves the verte singy, the traier moves on to v t p, the pertinent transspment verte. A verte v t p V t can ony be reached and eft by a singe orry and by a traier of the corresponding cass, where the atter comes from v d p. A verte v r p V r can ony be reached by a singe orry and by a traier of the corresponding cass, where the atter comes from the pertinent transspment verte v t p, and be eft by the orry puing that traier. At any verte of a transspment verte tupe, that is, aso at the decouping and the recouping verte, a orry visiting the verte can transfer oad to a traier. A veces are initiay at the start depot verte o and a veces that are used end their route at the end depot verte e. Lorries can visit a vertices of D, ecept for decouping and recouping vertices of incompatibe traiers. Traiers can ony visit their corresponding transspment vertices, traier customers, and, of course, the start and the end depot verte. F (i), for a i V I, is the cass of passive vece associated with i. For (i, j) A, F L and F T respectivey denote the set of orry and traier casses that can traverse (i, j). For a F, V (A ) is the set of vertices (arcs) that can be reached (traversed) by veces of cass. Moreover, for S V, et A (S) := {(i, j) A : i, j S}, δ (S) := {(i, j) A : i / S j}, and δ + (S) := {(i, j) A : i S j}. The arc set A contains the foowing eements: (o, i) and (i, e) for a customer vertices i V C (o, i) for a decouping and recouping vertices i V d V r (i, j) for a customer vertices i, j V C with i j (v d p, j) for a decouping vertices vd p V d and a vertices j V \ ({o} V d ) (v t p, j) for a transspment vertices vt p V t and a other vertices j V \ ({o, e} V d ) (v r p, j) for a recouping vertices vr p V r and a other vertices j V d \ {v d p : p p} V CLT {e} (i, j) for a i V CLT, j V I (i, j) for a i V CL, j V t V r Figure 2 visuaizes the verte types present in the networ and the arc types that can be traversed by orry-traier combinations, singe orries, and singe traiers. To eep the figure concise, there is ony one arc for each arc type present in the networ. For eampe, in the subfigure for LTCs, an arc from the eft traier customer to the right one is depicted to indicate that LTCs can freey move from any traier customer to any other traier customer (uness capacity or time window restrictions probit ts). The absence of an arc from the right traier customer to the eft one in the subfigure does not mean that there is no such arc. The arc from the recouping to the decouping verte in the subfigure for LTCs eists ony if the two transspment verte tupes (TVTs) represent the same traier cass, and if they ie at different ocations or the upper TVT has a gher p vaue. The indicated arcs to, from, and between transspment vertices in the subfigure for singe orries eist between a types of TVTs, that is, those representing the same or different traiers at the same or different ocations. For TVTs representing the same ocation, the arcs eist ony if the upper TVT has a gher p vaue. A cost coefficient c indicates the distance-dependent costs of a vece of cass F traversing arc (i, j) A. For the arcs emanating from the start depot verte, fied costs for using a vece of cass are aso contained in c oj. It is assumed that a arc costs and arc trave times are non-negative. For each arc (i, j) A, τ is the traversa time, wch is assumed to be the same for a vece casses. Moreover, it is assumed that τ oi a i for a i V \ {o}, and that b i + τ ie T for a i V \ {e}. 5

6 Lorry customers o Transspment verte tupes Decoupe Recoupe Transfer d Traier customers Networ vertices Arc types for orry-traier combinations Arc types for singe orries Arc types for singe traiers Figure 2: Networ vertices and arc types Lorries cannot traverse arcs (v d p, vt p ) and (vt p, vr p ). The corresponding traiers of cass can traverse these arcs, wch, consequenty, form the set A,s. An arc (v d p, vr p ) modes the possibiity that a orry can wait for a traier at a transspment ocation we another orry performs a oad transfer. If a orry currenty puing a traier wants to transfer oad to another traier at a certain ocation, the orry decoupes at a decouping verte at, transfers oad to, and re-coupes. 4 Two arc-variabe based formuations Both subsequent formuations are modifications of a formuation deveoped in Dre [11]. For both formuations, the foowing three fundamenta assumptions are made: (i) (ii) Each orry customer verte is visited by eacty one orry. Each traier customer verte is visited by eacty one orry and at most one traier. (iii) Each transspment verte is reached by at most one orry and at most one traier. These assumptions ensure that there is at most one transspment operation at each transspment verte, and that it is cear between wch orry and traier casses each of these operations is performed. In ts way, the three modeing issues mentioned at the beginning of Section 3 are addressed. With the networ defined as above, that is, with one transspment verte between a decouping and a recouping verte, these assumptions aow at most 3 F oad transfers per traier cass at each transspment ocation. These oad transfers can a be made into one and the same traier of cass, if ts traier is pued from v r p to vd,p+1 for p = 1,..., F 1, or into severa or a traiers of cass. 6

7 4.1 First formuation The foowing variabes are used: {0, 1} F, (i, j) A. = { 1, a vece of cass traverses arc (i, j) 0, otherwise i R0 + F, i V \ {o}. The amount of oad a vece of cass is carrying when reacng verte i. t i R 0 + i V \ {o}. The unique point in time when verte i is reached. The resuting formuation is: (VRPTT1): c min (1) F (i,j) A subject to = 1 i V C (2a) (h,i) A F L F (i) (h,i) A F (i) 1 F L F L F L i V I = F (i) i V d, (i, j) A F (i) (2c) (h,i) A F (i) i V t, (i, j) A F (i) (2d) (h,i) A = F (i) h i i V r, (h, i) A F (i) (2e) (h,i) A (2b) C( ) F L F T, i {o} V CLT, (i, j) A (2f) F (i) = C( ) F L i V r, (i, j) A F (i) (2g) ie (i,e) A F F (2h) = F, i V \ {o, e} (2i) (h,i) A (i,j) A = 1 = 1 i + i + s i j + j ) (i j (i) = 1 F i + i V CLT, F L, C(), (i, j) A A (3a) F (i) j i V d, C(F (i)), (i, j) A F (i), (i, j ) A (3b) 7

8 ) (i) = 1 F i + (i j F (i) j i V t, F L, (i, j) A F (i), (i, j ) A, j e (3c) ( ) = 1 F (i) i + i j F (i) j i V r, C(F (i)), (i, j) A F (i) A, j e (3d) = 1 i + s i j F L, i V C, (i, j) A, F T = (3e) = 1 i j = 1 F, i V CLT, (i, j) A, F T = 0 i + s i j F L, i V CLT, (i, j) A, F T (3f) (3g) C() F T = 1 j i F L, i V I, (i, j) A (3h) ( F (i) = 1 t i + F (i) j ) F (i) i τ b i i V r, (i, j) A F (i) (3i) = 1 t i + s i τ + τ t j F L, i V C, (i, j) A (3j) ( ) = 1 t i + i j τ + τ t j F L, i V I, (i, j) A (3) F (i) i t i + F (i) j i V t, (i, j) A F (i) (4a) ( ) F (i) j F (i) i τ + τ t j i V d V t, (i, j) A F (i) (4b) {0, 1} F, (i, j) A (5a) 0 i q F, i V \ {o} (5b) a i t i b i s i τ i V \ {o} (5c) (1) is the objective function, wch minimizes tota costs. Constraints (2) are those invoving ony fow variabes. (2a) (2g) are the routing synchronization constraints, and (2h) (2i) are vece-cass specific routing constraints. Constraints (3) are ogica impications ining routing and resource variabes. (3a) (3d) are oad synchronization constraints, (3e) (3h) are vece-cass specific oad update constraints, and (3i) (3) are time update constraints. Constraints (4) are resource update constraints not invoving fow variabes, and (5) determine the ranges of the variabes. (2a) are the usua customer covering constraints. (2b) ensure that each transspment verte is visited at most once by a corresponding traier. (2c) (2e) mae sure that a transspment verte i is ony visited by a orry if the corresponding traier visits ts verte, and, hence, that at most one orry visits a transspment verte. Observe that, for i V d, the ony arc in A F (i) eaving i is the arc to the subsequent transspment verte. Simiary, for i V t, the ony arc in A F (i) eaving i is the arc to the subsequent recouping verte. (2a) (2e) impy that each verte in V \ {o, e} is visited by at most one orry and one traier. (2f) and (2g) guarantee that a traier is pued by a compatibe orry if traverses a spacia arc. For the remaining constraints, note that a orry that brings a traier to a decouping verte and then moves on to the depot wi not transfer any oad at ts verte. Liewise, a orry that pus a traier from a recouping verte to the depot wi aso not transfer any oad. Moreover, a orry wi never visit a transspment verte i V t directy before moving to the depot; thus, as stated above, no arcs (i, e) with i V t eist. For each vece cass, (2h) imit the number of veces reacng the end depot to the number of veces of each cass. (2i) are the fow conservation constraints. 8

9 (3a) (3d) are the oad update constraints for both orries and traiers at traier customer and transspment vertices. (3a) state that the suppy of a traier customer is divided up arbitrariy between the orry and the traier visiting the customer. The net constraints, (3e), are the oad update constraints for the orries at customer vertices. Constraints (3f) state that, at a traier customer verte, no oad transfer is possibe. Ts is a sensibe requirement, because, as stated above, for each traier customer ocation, there are aso tupes of transspment vertices for each traier cass, and movements between vertices corresponding to the same physica ocation incur virtuay no costs and tae virtuay no time. Constraints (3g) are for the correct update of the oad variabes at traier customer vertices visited by a singe orry. Constraints (3h) mae sure that no oad transfer from a traier to a orry (passive to active vece) is possibe, respectivey, that the amount of oad transferred from a orry to a traier is non-negative. Without ts constraint, negative oad transfer times can resut. (3i) mae sure that the oad transfer at transspment vertices is finished by the end of the time window. Note that, because of (3a) (3d), it is sufficient to require (3i) for F (i). Moreover, it is sufficient to require (3i) for i V r, since then the time windows wi be maintained aso for the preceding transspment vertices. Constraints (3j) and (3) are for the timing update for orries on arcs emanating from customer vertices and at transspment vertices respectivey. (4a) mae sure that the oad of a traier does not decrease at transspment intermediate vertices not visited by a orry, and (4b) are for the timing update of traiers at their transspment vertices. The atter two constraint types are no impications, because these constraints can be fufied aso when the arc (i, j) A F (i) for i V d V t is not used. Synchronization constraints. In standard VRPs such as the capacitated VRP (CVRP) or the VRP with time windows (VRPTW), the customer covering constraints (2a) are the ony ogicay couping (ining, joint) ones. In the VRPTT, the cose interdependency between the veces must be deat with by severa additiona types of ogicay couping constraints, namey, (2b) (2g) and (3a) (3d). Infating resource variabe vaues. At the end depot verte, there is ony one oad variabe, e, for a veces of each cass F. Ts means that if one vece of a cass reaches the end depot fuy oaded (according to the variabe vaues), any other vece beonging to the same cass does so aso, even if ts vece actuay (in reaity) carries ess oad. The constraints aow to infate oad variabes, meaning that the oad variabe of a vece cass at the vertices on a tour may be set gher than the rea oad, we sti not eceeding the rea vece capacity. In ts way, the differences between the vaues of the oad variabes aong a tour refect the rea situation, so that the correct synchronization of oad transfer amounts and visit times, wch depends on differences in oad variabe vaues, remains possibe. Ts technique of infating the vaues of resource variabes aows to avoid introducing routing and resource variabes for each vece, and, when using variabes for vece casses instead of individua veces, maes the creation of start and end depot vertices for each vece unnecessary. 4.2 Second formuation The second formuation is based on the same networ as the first formuation, and it uses the same type of binary variabes. However, foowing an idea introduced independenty by van E [24] and Maffioi and Sciomachen [18], it uses different types of continuous resource variabes: R0 + F, (i, j) A. If the arc (i, j) is used by a vece of cass, the amount of oad the vece is carrying when eaving verte i; zero otherwise. t R 0 + (i, j) A. If the arc (i, j) is used by any vece, the unique point in time when verte i is eft; zero otherwise. The resuting formuation is: 9

10 (VRPTT2): (1) subject to (2) and ( ) + s i + + (h,i) A (i,j) A i V CLT (6a) (h,i) A F L C(F (i)) + F L (h,i) A + C(F (i)) (h,i) A F L (h,i) A F T + F (i) C(F (i)) F (i) h i (h,i) A F (i) (i,j ) A + + F L (i,j ) A F L F (i) h i (h,i) A F (i) (i,j) A C(F (i)) F L F L F T F (i) i V d (6b) (i,j) A F (i) F (i) i V t (6c) (i,j) A F (i) + F (i) i V r (6d) ( ) + s i i V CL (6e) (i,j) A F T (h,i) A F L (h,i) A F T ( ) + s i F L (i,j) A F L i V CLT (6f) (i,j) A s i F T h i (h,i) A F L i V CLT (6g) (i,j) A i V I (6h) (h,i) A F (i) F (i) i V t (6i) (h,i) A F (i) (i,j) A F (i) (h,i) A ( t + τ + s i τ t t + τ F (i) F L ) (h,i) A F (i) + F (i) (i,j) A F (i) (i,j) A F (i) (h,i) A F (i) i V C (6j) τ t i V I (6) (i,j) A F (i) t = t i V d (6) (i,j) A F (i) (i,j ) A\A F (i) t (i,j) A F (i) b i 1 t (i,j ) A\A F (i) t + τ F L (i,j ) A\A F (i) i V t (6m) (h,i) A\A F (i) F L 10

11 + (h,i) A \A F (i) F L (h,i) A\A F (i) (i,j ) A \A F (i) F L τ t i V t (6n) (i,j) A F (i) t + τ + τ (h,i) A (i,j) A t (i,j) A F L F L F L i V r {0, 1} F, (i, j) A (7a) 0 q F, (i, j) A (7b) ( a i + s i τ ) t b i (i, j) A, i {o} V C (6o) (7c) F L F L a i t b i i V I, (i, j) A \ A F (i),s (7d) F L F L a i F (i) t b i F (i) i V d V t, (i, j) A F (i) (7e) t ie b e τ ie (i, e) A (7f) Constraints (6a) (6i) and (6j) (6n) are for the update of the oad and time variabes respectivey; constraints (7) specify the ranges of the variabes. Constraints (6a) (6h) correspond to (3a) (3h) in formuation (VRPTT1). Simiary, (6i) corresponds to (4a), and (6j) (6) correspond to (3j) (3). (6) (6o) ensure the correct update of the timing variabes at decouping, transspment intermediate, and recouping vertices respectivey. The corresponding constraint to (3i), wch maes sure that the oad transfer at transspment vertices is finished by the end of the time window, is (7c). (7b) (7e) coupe the fow and resource variabes. These constraints aow to avoid impications in formuation (VRPTT2). 4.3 Vaid inequaities The foowing vaid inequaities were used in the computationa eperiments described in the net section. Lorry fow cut: ie n L,min (i,e) A F L (8) Ts inequaity requires that the fow of orries into the end depot be at east equa to the minimum number of necessary orries, n L,min. As the tota customer suppy must be brought to the depot, the inequaity is vaid. A ower bound on n L,min can be determined as foows. A possibe orry-traier combinations and singe orries are sorted by non-increasing tota capacity and the capacities are summed up unti the sum eceeds the tota customer suppy. For each summand, n L,min is increased by one. As traiers are usuay compatibe with more than one orry, a orry-traier combinations that are no onger possibe because the respective orry or traier or both has/have aready been used with another orry or traier must be deeted during the summation. A simiar cut is possibe for traiers, but was not usefu in the computationa eperiments described beow. 11

12 Connectivity cuts: i j F, S V \ {e}, (i, j) A, i S (9) (i,j ) δ (S),i j These cuts require that, for each vece cass and for each subset of the verte set not containing the end depot verte, if a vece of ts cass uses an arc (i, j) whose tai is in the subset, there must be another arc (i, j ) entering ts subset whose tai must not be j. Ts is impied by the fow conservation constraints (2h) and (2i). An eact procedure for the separation of vioated connectivity cuts wors as foows. For each > 0, a maimum fow probem from j to e is soved in the support graph where the arc capacities are equa to the vaues of the fow variabes of vece cass, but where verte i, and, hence, a arcs incident to i, are deeted. Ts yieds a cut with j on one side and e on the other. If the vaue of the maimum fow is ess than, a vioated connectivity cut has been found. Ts procedure is correct, because in any feasibe soution, > 0 means = 1, and i is on the same side as j in any j-e-cut, because each verte is visited by each vece cass at most once. In other words, there is no feasibe soution where i is on the unique path of a vece of cass from j to e, if arc (i, j) is used by a vece of cass. Ts means that there must be a fow of at east from the j side of the cut to the e side of the cut, but ts fow must not pass through i. Cuts (9) can sti be ifted. If one of the incident vertices i, j of an arc (i, j) is a recouping verte, the corresponding decouping and intermediate vertices cannot be visited any more, and, hence, any potentia cut arc emanating from or eading to such a verte can be discarded. If i or j is an intermediate verte, the same hods for arcs incident to the corresponding decouping verte. κ-path cuts: κ-path cuts are we-nown vaid inequaities for the VRP(TW), cf. Koh et a. [17]. In the VRPTW with homogeneous feet (in particuar, without traiers), the idea of κ-path cuts for κ 1 is that, for any customer subset S for wch at east κ(s) veces are necessary to serve a customers in the subset, at east κ veces must visit ts subset, and, consequenty, the fow into ts subset must be at east κ. 2-path cuts for heterogeneous feet: Koh et a. [17] have very successfuy used 2-path cuts in a branch-and-price-and-cut agorithm for the VRPTW. For vece routing probems with heterogeneous feet as we as for the VRPTT, 2-path cuts can be generaized as foows. For each subset S V C, et κ (S) be a ower bound on the minima number of routes of veces of cass F necessary to coect the compete suppy of the customers in S when no transspments are aowed and no vece of a different cass is used (that is, when it is assumed that each vece cass can visit each customer and that traiers can visit customers without being pued by a orry). For eampe, if there are no time windows, the bound s S /q can be used. Then, the foowing inequaity is vaid: S V C. (10) { F :κ (S) 2} (i,j) δ (S) { F :κ (S)=1} (i,j) δ (S) The vaidity of (10) foows from the customer covering constraints (2a), the oad update constraints (3a) (3h), and the vece capacity constraints (5b). κ-path cuts for heterogeneous feet for genera κ 2: Setting κ := min F {κ (S)} for a S V C, a κ-path cut for genera κ 2 can be written as foows: κ S V C (11) F (i,j) δ (S) 12

13 Note that (10) and (11) require summation over a F, that is, over a orry and traier casses, and that (10) as we as (11) are vaid for sets S containing an arbitrary number of orry and/or traier customers. Lifted generaized arge mutistar cuts: Yaman [25] has shown that the foowing cuts are vaid for heterogeneous feet VRPs: F ξ i (i,j) δ (S) s S + F sj S V C, (12) (i,j) δ + (S) where ξ i := min{q s i, s S + ma (i,j) δ + (S){s j}}. Essentiay, ts type of cut requires that the remaining capacity of the veces visiting customers in S be at east equa to the suppy of the customers in S pus the suppy of those customers (if any) visited immediatey after eaving S. Symmetry-breaing cuts: A traier routes visiting the same ocations in the same order and receiving the same amount of oad at each ocation are equivaent, irrespective of wch tupe of transspment vertices is used at a certain ocation. That is, two otherwise identica traier routes r 1 and r 2, where r 1 uses the transspment vertices v d p 1, v t p 1, v r p 1 and r 2 uses v d p 2, v t p 2, v r p 2, are equivaent and incur the same costs, irrespective of whether or not p 1 = p 2, since they describe the same movements in reaity. Such symmetries can be removed by the foowing cuts: F L i, i V d, L(i) = L(i ), F (i) = F (i ), p(i) = p(i ) 1 (13) F L These cuts require that a transspment verte tupe with order number p is ony visited if a preceding transspment verte tupes, that is, those with ower order number, are aso visited. If two traiers 1 and 2 of the same cass use the same transspment ocation, simiar symmetries arise: It does not mae a difference whether 1 uses the transspment verte tupe with order number p 1 and 2 uses the tupe with order number p 2 or the other way round. To remove such symmetries, the foowing cuts can be used in formuation (VRPTT1): t i t i i, i V d, L(i) = L(i ), F (i) = F (i ), p(i) = p(i ) 1 (14) These cuts require that the service at the decouping verte of a transspment verte tupe with order number p starts not earier than the service at the decouping verte of the transspment verte tupe with order number p 1. For formuation (VRPTT2), the foowing cuts can be used instead of (14): t (b i + 1) F (i) (i,j) A F (i) ti j (b i + 1)F (i ) i j (i,j ) A F (i ) i, i V d, L(i) = L(i ), F (i) = F (i ), p(i) = p(i ) 1 (15) Some remars on vaid inequaities foow. The orry fow, connectivity, and 2-path cuts were aready used in Dre [11]. Badacci et a. [2] propose a ifting for the κ-path cuts of Koh et a. [17], but ts cannot be used for arc variabe formuations. Desauniers et a. [10] have introduced generaized κ-path cuts for the VRPTW with homogeneous feet. Athough these can be generaized to heterogeneous feet, they were not used in the computationa eperiments described beow, because the wide time windows in the test instances used essentiay made these cuts equivaent to cuts (11). Ascheuer et a. [1] have introduced, for the asymmetric TSP with time windows, the so-caed infeasibe path constraints. These cuts are vaid aso for the VRPTT, but are not usefu when time windows are non-restrictive. Finay, Pessoa et a. [19] describe cuts for a heterogeneous feet VRP, but these are based on a different formuation. 13

14 5 Soution approaches Formuation (VRPTT1), the origina formuation, contains a ot of ogica constraints (impications). To dea with these, the foowing five soution approaches were impemented in C++, using Visua C and IBM Iog Cpe Concert Technoogy, version 12.2: (i) Formuation (VRPTT1), inearizing the impications using taiored Big-M vaues, taing into account the vece capacities and the ength of the panning horizon. (ii) (iii) (iv) Formuation (VRPTT2), thereby avoiding ogica constraints atogether. Formuation (VRPTT1), empoying a Cpe feature that aows entering impications directy as such and etting the sover automaticay hande them. How ts is done eacty, though, is not documented. Formuation (VRPTT1), appying the combinatoria Benders cuts introduced by Codato and Fischetti [6] and successfuy used by Cortés et a. [7]. The basic idea of ts approach is to decompose the probem into a master probem containing ony the variabes and the constraints invoving ony these variabes, and a subprobem containing ony the continuous variabes and a constraints not invoving variabes. Then, after soving the LP reaation of the master probem, a impications activated by variabes that are one or zero in the soution of the LP reaation are added to the subprobem. If the subprobem, wch has no objective function, is feasibe, then evidenty the variabes soution to the master probem is feasibe and optima for the compete probem. Otherwise, at east one of the integer variabes must change its vaue. Ts is enforced by identifying an incusion-minima infeasibe subset (by means of a buit-in Cpe function) of the subprobem constraints and adding the constraint (1 ) S 1 + S 0 1 (16) to the master, where the set S 1 (S 0 ) contains those variabes that activate a constraint when they tae vaue one (zero) and do tae ts vaue in the current LP soution. (v) Formuation (VRPTT1), using deferred consideration of ogica constraints. Ts wors as foows. At the root node of the branch-and-bound tree, the LP reaation of a reaed probem is soved, where a impications are ignored. Then, at each subsequent node, it is checed for wch variabes the ower bound is greater than zero. A such variabes wi have vaue one in each integer feasibe soution (if any) at the subtree rooted at ts node. Simiary, it is checed wch variabes have an upper bound of ess than one. A such variabes wi have vaue zero in each integer feasibe soution at the subtree rooted at ts node. Then, instead of having inearized constraints containing Big-M or adding combinatoria Benders cuts, ony the right hand sides of the activated impications, that is, the respective inear constraints to the right of in (3), are added as ocay vaid constraints to the LP at the respective node. A approaches use the cuts described in Section 4.3. For formuation (VRPTT2), cuts (15) are used instead of (14). The orry fow cut as we as both types of symmetry-breaing cuts are directy added to the formuation as static cuts. The connectivity cuts are dynamicay separated in the course of the agorithm. As for the κ-path and mutistar cuts, it turned out that a instances that coud be soved with or without them are sma enough to aow compete enumeration of a cuts e ante. Therefore, a cuts (11) and (12) are aso added as static cuts. As brancng strategy, the foowing four-stage strategy outperformed the defaut best-bound singe variabe brancng strategy provided by Cpe for approaches (i), (ii), and (iv). (i) Branch on the sum of a arc variabes eaving the start depot verte. (ii) Branch on the sum of a orry cass arc variabes eaving the start depot verte. (iii) Branch on the sum of a traier cass arc variabes eaving the start depot verte. (iv) Use the defaut Cpe brancng strategy. For approaches (iii) and (v), the Cpe defaut brancng performed better. Strong brancng was used in a cases. 14

15 6 Computationa eperiments In ts section, the resuts of an etensive computationa study with the approaches described in the previous section are presented. 6.1 Test instances Dre [11] has created a set of VRPTT test instances structured to resembe the situation in raw mi coection. These use two orry and two traier casses as specified in Tabe 1. Capacity Fied costs Costs per m Compatibe traier casses Lorry cass 1 10, , 2 Lorry cass 2 15, Traier cass 1 10, Traier cass 2 15, Tabe 1: Feet data The customer and transspment ocations are randomy seected on a m grid with the depot ocated in the centre. The resuting Eucidean distances between each pair of vertices are mutipied by a distance factor of 1.3. The customer suppies are chosen randomy from [1,000; 10,000]. As oad transfer time, two minutes per 1,000 units of suppy are assumed. The ength of the panning horizon is 12 hours (1,320 minutes). A customers and transspment ocations have a time window of [0; 1,320]. Such non-restrictive time windows represent the situation in raw mi coection where ony very few customers or transspment ocations actuay have time windows. The test instances were created with enough veces of each cass such that the compete suppy coud be coected with either ony orries of cass 1 and traiers of cass 2 or ony orries of cass 2 and traiers of cass 1. With these parameters, a set of so-caed y z instances was created, where 2 20 stands for the number of customers, 2 y 20 for the number of potentia transspment ocations, and 4 z 24 for the number of veces. It is aways = y, and there are aways /2 orry customers and /2 traier customers. y/2 potentia transspment ocations correspond to the traier customer ocations, and y/2 are pure transspment ocations. For each of the four vece casses, there are z/4 veces. Ts means that an y z instance has 1++3 F T F y+1 vertices: one for the start depot, for the customers, three for each traier at each of the y potentia transspment ocations to represent decouping, transfer, and recouping, and one for the end depot. Tabe 2 shows detais on instance sizes. Instance No. of Vertices Arcs Variabes Constraints Cuts cass instances Binary Continuous Symmetry- κ-path Mutistar (VRPTT1) (VRPTT2) (VRPTT1) (VRPTT2) breaing 1 / 2 (avg.) / ,340 2,384 2,176 0 / ,516 2, ,116 7,408 6,344 8 / ,074 1, ,976 5,299 4,630 0 / ,390 5, ,216 16,597 13, / ,008 10, ,344 29,444 24, / Tabe 2: Instance sizes 6.2 Agorithm setup and system parameters Some remars on the setup of the agorithms are appropriate. In approach (iv), combinatoria Benders cuts can either be separated ony when the LP soution is integer or aso when some variabes are fractiona. Moreover, it is possibe to add the cuts as goba or oca ones, and either ony one or severa vioated cuts can be added in each iteration. Tests showed that the 15

16 best setup is to separate cuts aso when there are fractiona variabes, and to add one vioated cut per iteration as a goba one. Simiary, in approach (v), it is possibe to chec for newy fied variabes either ony when the LP soution is integer in the variabes or after each soution of the LP reaation. Here, tests showed that the former variant is sighty but consistenty better. In a approaches, connectivity cuts are added ony if the vioation is at east 0.3. In each iteration, a connectivity cuts that reach ts vioation threshod are added. Tests with adding ony one vioated connectivity cut per vece cass and iteration yieded a worse performance. The maimum fow probems for the separation of the connectivity cuts are soved with the Edmonds-Karp agorithm provided by the Boost Graph ibrary (boost.org). The resuts presented subsequenty were obtained on a computer with a 2.80 GHz CPU and 16 GB of main memory for instance casses , and on a macne with a 2.83 GHz CPU and 8 GB of main memory for instance cass 8 8 8, both running Win XP Sp 2, 64-bit. A CPU time imit of 10,800 seconds was set. 6.3 Evauation of the vaid inequaities Tabe 3 reports the average reative increase in the root ower bound for formuations (VRPTT1) and (VRPTT2) when the different casses of vaid inequaities are used. As can be seen, the root ower bounds of formuation (VRPTT2) are significanty gher than those of (VRPTT1). Moreover, the tabe shows that the orry fow cut is very important: It cuts off fractiona soutions where orries are used ony partiay, and the resuting better consideration of the fied costs raises the ower bounds consideraby. Formuation: (VRPTT1) (VRPTT2) (VRPTT2) Average reative increase compared to: (VRPTT1), no cuts (VRPTT1), no cuts (VRPTT2), no cuts No cuts n.a % n.a. Ony orry fow cut % % 36.8 % Ony connectivity cuts 67.8 % % 3.3 % Ony κ-path cuts 7.6 % % 0.4 % Ony mutistar cuts 17.7 % % 0.3 % Tabe 3: Comparison of root ower bounds (instances ) 6.4 Comparison of the different soution approaches Tabe 4 presents the computationa resuts obtained with the different approaches and the described settings. Coumns Optima and Feasibe respectivey indicate the percentage of instances that coud be soved optimay and for wch a feasibe soution coud be computed witn 10,800 seconds. Coumn Gap specifies the average reative gap between the upper and the ower bound, UB and LB respectivey, upon termination of the optimization, for those instances for wch a feasibe soution coud be computed, as (UB LB)/LB. Coumn CPU time indicates the average running time over a instances of a cass, whether or not a feasibe soution coud be found. Coumn None better reports the percentage of instances that no approach soved better. No approach soves an instance i better than a certain approach A if either A soves i to optimaity and no other approach that aso soves i optimay is faster, or if no approach finds the optima soution witn the time imit and no approach finds a better feasibe soution than A. Coumn Best indicates the percentage of instances that were soved best with the respective approach. An approach A is best in soving an instance i if A finds the optima soution faster than any other approach, or if it finds a better soution than any other approach. To account for measuring inaccuracies, computation times in seconds were rounded to integer mutipes of 10 seconds when computing the vaues in coumns None better and Best. Finay, coumn No. B & B nodes gives the average number of nodes in the branch-and-bound tree upon termination of the optimization, irrespective of the soution status. 16

17 Approach Instance Optima Feasibe Gap CPU time None better Best No. B & B cass [%] [%] [%] [secs.] [%] [%] nodes (i): (VRPTT1), inearizing the impications using taiored Big-M vaues , , , , , , ,221 Overa , ,382 (ii): (VRPTT2), avoiding ogica constraints atogether , , , Overa , (iii): (VRPTT1), empoying a Cpe feature that aows entering impications directy , , , , Overa , ,478 (iv): (VRPTT1), appying the combinatoria Benders cuts , , , , , , ,060 Overa , ,997 (v): (VRPTT1), using deferred consideration of ogica constraints , , , , , , , ,642 Overa , ,310 Tabe 4: Computationa resuts The foowing observations can be made in Tabe 4: Overa, there is no approach that outperforms a others with respect to every criterion. However, approach (v) performs best in four out of seven criteria, and, for one additiona criterion, there is no better one. Approach (iii) ceary performs worst. Instance casses were not even tried with ts approach. It seems that the automatic handing of impications buit into Cpe is inadequate for the probem structure at hand. Athough formuation (VRPTT2) has a significanty stronger LP reaation than formuation (VRPTT1), approach (ii) soves fewer instances to optimaity than approach (i), and it requires sighty more computation time on average. Apparenty, the soution of the LP reaations taes onger. Approach (v) soves the ghest number of instances to optimaity, 75 out of 109, wch corresponds to 68.8 %. It is the ony approach to sove an instance of cass to optimaity. However, it fais to find a feasibe soution for five instances, wch is worse than approaches (i), (ii), and (iv). There is one particuary hard instance in cass 4 4 8, wch gets soved optimay ony by approach (ii). The overa optimaity gap is comparativey gh for approach (i). In particuar, though feasibe soutions are found for a but two instances of cass 8 8 8, the gap for ts cass is amost twice as gh as for the other approaches. 17

XSAT of linear CNF formulas

XSAT of linear CNF formulas XSAT of inear CN formuas Bernd R. Schuh Dr. Bernd Schuh, D-50968 Kön, Germany; bernd.schuh@netcoogne.de eywords: compexity, XSAT, exact inear formua, -reguarity, -uniformity, NPcompeteness Abstract. Open

More information

Asynchronous Control for Coupled Markov Decision Systems

Asynchronous Control for Coupled Markov Decision Systems INFORMATION THEORY WORKSHOP (ITW) 22 Asynchronous Contro for Couped Marov Decision Systems Michae J. Neey University of Southern Caifornia Abstract This paper considers optima contro for a coection of

More information

Problem set 6 The Perron Frobenius theorem.

Problem set 6 The Perron Frobenius theorem. Probem set 6 The Perron Frobenius theorem. Math 22a4 Oct 2 204, Due Oct.28 In a future probem set I want to discuss some criteria which aow us to concude that that the ground state of a sef-adjoint operator

More information

A Brief Introduction to Markov Chains and Hidden Markov Models

A Brief Introduction to Markov Chains and Hidden Markov Models A Brief Introduction to Markov Chains and Hidden Markov Modes Aen B MacKenzie Notes for December 1, 3, &8, 2015 Discrete-Time Markov Chains You may reca that when we first introduced random processes,

More information

MARKOV CHAINS AND MARKOV DECISION THEORY. Contents

MARKOV CHAINS AND MARKOV DECISION THEORY. Contents MARKOV CHAINS AND MARKOV DECISION THEORY ARINDRIMA DATTA Abstract. In this paper, we begin with a forma introduction to probabiity and expain the concept of random variabes and stochastic processes. After

More information

Combining reaction kinetics to the multi-phase Gibbs energy calculation

Combining reaction kinetics to the multi-phase Gibbs energy calculation 7 th European Symposium on Computer Aided Process Engineering ESCAPE7 V. Pesu and P.S. Agachi (Editors) 2007 Esevier B.V. A rights reserved. Combining reaction inetics to the muti-phase Gibbs energy cacuation

More information

Separation of Variables and a Spherical Shell with Surface Charge

Separation of Variables and a Spherical Shell with Surface Charge Separation of Variabes and a Spherica She with Surface Charge In cass we worked out the eectrostatic potentia due to a spherica she of radius R with a surface charge density σθ = σ cos θ. This cacuation

More information

Approximated MLC shape matrix decomposition with interleaf collision constraint

Approximated MLC shape matrix decomposition with interleaf collision constraint Approximated MLC shape matrix decomposition with intereaf coision constraint Thomas Kainowski Antje Kiese Abstract Shape matrix decomposition is a subprobem in radiation therapy panning. A given fuence

More information

Uniprocessor Feasibility of Sporadic Tasks with Constrained Deadlines is Strongly conp-complete

Uniprocessor Feasibility of Sporadic Tasks with Constrained Deadlines is Strongly conp-complete Uniprocessor Feasibiity of Sporadic Tasks with Constrained Deadines is Strongy conp-compete Pontus Ekberg and Wang Yi Uppsaa University, Sweden Emai: {pontus.ekberg yi}@it.uu.se Abstract Deciding the feasibiity

More information

Cryptanalysis of PKP: A New Approach

Cryptanalysis of PKP: A New Approach Cryptanaysis of PKP: A New Approach Éiane Jaumes and Antoine Joux DCSSI 18, rue du Dr. Zamenhoff F-92131 Issy-es-Mx Cedex France eiane.jaumes@wanadoo.fr Antoine.Joux@ens.fr Abstract. Quite recenty, in

More information

8 Digifl'.11 Cth:uits and devices

8 Digifl'.11 Cth:uits and devices 8 Digif'. Cth:uits and devices 8. Introduction In anaog eectronics, votage is a continuous variabe. This is usefu because most physica quantities we encounter are continuous: sound eves, ight intensity,

More information

Componentwise Determination of the Interval Hull Solution for Linear Interval Parameter Systems

Componentwise Determination of the Interval Hull Solution for Linear Interval Parameter Systems Componentwise Determination of the Interva Hu Soution for Linear Interva Parameter Systems L. V. Koev Dept. of Theoretica Eectrotechnics, Facuty of Automatics, Technica University of Sofia, 1000 Sofia,

More information

Lobontiu: System Dynamics for Engineering Students Website Chapter 3 1. z b z

Lobontiu: System Dynamics for Engineering Students Website Chapter 3 1. z b z Chapter W3 Mechanica Systems II Introduction This companion website chapter anayzes the foowing topics in connection to the printed-book Chapter 3: Lumped-parameter inertia fractions of basic compiant

More information

T.C. Banwell, S. Galli. {bct, Telcordia Technologies, Inc., 445 South Street, Morristown, NJ 07960, USA

T.C. Banwell, S. Galli. {bct, Telcordia Technologies, Inc., 445 South Street, Morristown, NJ 07960, USA ON THE SYMMETRY OF THE POWER INE CHANNE T.C. Banwe, S. Gai {bct, sgai}@research.tecordia.com Tecordia Technoogies, Inc., 445 South Street, Morristown, NJ 07960, USA Abstract The indoor power ine network

More information

First-Order Corrections to Gutzwiller s Trace Formula for Systems with Discrete Symmetries

First-Order Corrections to Gutzwiller s Trace Formula for Systems with Discrete Symmetries c 26 Noninear Phenomena in Compex Systems First-Order Corrections to Gutzwier s Trace Formua for Systems with Discrete Symmetries Hoger Cartarius, Jörg Main, and Günter Wunner Institut für Theoretische

More information

Approximated MLC shape matrix decomposition with interleaf collision constraint

Approximated MLC shape matrix decomposition with interleaf collision constraint Agorithmic Operations Research Vo.4 (29) 49 57 Approximated MLC shape matrix decomposition with intereaf coision constraint Antje Kiese and Thomas Kainowski Institut für Mathematik, Universität Rostock,

More information

12.2. Maxima and Minima. Introduction. Prerequisites. Learning Outcomes

12.2. Maxima and Minima. Introduction. Prerequisites. Learning Outcomes Maima and Minima 1. Introduction In this Section we anayse curves in the oca neighbourhood of a stationary point and, from this anaysis, deduce necessary conditions satisfied by oca maima and oca minima.

More information

Stochastic Complement Analysis of Multi-Server Threshold Queues. with Hysteresis. Abstract

Stochastic Complement Analysis of Multi-Server Threshold Queues. with Hysteresis. Abstract Stochastic Compement Anaysis of Muti-Server Threshod Queues with Hysteresis John C.S. Lui The Dept. of Computer Science & Engineering The Chinese University of Hong Kong Leana Goubchik Dept. of Computer

More information

Lower Bounds for the Relative Greedy Algorithm for Approximating Steiner Trees

Lower Bounds for the Relative Greedy Algorithm for Approximating Steiner Trees This paper appeared in: Networks 47:2 (2006), -5 Lower Bounds for the Reative Greed Agorithm for Approimating Steiner Trees Stefan Hougard Stefan Kirchner Humbodt-Universität zu Berin Institut für Informatik

More information

Explicit overall risk minimization transductive bound

Explicit overall risk minimization transductive bound 1 Expicit overa risk minimization transductive bound Sergio Decherchi, Paoo Gastado, Sandro Ridea, Rodofo Zunino Dept. of Biophysica and Eectronic Engineering (DIBE), Genoa University Via Opera Pia 11a,

More information

Optimal Control of Assembly Systems with Multiple Stages and Multiple Demand Classes 1

Optimal Control of Assembly Systems with Multiple Stages and Multiple Demand Classes 1 Optima Contro of Assemby Systems with Mutipe Stages and Mutipe Demand Casses Saif Benjaafar Mohsen EHafsi 2 Chung-Yee Lee 3 Weihua Zhou 3 Industria & Systems Engineering, Department of Mechanica Engineering,

More information

Integrating Factor Methods as Exponential Integrators

Integrating Factor Methods as Exponential Integrators Integrating Factor Methods as Exponentia Integrators Borisav V. Minchev Department of Mathematica Science, NTNU, 7491 Trondheim, Norway Borko.Minchev@ii.uib.no Abstract. Recenty a ot of effort has been

More information

Schedulability Analysis of Deferrable Scheduling Algorithms for Maintaining Real-Time Data Freshness

Schedulability Analysis of Deferrable Scheduling Algorithms for Maintaining Real-Time Data Freshness 1 Scheduabiity Anaysis of Deferrabe Scheduing Agorithms for Maintaining Rea-Time Data Freshness Song Han, Deji Chen, Ming Xiong, Kam-yiu Lam, Aoysius K. Mok, Krithi Ramamritham UT Austin, Emerson Process

More information

Theory and implementation behind: Universal surface creation - smallest unitcell

Theory and implementation behind: Universal surface creation - smallest unitcell Teory and impementation beind: Universa surface creation - smaest unitce Bjare Brin Buus, Jaob Howat & Tomas Bigaard September 15, 218 1 Construction of surface sabs Te aim for tis part of te project is

More information

Schedulability Analysis of Deferrable Scheduling Algorithms for Maintaining Real-Time Data Freshness

Schedulability Analysis of Deferrable Scheduling Algorithms for Maintaining Real-Time Data Freshness 1 Scheduabiity Anaysis of Deferrabe Scheduing Agorithms for Maintaining Rea- Data Freshness Song Han, Deji Chen, Ming Xiong, Kam-yiu Lam, Aoysius K. Mok, Krithi Ramamritham UT Austin, Emerson Process Management,

More information

Primal and dual active-set methods for convex quadratic programming

Primal and dual active-set methods for convex quadratic programming Math. Program., Ser. A 216) 159:469 58 DOI 1.17/s117-15-966-2 FULL LENGTH PAPER Prima and dua active-set methods for convex quadratic programming Anders Forsgren 1 Phiip E. Gi 2 Eizabeth Wong 2 Received:

More information

Capacity sharing among truck owners: A collaborative approach to overcome overloading

Capacity sharing among truck owners: A collaborative approach to overcome overloading Capacity sharing among truck owners: A coaborative approach to overcome overoading Arindam Debroy 1 Research Schoar debroyarindam1@gmai.com S. P. Sarmah 1 Professor 1 Department of Industria and Systems

More information

Discrete Applied Mathematics

Discrete Applied Mathematics Discrete Appied Mathematics 159 (2011) 812 825 Contents ists avaiabe at ScienceDirect Discrete Appied Mathematics journa homepage: www.esevier.com/ocate/dam A direct barter mode for course add/drop process

More information

APPENDIX C FLEXING OF LENGTH BARS

APPENDIX C FLEXING OF LENGTH BARS Fexing of ength bars 83 APPENDIX C FLEXING OF LENGTH BARS C.1 FLEXING OF A LENGTH BAR DUE TO ITS OWN WEIGHT Any object ying in a horizonta pane wi sag under its own weight uness it is infinitey stiff or

More information

Power Control and Transmission Scheduling for Network Utility Maximization in Wireless Networks

Power Control and Transmission Scheduling for Network Utility Maximization in Wireless Networks ower Contro and Transmission Scheduing for Network Utiity Maximization in Wireess Networks Min Cao, Vivek Raghunathan, Stephen Hany, Vinod Sharma and. R. Kumar Abstract We consider a joint power contro

More information

PHYS 110B - HW #1 Fall 2005, Solutions by David Pace Equations referenced as Eq. # are from Griffiths Problem statements are paraphrased

PHYS 110B - HW #1 Fall 2005, Solutions by David Pace Equations referenced as Eq. # are from Griffiths Problem statements are paraphrased PHYS 110B - HW #1 Fa 2005, Soutions by David Pace Equations referenced as Eq. # are from Griffiths Probem statements are paraphrased [1.] Probem 6.8 from Griffiths A ong cyinder has radius R and a magnetization

More information

More Scattering: the Partial Wave Expansion

More Scattering: the Partial Wave Expansion More Scattering: the Partia Wave Expansion Michae Fower /7/8 Pane Waves and Partia Waves We are considering the soution to Schrödinger s equation for scattering of an incoming pane wave in the z-direction

More information

DOCUMENT DE TRAVAIL

DOCUMENT DE TRAVAIL Pubié par : Pubished by: Pubicación de a: Facuté des sciences de administration 2325, rue de a Terrasse Pavion Paasis-Prince, Université Lava Québec (Québec) Canada G1V 0A6 Té. Ph. Te. : (418) 656-3644

More information

Partial permutation decoding for MacDonald codes

Partial permutation decoding for MacDonald codes Partia permutation decoding for MacDonad codes J.D. Key Department of Mathematics and Appied Mathematics University of the Western Cape 7535 Bevie, South Africa P. Seneviratne Department of Mathematics

More information

An Algorithm for Pruning Redundant Modules in Min-Max Modular Network

An Algorithm for Pruning Redundant Modules in Min-Max Modular Network An Agorithm for Pruning Redundant Modues in Min-Max Moduar Network Hui-Cheng Lian and Bao-Liang Lu Department of Computer Science and Engineering, Shanghai Jiao Tong University 1954 Hua Shan Rd., Shanghai

More information

A Novel Learning Method for Elman Neural Network Using Local Search

A Novel Learning Method for Elman Neural Network Using Local Search Neura Information Processing Letters and Reviews Vo. 11, No. 8, August 2007 LETTER A Nove Learning Method for Eman Neura Networ Using Loca Search Facuty of Engineering, Toyama University, Gofuu 3190 Toyama

More information

Traffic data collection

Traffic data collection Chapter 32 Traffic data coection 32.1 Overview Unike many other discipines of the engineering, the situations that are interesting to a traffic engineer cannot be reproduced in a aboratory. Even if road

More information

14 Separation of Variables Method

14 Separation of Variables Method 14 Separation of Variabes Method Consider, for exampe, the Dirichet probem u t = Du xx < x u(x, ) = f(x) < x < u(, t) = = u(, t) t > Let u(x, t) = T (t)φ(x); now substitute into the equation: dt

More information

MATH 172: MOTIVATION FOR FOURIER SERIES: SEPARATION OF VARIABLES

MATH 172: MOTIVATION FOR FOURIER SERIES: SEPARATION OF VARIABLES MATH 172: MOTIVATION FOR FOURIER SERIES: SEPARATION OF VARIABLES Separation of variabes is a method to sove certain PDEs which have a warped product structure. First, on R n, a inear PDE of order m is

More information

A NOTE ON QUASI-STATIONARY DISTRIBUTIONS OF BIRTH-DEATH PROCESSES AND THE SIS LOGISTIC EPIDEMIC

A NOTE ON QUASI-STATIONARY DISTRIBUTIONS OF BIRTH-DEATH PROCESSES AND THE SIS LOGISTIC EPIDEMIC (January 8, 2003) A NOTE ON QUASI-STATIONARY DISTRIBUTIONS OF BIRTH-DEATH PROCESSES AND THE SIS LOGISTIC EPIDEMIC DAMIAN CLANCY, University of Liverpoo PHILIP K. POLLETT, University of Queensand Abstract

More information

A. Distribution of the test statistic

A. Distribution of the test statistic A. Distribution of the test statistic In the sequentia test, we first compute the test statistic from a mini-batch of size m. If a decision cannot be made with this statistic, we keep increasing the mini-batch

More information

Asymptotic Properties of a Generalized Cross Entropy Optimization Algorithm

Asymptotic Properties of a Generalized Cross Entropy Optimization Algorithm 1 Asymptotic Properties of a Generaized Cross Entropy Optimization Agorithm Zijun Wu, Michae Koonko, Institute for Appied Stochastics and Operations Research, Caustha Technica University Abstract The discrete

More information

$, (2.1) n="# #. (2.2)

$, (2.1) n=# #. (2.2) Chapter. Eectrostatic II Notes: Most of the materia presented in this chapter is taken from Jackson, Chap.,, and 4, and Di Bartoo, Chap... Mathematica Considerations.. The Fourier series and the Fourier

More information

Efficiently Generating Random Bits from Finite State Markov Chains

Efficiently Generating Random Bits from Finite State Markov Chains 1 Efficienty Generating Random Bits from Finite State Markov Chains Hongchao Zhou and Jehoshua Bruck, Feow, IEEE Abstract The probem of random number generation from an uncorreated random source (of unknown

More information

MONOCHROMATIC LOOSE PATHS IN MULTICOLORED k-uniform CLIQUES

MONOCHROMATIC LOOSE PATHS IN MULTICOLORED k-uniform CLIQUES MONOCHROMATIC LOOSE PATHS IN MULTICOLORED k-uniform CLIQUES ANDRZEJ DUDEK AND ANDRZEJ RUCIŃSKI Abstract. For positive integers k and, a k-uniform hypergraph is caed a oose path of ength, and denoted by

More information

Algorithms to solve massively under-defined systems of multivariate quadratic equations

Algorithms to solve massively under-defined systems of multivariate quadratic equations Agorithms to sove massivey under-defined systems of mutivariate quadratic equations Yasufumi Hashimoto Abstract It is we known that the probem to sove a set of randomy chosen mutivariate quadratic equations

More information

Distribution Systems Voltage Profile Improvement with Series FACTS Devices Using Line Flow-Based Equations

Distribution Systems Voltage Profile Improvement with Series FACTS Devices Using Line Flow-Based Equations 16th NATIONAL POWER SYSTEMS CONFERENCE, 15th-17th DECEMBER, 010 386 Distribution Systems otage Profie Improvement with Series FACTS Devices Using Line Fow-Based Equations K. enkateswararao, P. K. Agarwa

More information

(Refer Slide Time: 2:34) L C V

(Refer Slide Time: 2:34) L C V Microwave Integrated Circuits Professor Jayanta Mukherjee Department of Eectrica Engineering Indian Intitute of Technoogy Bombay Modue 1 Lecture No 2 Refection Coefficient, SWR, Smith Chart. Heo wecome

More information

Tight Approximation Algorithms for Maximum Separable Assignment Problems

Tight Approximation Algorithms for Maximum Separable Assignment Problems MATHEMATICS OF OPERATIONS RESEARCH Vo. 36, No. 3, August 011, pp. 416 431 issn 0364-765X eissn 156-5471 11 3603 0416 10.187/moor.1110.0499 011 INFORMS Tight Approximation Agorithms for Maximum Separabe

More information

CS229 Lecture notes. Andrew Ng

CS229 Lecture notes. Andrew Ng CS229 Lecture notes Andrew Ng Part IX The EM agorithm In the previous set of notes, we taked about the EM agorithm as appied to fitting a mixture of Gaussians. In this set of notes, we give a broader view

More information

FRIEZE GROUPS IN R 2

FRIEZE GROUPS IN R 2 FRIEZE GROUPS IN R 2 MAXWELL STOLARSKI Abstract. Focusing on the Eucidean pane under the Pythagorean Metric, our goa is to cassify the frieze groups, discrete subgroups of the set of isometries of the

More information

Efficient Generation of Random Bits from Finite State Markov Chains

Efficient Generation of Random Bits from Finite State Markov Chains Efficient Generation of Random Bits from Finite State Markov Chains Hongchao Zhou and Jehoshua Bruck, Feow, IEEE Abstract The probem of random number generation from an uncorreated random source (of unknown

More information

Convergence Property of the Iri-Imai Algorithm for Some Smooth Convex Programming Problems

Convergence Property of the Iri-Imai Algorithm for Some Smooth Convex Programming Problems Convergence Property of the Iri-Imai Agorithm for Some Smooth Convex Programming Probems S. Zhang Communicated by Z.Q. Luo Assistant Professor, Department of Econometrics, University of Groningen, Groningen,

More information

Optimal islanding and load shedding

Optimal islanding and load shedding Optima isanding and oad shedding Pau Trodden, Waqquas Bukhsh, Andreas Grothey, Jacek Gondzio, Ken McKinnon March 11, 2011 Contents 1 Introduction 3 2 Optima oad shedding 3 2.1 AC OLS probem...............................

More information

Reliability: Theory & Applications No.3, September 2006

Reliability: Theory & Applications No.3, September 2006 REDUNDANCY AND RENEWAL OF SERVERS IN OPENED QUEUING NETWORKS G. Sh. Tsitsiashvii M.A. Osipova Vadivosto, Russia 1 An opened queuing networ with a redundancy and a renewa of servers is considered. To cacuate

More information

K a,k minors in graphs of bounded tree-width *

K a,k minors in graphs of bounded tree-width * K a,k minors in graphs of bounded tree-width * Thomas Böhme Institut für Mathematik Technische Universität Imenau Imenau, Germany E-mai: tboehme@theoinf.tu-imenau.de and John Maharry Department of Mathematics

More information

Stat 155 Game theory, Yuval Peres Fall Lectures 4,5,6

Stat 155 Game theory, Yuval Peres Fall Lectures 4,5,6 Stat 155 Game theory, Yuva Peres Fa 2004 Lectures 4,5,6 In the ast ecture, we defined N and P positions for a combinatoria game. We wi now show more formay that each starting position in a combinatoria

More information

Approximate Bandwidth Allocation for Fixed-Priority-Scheduled Periodic Resources (WSU-CS Technical Report Version)

Approximate Bandwidth Allocation for Fixed-Priority-Scheduled Periodic Resources (WSU-CS Technical Report Version) Approximate Bandwidth Aocation for Fixed-Priority-Schedued Periodic Resources WSU-CS Technica Report Version) Farhana Dewan Nathan Fisher Abstract Recent research in compositiona rea-time systems has focused

More information

Noname manuscript No. (will be inserted by the editor) Can Li Ignacio E. Grossmann

Noname manuscript No. (will be inserted by the editor) Can Li Ignacio E. Grossmann Noname manuscript No. (wi be inserted by the editor) A finite ɛ-convergence agorithm for two-stage convex 0-1 mixed-integer noninear stochastic programs with mixed-integer first and second stage variabes

More information

Nonlinear Analysis of Spatial Trusses

Nonlinear Analysis of Spatial Trusses Noninear Anaysis of Spatia Trusses João Barrigó October 14 Abstract The present work addresses the noninear behavior of space trusses A formuation for geometrica noninear anaysis is presented, which incudes

More information

Target Location Estimation in Wireless Sensor Networks Using Binary Data

Target Location Estimation in Wireless Sensor Networks Using Binary Data Target Location stimation in Wireess Sensor Networks Using Binary Data Ruixin Niu and Pramod K. Varshney Department of ectrica ngineering and Computer Science Link Ha Syracuse University Syracuse, NY 344

More information

Reliability Improvement with Optimal Placement of Distributed Generation in Distribution System

Reliability Improvement with Optimal Placement of Distributed Generation in Distribution System Reiabiity Improvement with Optima Pacement of Distributed Generation in Distribution System N. Rugthaicharoencheep, T. Langtharthong Abstract This paper presents the optima pacement and sizing of distributed

More information

arxiv: v1 [math.co] 17 Dec 2018

arxiv: v1 [math.co] 17 Dec 2018 On the Extrema Maximum Agreement Subtree Probem arxiv:1812.06951v1 [math.o] 17 Dec 2018 Aexey Markin Department of omputer Science, Iowa State University, USA amarkin@iastate.edu Abstract Given two phyogenetic

More information

Gauss Law. 2. Gauss s Law: connects charge and field 3. Applications of Gauss s Law

Gauss Law. 2. Gauss s Law: connects charge and field 3. Applications of Gauss s Law Gauss Law 1. Review on 1) Couomb s Law (charge and force) 2) Eectric Fied (fied and force) 2. Gauss s Law: connects charge and fied 3. Appications of Gauss s Law Couomb s Law and Eectric Fied Couomb s

More information

Lecture Note 3: Stationary Iterative Methods

Lecture Note 3: Stationary Iterative Methods MATH 5330: Computationa Methods of Linear Agebra Lecture Note 3: Stationary Iterative Methods Xianyi Zeng Department of Mathematica Sciences, UTEP Stationary Iterative Methods The Gaussian eimination (or

More information

Discrete Techniques. Chapter Introduction

Discrete Techniques. Chapter Introduction Chapter 3 Discrete Techniques 3. Introduction In the previous two chapters we introduced Fourier transforms of continuous functions of the periodic and non-periodic (finite energy) type, as we as various

More information

BP neural network-based sports performance prediction model applied research

BP neural network-based sports performance prediction model applied research Avaiabe onine www.jocpr.com Journa of Chemica and Pharmaceutica Research, 204, 6(7:93-936 Research Artice ISSN : 0975-7384 CODEN(USA : JCPRC5 BP neura networ-based sports performance prediction mode appied

More information

School of Electrical Engineering, University of Bath, Claverton Down, Bath BA2 7AY

School of Electrical Engineering, University of Bath, Claverton Down, Bath BA2 7AY The ogic of Booean matrices C. R. Edwards Schoo of Eectrica Engineering, Universit of Bath, Caverton Down, Bath BA2 7AY A Booean matrix agebra is described which enabes man ogica functions to be manipuated

More information

A Branch and Cut Algorithm to Design. LDPC Codes without Small Cycles in. Communication Systems

A Branch and Cut Algorithm to Design. LDPC Codes without Small Cycles in. Communication Systems A Branch and Cut Agorithm to Design LDPC Codes without Sma Cyces in Communication Systems arxiv:1709.09936v1 [cs.it] 28 Sep 2017 Banu Kabakuak 1, Z. Caner Taşkın 1, and Ai Emre Pusane 2 1 Department of

More information

School of Electrical Engineering, University of Bath, Claverton Down, Bath BA2 7AY

School of Electrical Engineering, University of Bath, Claverton Down, Bath BA2 7AY The ogic of Booean matrices C. R. Edwards Schoo of Eectrica Engineering, Universit of Bath, Caverton Down, Bath BA2 7AY A Booean matrix agebra is described which enabes man ogica functions to be manipuated

More information

Rate-Distortion Theory of Finite Point Processes

Rate-Distortion Theory of Finite Point Processes Rate-Distortion Theory of Finite Point Processes Günther Koiander, Dominic Schuhmacher, and Franz Hawatsch, Feow, IEEE Abstract We study the compression of data in the case where the usefu information

More information

1D Heat Propagation Problems

1D Heat Propagation Problems Chapter 1 1D Heat Propagation Probems If the ambient space of the heat conduction has ony one dimension, the Fourier equation reduces to the foowing for an homogeneous body cρ T t = T λ 2 + Q, 1.1) x2

More information

In-plane shear stiffness of bare steel deck through shell finite element models. G. Bian, B.W. Schafer. June 2017

In-plane shear stiffness of bare steel deck through shell finite element models. G. Bian, B.W. Schafer. June 2017 In-pane shear stiffness of bare stee deck through she finite eement modes G. Bian, B.W. Schafer June 7 COLD-FORMED STEEL RESEARCH CONSORTIUM REPORT SERIES CFSRC R-7- SDII Stee Diaphragm Innovation Initiative

More information

Lecture 6: Moderately Large Deflection Theory of Beams

Lecture 6: Moderately Large Deflection Theory of Beams Structura Mechanics 2.8 Lecture 6 Semester Yr Lecture 6: Moderatey Large Defection Theory of Beams 6.1 Genera Formuation Compare to the cassica theory of beams with infinitesima deformation, the moderatey

More information

New Efficiency Results for Makespan Cost Sharing

New Efficiency Results for Makespan Cost Sharing New Efficiency Resuts for Makespan Cost Sharing Yvonne Beischwitz a, Forian Schoppmann a, a University of Paderborn, Department of Computer Science Fürstenaee, 3302 Paderborn, Germany Abstract In the context

More information

A Statistical Framework for Real-time Event Detection in Power Systems

A Statistical Framework for Real-time Event Detection in Power Systems 1 A Statistica Framework for Rea-time Event Detection in Power Systems Noan Uhrich, Tim Christman, Phiip Swisher, and Xichen Jiang Abstract A quickest change detection (QCD) agorithm is appied to the probem

More information

MULTI-PERIOD MODEL FOR PART FAMILY/MACHINE CELL FORMATION. Objectives included in the multi-period formulation

MULTI-PERIOD MODEL FOR PART FAMILY/MACHINE CELL FORMATION. Objectives included in the multi-period formulation ationa Institute of Technoogy aicut Department of echanica Engineering ULTI-PERIOD ODEL FOR PART FAILY/AHIE ELL FORATIO Given a set of parts, processing requirements, and avaiabe resources The objective

More information

<C 2 2. λ 2 l. λ 1 l 1 < C 1

<C 2 2. λ 2 l. λ 1 l 1 < C 1 Teecommunication Network Contro and Management (EE E694) Prof. A. A. Lazar Notes for the ecture of 7/Feb/95 by Huayan Wang (this document was ast LaT E X-ed on May 9,995) Queueing Primer for Muticass Optima

More information

Physics 566: Quantum Optics Quantization of the Electromagnetic Field

Physics 566: Quantum Optics Quantization of the Electromagnetic Field Physics 566: Quantum Optics Quantization of the Eectromagnetic Fied Maxwe's Equations and Gauge invariance In ecture we earned how to quantize a one dimensiona scaar fied corresponding to vibrations on

More information

Throughput Rate Optimization in High Multiplicity Sequencing Problems

Throughput Rate Optimization in High Multiplicity Sequencing Problems Throughput Rate Optimization in High Mutipicity Sequencing Probems Aexander Grigoriev A.Grigoriev@KE.unimaas.n Maastricht University, 6200 MD, Maastricht, The Netherands Joris van de Kundert J.vandeKundert@MATH.unimaas.n

More information

Some Measures for Asymmetry of Distributions

Some Measures for Asymmetry of Distributions Some Measures for Asymmetry of Distributions Georgi N. Boshnakov First version: 31 January 2006 Research Report No. 5, 2006, Probabiity and Statistics Group Schoo of Mathematics, The University of Manchester

More information

Solving Elementary Shortest-Path Problems as Mixed-Integer Programs

Solving Elementary Shortest-Path Problems as Mixed-Integer Programs Gutenberg School of Management and Economics Discussion Paper Series Solving Elementary Shortest-Path Problems as Mixed-Integer Programs Michael Drexl and Stefan Irnich Januar 2012 Discussion paper number

More information

Discrete Techniques. Chapter Introduction

Discrete Techniques. Chapter Introduction Chapter 3 Discrete Techniques 3. Introduction In the previous two chapters we introduced Fourier transforms of continuous functions of the periodic and non-periodic (finite energy) type, we as various

More information

THE REACHABILITY CONES OF ESSENTIALLY NONNEGATIVE MATRICES

THE REACHABILITY CONES OF ESSENTIALLY NONNEGATIVE MATRICES THE REACHABILITY CONES OF ESSENTIALLY NONNEGATIVE MATRICES by Michae Neumann Department of Mathematics, University of Connecticut, Storrs, CT 06269 3009 and Ronad J. Stern Department of Mathematics, Concordia

More information

Copyright information to be inserted by the Publishers. Unsplitting BGK-type Schemes for the Shallow. Water Equations KUN XU

Copyright information to be inserted by the Publishers. Unsplitting BGK-type Schemes for the Shallow. Water Equations KUN XU Copyright information to be inserted by the Pubishers Unspitting BGK-type Schemes for the Shaow Water Equations KUN XU Mathematics Department, Hong Kong University of Science and Technoogy, Cear Water

More information

A proposed nonparametric mixture density estimation using B-spline functions

A proposed nonparametric mixture density estimation using B-spline functions A proposed nonparametric mixture density estimation using B-spine functions Atizez Hadrich a,b, Mourad Zribi a, Afif Masmoudi b a Laboratoire d Informatique Signa et Image de a Côte d Opae (LISIC-EA 4491),

More information

Interdiction of a Mixed-Integer Linear System

Interdiction of a Mixed-Integer Linear System Interdiction of a Mixed-Integer Linear System Bowen Hua, Ross Badick Department of Eectrica and Computer Engineering, University of Texas at Austin, Austin, TX 78712 bhua@utexas.edu, badick@ece.utexas.edu

More information

arxiv: v1 [math.ca] 6 Mar 2017

arxiv: v1 [math.ca] 6 Mar 2017 Indefinite Integras of Spherica Besse Functions MIT-CTP/487 arxiv:703.0648v [math.ca] 6 Mar 07 Joyon K. Boomfied,, Stephen H. P. Face,, and Zander Moss, Center for Theoretica Physics, Laboratory for Nucear

More information

The Group Structure on a Smooth Tropical Cubic

The Group Structure on a Smooth Tropical Cubic The Group Structure on a Smooth Tropica Cubic Ethan Lake Apri 20, 2015 Abstract Just as in in cassica agebraic geometry, it is possibe to define a group aw on a smooth tropica cubic curve. In this note,

More information

Routing optimization with time windows under uncertainty

Routing optimization with time windows under uncertainty Math. Program., Ser. A https://doi.org/10.1007/s10107-018-143-y FULL LENGTH PAPER Routing optimization with time windows under uncertainty Yu Zhang 1, Roberto Badacci 3 Mevyn Sim 4 Jiafu Tang 5 Received:

More information

VALIDATED CONTINUATION FOR EQUILIBRIA OF PDES

VALIDATED CONTINUATION FOR EQUILIBRIA OF PDES SIAM J. NUMER. ANAL. Vo. 0, No. 0, pp. 000 000 c 200X Society for Industria and Appied Mathematics VALIDATED CONTINUATION FOR EQUILIBRIA OF PDES SARAH DAY, JEAN-PHILIPPE LESSARD, AND KONSTANTIN MISCHAIKOW

More information

Maximizing Sum Rate and Minimizing MSE on Multiuser Downlink: Optimality, Fast Algorithms and Equivalence via Max-min SIR

Maximizing Sum Rate and Minimizing MSE on Multiuser Downlink: Optimality, Fast Algorithms and Equivalence via Max-min SIR 1 Maximizing Sum Rate and Minimizing MSE on Mutiuser Downink: Optimaity, Fast Agorithms and Equivaence via Max-min SIR Chee Wei Tan 1,2, Mung Chiang 2 and R. Srikant 3 1 Caifornia Institute of Technoogy,

More information

arxiv:math/ v2 [math.pr] 6 Mar 2005

arxiv:math/ v2 [math.pr] 6 Mar 2005 ASYMPTOTIC BEHAVIOR OF RANDOM HEAPS arxiv:math/0407286v2 [math.pr] 6 Mar 2005 J. BEN HOUGH Abstract. We consider a random wa W n on the ocay free group or equivaenty a signed random heap) with m generators

More information

Data Mining Technology for Failure Prognostic of Avionics

Data Mining Technology for Failure Prognostic of Avionics IEEE Transactions on Aerospace and Eectronic Systems. Voume 38, #, pp.388-403, 00. Data Mining Technoogy for Faiure Prognostic of Avionics V.A. Skormin, Binghamton University, Binghamton, NY, 1390, USA

More information

Mat 1501 lecture notes, penultimate installment

Mat 1501 lecture notes, penultimate installment Mat 1501 ecture notes, penutimate instament 1. bounded variation: functions of a singe variabe optiona) I beieve that we wi not actuay use the materia in this section the point is mainy to motivate the

More information

4 Separation of Variables

4 Separation of Variables 4 Separation of Variabes In this chapter we describe a cassica technique for constructing forma soutions to inear boundary vaue probems. The soution of three cassica (paraboic, hyperboic and eiptic) PDE

More information

Mixed Volume Computation, A Revisit

Mixed Volume Computation, A Revisit Mixed Voume Computation, A Revisit Tsung-Lin Lee, Tien-Yien Li October 31, 2007 Abstract The superiority of the dynamic enumeration of a mixed ces suggested by T Mizutani et a for the mixed voume computation

More information

Energy-Efficient Task Scheduling on Multiple Heterogeneous Computers: Algorithms, Analysis, and Performance Evaluation

Energy-Efficient Task Scheduling on Multiple Heterogeneous Computers: Algorithms, Analysis, and Performance Evaluation IEEE TRANSACTIONS ON SUSTAINABLE COMPUTING, VOL., NO., JANUARY-JUNE 206 7 Energy-Efficient Tas Scheduing on Mutipe Heterogeneous Computers: Agorithms, Anaysis, and Performance Evauation Keqin Li, Feow,

More information

arxiv: v2 [cond-mat.stat-mech] 14 Nov 2008

arxiv: v2 [cond-mat.stat-mech] 14 Nov 2008 Random Booean Networks Barbara Drosse Institute of Condensed Matter Physics, Darmstadt University of Technoogy, Hochschustraße 6, 64289 Darmstadt, Germany (Dated: June 27) arxiv:76.335v2 [cond-mat.stat-mech]

More information

BALANCING REGULAR MATRIX PENCILS

BALANCING REGULAR MATRIX PENCILS BALANCING REGULAR MATRIX PENCILS DAMIEN LEMONNIER AND PAUL VAN DOOREN Abstract. In this paper we present a new diagona baancing technique for reguar matrix pencis λb A, which aims at reducing the sensitivity

More information