Presentation is loading. Please wait.

Presentation is loading. Please wait.

1.3 Modeling with exponentially many constr.

Similar presentations


Presentation on theme: "1.3 Modeling with exponentially many constr."β€” Presentation transcript:

1 1.3 Modeling with exponentially many constr.
Some strong formulations (or even formulation itself) may involve exponentially many constraints (cutting plane method is used to solve the LP relaxations of them) The minimum spanning tree problem 𝐺=(𝑉,𝐸) undirected graph ( 𝑉 =𝑛, 𝐸 =π‘š). Every edge π‘’βˆˆπΈ has cost 𝑐 𝑒 . Find a spanning tree (acyclic connected subgraph of 𝐺) of minimum cost. Let π‘†βŠ‚π‘‰, define 𝐸 𝑆 = 𝑖,𝑗 ∈𝐸:𝑖,π‘—βˆˆπ‘† , 𝛿 𝑆 ={ 𝑖,𝑗 ∈𝐸: one end node of 𝑖,𝑗 is in 𝑆} (cut defined by 𝑆) Integer Programming 2017

2 Subtour elimination formulation min π‘’βˆˆπΈ 𝑐 𝑒 π‘₯ 𝑒 (π‘›βˆ’1 edges, no cycle)
π‘’βˆˆπΈ π‘₯ 𝑒 =π‘›βˆ’1, π‘’βˆˆπΈ(𝑆) π‘₯ 𝑒 ≀ 𝑆 βˆ’1, π‘†βŠ‚π‘‰, π‘†β‰ βˆ…, 𝑉, π‘₯ 𝑒 ∈ 0,1 Cutset formulation min π‘’βˆˆπΈ 𝑐 𝑒 π‘₯ 𝑒 (π‘›βˆ’1 edges, connected) π‘’βˆˆπ›Ώ(𝑆) π‘₯ 𝑒 β‰₯1, π‘†βŠ‚π‘‰, π‘†β‰ βˆ…, 𝑉, Thm 1.1 : (a) 𝑃 𝑠𝑒𝑏 βŠ‚ 𝑃 𝑐𝑒𝑑 , and there exist examples for which the inclusion is strict. (b) 𝑃 𝑐𝑒𝑑 can have fractional extreme points. Integer Programming 2017

3 It can be shown that LP relaxation of subtour elimination formulation gives integer optimal solutions. (polymatroid) Why consider IP formulation although there exist good algorithms (e.g., Kruskal, Prim)? Algorithms may fail if problem structure changed a little bit: degree constrained spanning tree problem, Shortest total path length spanning tree problem, Steiner tree problem, capacitated spanning tree problem, … Formulation of a basic problem may be used as part of a formulation for a larger complicated problem. Theoretical analysis, e.g. strength of 1-tree relaxation of TSP. Integer Programming 2017

4 The traveling salesman problem
𝐺=(𝑉,𝐸) undirected graph. Every edge π‘’βˆˆπΈ has cost 𝑐 𝑒 . Find a tour (a cycle that visits all nodes exactly once) of minimum cost. Cutset formulation minimize π‘’βˆˆπΈ 𝑐 𝑒 π‘₯ 𝑒 subject to π‘’βˆˆπ›Ώ( 𝑖 ) π‘₯ 𝑒 =2, π‘–βˆˆπ‘‰ π‘’βˆˆπ›Ώ(𝑆) π‘₯ 𝑒 β‰₯2, π‘†βŠ‚π‘‰, π‘†β‰ βˆ…, 𝑉, π‘₯ 𝑒 ∈ 0,1 . Subtour elimination formulation π‘’βˆˆπΈ(𝑆) π‘₯ 𝑒 ≀ 𝑆 βˆ’1, π‘†βŠ‚π‘‰, π‘†β‰ βˆ…, 𝑉, LP relaxations of both formulations give the same solution set. Integer Programming 2017

5 Remarks For directed version of the problem, the following formulation is possible, which is smaller in size. But it is a bad formulation. (refer exercise 1.21 in text page 32) 𝑒 𝑖 βˆ’ 𝑒 𝑗 +𝑛 𝑦 𝑖𝑗 β‰€π‘›βˆ’1, 𝑖,𝑗 ∈𝐴, 𝑖,𝑗≠1, 𝑖:(𝑖,𝑗)∈𝐴 𝑦 𝑖𝑗 =1, π‘—βˆˆπ‘‰ 𝑗:(𝑖,𝑗)∈𝐴 𝑦 𝑖𝑗 =1, π‘–βˆˆπ‘‰ 𝑦 𝑖𝑗 ∈ 0,1 , 𝑖,π‘—βˆˆπ‘‰ Note that, 𝑒 𝑗 ′𝑠 are continuous variables in the above formulation. Undirected TSP is a special case of directed case, we may replace each edge by two directed arcs with opposite direction and having the same costs as the edge. Integer Programming 2017

6 Is the formulation correct?
The formulation has 𝑒,𝑦 variables. If ( 𝑒 βˆ— , 𝑦 βˆ— ) feasible, we only read 𝑦 βˆ— values ( projection of ( 𝑒 βˆ— , 𝑦 βˆ— ) to 𝑦 space) We need to show that (1) any tour solution 𝑦 βˆ— satisfies the constraints and (2) any non-tour solution does not satisfy the constraints. (1) For any tour 𝑦 βˆ— , if node 𝑖 is π‘˜βˆ’th node in the tour, assign 𝑒 𝑖 =π‘˜. (2) If 𝑦 βˆ— is 0,1 and satisfies degree constraints, it is either a tour or consists of subtours. If subtours exist, there is one that does not include node 1. Add the constraints 𝑒 𝑖 βˆ’ 𝑒 𝑗 +𝑛 𝑦 𝑖𝑗 β‰€π‘›βˆ’1 along the arcs in the subtour. Integer Programming 2017

7 Comparing the LP relaxation of the cutset formulation (A) (in directed case version) and the LP relaxation of the previous formulation (B): It can be shown that the projection of the polyhedron B onto y space gives a polyhedron which completely contains A (the inclusion can be strict), hence cutset formulation (or subtour elimination formulation) is stronger. Although the previous formulation is not strong, it can be an alternative to use if you only have a generic IP software to use, not the sophisticated one to handle the cutset constraints. Integer Programming 2017

8 How to Solve the LP relaxation of the Cut-Set Formulation? (many constr.)
Solve LP relaxation (w/o cut-set constraints) If y* tour, stop. O/w find violated cut-set Solve LP after adding the Cut-set constraint. ο€€ violated cut-set? Y N Stop Integer Programming 2017

9 Branch-and-cut approach ( cutting plane alg.)
If the obtained solution is not a tour, branch and apply the same procedure again. Choose the best solution Branching : If 𝑦 𝑖𝑗 βˆ— β‰ 0,1, solve two subproblems after setting 𝑦 𝑖𝑗 =0, and 𝑦 𝑖𝑗 =1. Branch-and-cut approach ( cutting plane alg.) Ideas for TSP formulation can be used for various routing, sequencing problems. Branch-and-cut Ideas useful to solve many difficult IP problems. What can we do for the LP with many variables? For the LP with many vars. and constraints? TSP site: Integer Programming 2017

10 The perfect matching problem
Match 𝑛 persons into pairs perfectly. Cost 𝑐 𝑒 , 𝑒={𝑖,𝑗}, if person 𝑖 is matched with person 𝑗. Construct graph 𝐺=(𝑉,𝐸) with edge cost 𝑐 𝑒 ,𝑒={𝑖,𝑗} for all π‘’βˆˆπΈ. If {𝑖,𝑗}βˆ‰πΈ, 𝑖 and 𝑗 cannot be matched. minimize π‘’βˆˆπΈ 𝑐 𝑒 π‘₯ 𝑒 subject to π‘’βˆˆπ›Ώ( 𝑖 ) π‘₯ 𝑒 =1, π‘–βˆˆπ‘‰ π‘₯ 𝑒 ∈ 0,1 . 𝑃 π‘‘π‘’π‘”π‘Ÿπ‘’π‘’ β‰ π‘π‘œπ‘›π‘£(𝐹) (see Fig 1.7) Add π‘’βˆˆπ›Ώ(𝑆) π‘₯ 𝑒 β‰₯1, π‘†βŠ‚π‘‰, 𝑆≠𝑉, 𝑆 odd or π‘’βˆˆπΈ(𝑆) π‘₯ 𝑒 ≀( 𝑆 βˆ’1)/2, π‘†βŠ‚π‘‰, 𝑆≠𝑉, 𝑆 odd Both have 𝑃 π‘šπ‘Žπ‘‘π‘β„Žπ‘–π‘›π‘” =π‘π‘œπ‘›π‘£(𝐹). Integer Programming 2017

11 Cut covering problems General problem class that includes many problems on networks and graphs 𝐺= 𝑉,𝐸 , 𝑉 =𝑛, undirected graph 𝑓: 2 𝑉 β†’ 𝑍 + , π·βŠ†π‘‰, costs 𝑐 𝑒 β‰₯0 for π‘’βˆˆπΈ Cut covering problems minimize π‘’βˆˆπΈ 𝑐 𝑒 π‘₯ 𝑒 subject to π‘’βˆˆπ›Ώ( 𝑖 ) π‘₯ 𝑒 =𝑓 𝑖 , π‘–βˆˆπ·βŠ†π‘‰, π‘’βˆˆπ›Ώ(𝑆) π‘₯ 𝑒 β‰₯𝑓 𝑆 , π‘†βŠ†π‘‰, π‘₯ 𝑒 ∈ 0,1 for all π‘’βˆˆπΈ. There exists an optimal solution which is minimal w.r.t. inclusion. ( 𝑐 𝑒 β‰₯0 ) Integer Programming 2017

12 subject to π‘’βˆˆπ›Ώ( 𝑖 ) π‘₯ 𝑒 =𝑓 𝑖 , π‘–βˆˆπ·βŠ†π‘‰, π‘’βˆˆπ›Ώ(𝑆) π‘₯ 𝑒 β‰₯𝑓 𝑆 , π‘†βŠ†π‘‰,
minimize π‘’βˆˆπΈ 𝑐 𝑒 π‘₯ 𝑒 subject to π‘’βˆˆπ›Ώ( 𝑖 ) π‘₯ 𝑒 =𝑓 𝑖 , π‘–βˆˆπ·βŠ†π‘‰, π‘’βˆˆπ›Ώ(𝑆) π‘₯ 𝑒 β‰₯𝑓 𝑆 , π‘†βŠ†π‘‰, π‘₯ 𝑒 ∈ 0,1 . The minimum spanning tree 𝐷=βˆ…, 𝑓 𝑆 =1, for all π‘†β‰ βˆ…, 𝑉 The traveling salesman problem 𝐷=𝑉, 𝑓 𝑆 =2, for all π‘†β‰ βˆ…, 𝑉 The perfect matching problem 𝐷=𝑉, 𝑓 𝑆 =1, for all π‘†β‰ βˆ…, 𝑉 with 𝑆 odd The Steiner tree problem π‘‡βŠ‚π‘‰ needs to be connected by a tree possibly using nodes in π‘‰βˆ–π‘‡. 𝐷=βˆ…, 𝑓 𝑆 =1, for all 𝑆 with π‘†βˆ©π‘‡β‰ βˆ…, 𝑇 =0, otherwise Integer Programming 2017

13 The survivable network design problem Costs 𝑐 𝑒 , for all π‘’βˆˆπΈ,
requirements π‘Ÿ 𝑖𝑗 for every pair of nodes 𝑖,π‘—βˆˆπ‘‰ Select a set of edges from 𝐸 at minimum cost, so that between every pair of nodes 𝑖 and 𝑗 there are at least π‘Ÿ 𝑖𝑗 paths that do not share any edges ( π‘Ÿ 𝑖𝑗 edge-disjoint paths) 𝐷=βˆ…, 𝑓 𝑆 = max π‘–βˆˆπ‘†, π‘—βˆˆπ‘‰βˆ–π‘† π‘Ÿ 𝑖𝑗 , π‘†β‰ βˆ…, 𝑉 The vehicle routing problem Integer Programming 2017

14 Dircted vs. undirected formulations
Steiner tree problem minimize π‘’βˆˆπΈ 𝑐 𝑒 π‘₯ 𝑒 subject to π‘’βˆˆπ›Ώ(𝑆) π‘₯ 𝑒 β‰₯1, βˆ€ π‘†βŠ†π‘‰, π‘†βˆ©π‘‡β‰ βˆ…, 𝑇, (1.8) π‘₯ 𝑒 ∈ 0,1 . (a) 𝑉 𝑖 βˆ©π‘‡β‰ βˆ…, 𝑖=1,…,𝑝. (b) 𝑉 𝑖 ∩ 𝑉 𝑗 =βˆ…, 𝑖,𝑗=1,…,𝑝, 𝑖≠𝑗. (c) 𝑖=1 𝑝 𝑉 𝑖 =𝑉 Let 𝛿( 𝑉 1 ,…, 𝑉 𝑝 ) be the set of edges, whose endpoints lie in different 𝑉 𝑖 . minimize π‘’βˆˆπΈ 𝑐 𝑒 π‘₯ 𝑒 subject to π‘’βˆˆπ›Ώ( 𝑉 1 ,…, 𝑉 𝑝 ) π‘₯ 𝑒 β‰₯π‘βˆ’1, ( 𝑉 1 ,…, 𝑉 𝑝 ) satisfying (a)-(c) π‘₯ 𝑒 ∈ 0, (1.9) Integer Programming 2017

15 ( 𝑖,𝑗 ∈𝐸 β†’ π‘‘π‘€π‘œ π‘Žπ‘Ÿπ‘π‘  𝑖,𝑗 , (𝑗,𝑖)∈𝐴, 𝑐 𝑖𝑗 = 𝑐 𝑗𝑖 β‰₯0)
Directed version 𝐺= 𝑉,𝐸 β†’ 𝐺=(𝑉,𝐴) ( 𝑖,𝑗 ∈𝐸 β†’ π‘‘π‘€π‘œ π‘Žπ‘Ÿπ‘π‘  𝑖,𝑗 , (𝑗,𝑖)∈𝐴, 𝑐 𝑖𝑗 = 𝑐 𝑗𝑖 β‰₯0) Find a minimum cost directed subtree that contains a directed path between some given root vertex 1 (1βˆˆπ‘‡), and every other terminal in 𝑇. minimize (𝑖,𝑗)∈𝐴 𝑐 𝑖𝑗 𝑦 𝑖𝑗 subject to (𝑖,𝑗)∈ 𝛿 + (𝑆) 𝑦 𝑖𝑗 β‰₯1, βˆ€ π‘†βŠ‚π‘‰, 1βˆˆπ‘†, π‘‡βˆ–π‘†β‰ βˆ…, (1.10) 𝑦 𝑖𝑗 + 𝑦 𝑗𝑖 ≀1, 𝑒= 𝑖,𝑗 ∈𝐸, 𝑦 𝑖𝑗 ∈{0,1}. can recover π‘₯ by setting π‘₯ 𝑒 = 𝑦 𝑖𝑗 + 𝑦 𝑗𝑖 , for all 𝑒= 𝑖,𝑗 ∈𝐸, 𝑍 π‘ π‘‘π‘’π‘–π‘›π‘’π‘Ÿ (𝑇)≀ 𝑍 π‘π‘Žπ‘Ÿπ‘‘π‘–π‘‘π‘–π‘œπ‘› (𝑇)≀ 𝑍𝐷 π‘ π‘‘π‘’π‘–π‘›π‘’π‘Ÿ (𝑇) (for linear relaxations) Integer Programming 2017

16 Assume that the root vertex 1∈ 𝑉 1 , and consider
(1.8) is a special case of (1.9) with 𝑝=2. ⟹ 𝑍 π‘ π‘‘π‘’π‘–π‘›π‘’π‘Ÿ (𝑇)≀ 𝑍 π‘π‘Žπ‘Ÿπ‘‘π‘–π‘‘π‘–π‘œπ‘› (𝑇) Assume that the root vertex 1∈ 𝑉 1 , and consider (𝑖,𝑗)∈ 𝛿 + (π‘‰βˆ– 𝑉 π‘˜ ) 𝑦 𝑖𝑗 β‰₯1, π‘˜=2,…,𝑝. Add above together with 𝑦 𝑗𝑖 β‰₯0 for π‘—βˆˆ 𝑉 π‘˜ , π‘˜=2,…,𝑝 and π‘–βˆˆ 𝑉 1 ,(𝑖,𝑗)∈𝐸 ⟹ π‘’βˆˆπ›Ώ( 𝑉 1 ,…, 𝑉 𝑝 ) ( 𝑦 𝑖𝑗 + 𝑦 𝑗𝑖 ) β‰₯π‘βˆ’1. setting π‘₯ 𝑒 = 𝑦 𝑖𝑗 + 𝑦 𝑗𝑖 , we get feasible π‘₯ for the linear relaxation of (1.9) ⟹ 𝑍 π‘π‘Žπ‘Ÿπ‘‘π‘–π‘‘π‘–π‘œπ‘› (𝑇)≀ 𝑍𝐷 π‘ π‘‘π‘’π‘–π‘›π‘’π‘Ÿ (𝑇) There are examples such that 𝑍 π‘ π‘‘π‘’π‘–π‘›π‘’π‘Ÿ 𝑇 < 𝑍𝐷 π‘ π‘‘π‘’π‘–π‘›π‘’π‘Ÿ (𝑇) For TSP, directed formulation has the same strength Integer Programming 2017

17 1.4 Modeling with exponentially many variables
Column generation method Enumerate partial feasible solutions and represent their interactions in the master model. (Decomposition) Important modeling tool in applications The cutting stock problem Large rolls of paper of width π‘Š (raw). Customer demand 𝑏 𝑖 rolls of width 𝑀 𝑖 (final), 𝑖=1,…,π‘š. ( 𝑀 𝑖 β‰€π‘Š) Minimize the number of large rolls used while satisfying customer demand. Cutting pattern 𝑗, π‘Ž 1𝑗 ,…, π‘Ž π‘šπ‘— : produce π‘Ž 𝑖𝑗 rolls of width 𝑀 𝑖 in π‘—βˆ’th cutting pattern (number of possible cutting patterns can be enormous) A feasible cutting pattern 𝑗 must satisfy 𝑖=1 π‘š π‘Ž 𝑖𝑗 𝑀 𝑖 β‰€π‘Š and π‘Ž 𝑖𝑗 is nonnegative integer. (integer knapsack constraint) Integer Programming 2017

18 subject to 𝑗=1 𝑛 π‘Ž 𝑖𝑗 π‘₯ 𝑗 = 𝑏 𝑖 , 𝑖=1,…,π‘š, π‘₯ 𝑗 ∈ 𝑍 + , 𝑗=1,…,𝑛.
Formulation minimize 𝑗=1 𝑛 π‘₯ 𝑗 subject to 𝑗=1 𝑛 π‘Ž 𝑖𝑗 π‘₯ 𝑗 = 𝑏 𝑖 , 𝑖=1,…,π‘š, π‘₯ 𝑗 ∈ 𝑍 + , 𝑗=1,…,𝑛. π‘₯ 𝑗 is the number of rolls of width π‘Š (raw) cut by cutting pattern 𝑗. LP relaxation can be solved by column generation. Fractional optimal solution may be rounded down and a few more raws may be used to produce additional finals. (close to optimal) Integer Programming 2017

19 Combinatorial auctions
𝑁: set of bidders, 𝑀: set of items being auctioned 𝑏 𝑗 (𝑆) : bid that bidder 𝑗 is willing to pay for π‘†βŠ†π‘€ Assume that if 𝑆⋂𝑇=βˆ…, 𝑏 𝑗 𝑆 + 𝑏 𝑗 (𝑇)≀ 𝑏 𝑗 (𝑆βˆͺ𝑇) Bidders are allowed to bid on combinations of different items. Let 𝑏 𝑆 = max π‘—βˆˆπ‘ 𝑏 𝑗 (𝑆) maximize π‘†βŠ‚π‘€ 𝑏(𝑆) π‘₯ 𝑠 subject to 𝑆:π‘–βˆˆπ‘† π‘₯ 𝑆 ≀1, π‘–βˆˆπ‘€, π‘₯ 𝑆 ∈ 0,1 , π‘†βŠ‚π‘€ Integer Programming 2017

20 The vehicle routing problem
transportation network: 𝐺= 𝑉,𝐸 , undirected, cost 𝑐 𝑒 , π‘’βˆˆπΈ. Node 0 is central depot. Node π‘–βˆˆπ‘‰ represents customers with demand 𝑑 𝑖 . Company has π‘š vehicles with capacities π‘ž π‘˜ , π‘˜=1,…,π‘š Assume demand of each customer cannot be divided into several vehicles. Let π‘₯ 𝑗 =1 if partial tour 𝑗 is used, and zero, otherwise (𝑗=1,…,𝑁) π‘Ž 𝑖𝑗 : equals one if node 𝑖 is visited in partial solution 𝑗. 𝑐 𝑗 : cost of tour 𝑗 minimize 𝑐 β€² π‘₯ subject to 𝐴π‘₯=𝑒 π‘₯∈ 0,1 𝑁 . (may have additional constraint, π‘₯ 𝑗 β‰€π‘š ) Integer Programming 2017


Download ppt "1.3 Modeling with exponentially many constr."

Similar presentations


Ads by Google