Backbone analysis and algorithm design for the quadratic assignment problem

Size: px
Start display at page:

Download "Backbone analysis and algorithm design for the quadratic assignment problem"

Transcription

1 Science in China Series F: Information Sciences 008 SCIENCE IN CHINA PRESS Springer wwwscichinacom infoscichinacom wwwspringerlinkcom Backbone analysis and algorithm design for the quadratic assignment problem JIANG He,, ZHANG XianChao, CHEN GuoLiang & LI MingChu School of Software, Dalian University of Technology, Dalian 66, China; Department of Computer Science, University of Science and Technology of China, Hefei 3007, China As the hot line in NP-hard problems research in recent years, backbone analysis is crucial for phase transition, hardness, and algorithm design Whereas theoretical analysis of backbone and its applications in algorithm design are still at a beginning state yet, this paper took the quadratic assignment problem (QAP) as a case study and proved by theoretical analysis that it is NP-hard to find the backbone, ie, no algorithm exists to obtain the backbone of a QAP in polynomial time Results of this paper showed that it is reasonable to acquire approximate backbone by intersection of local optimal solutions Furthermore, with the method of constructing biased instances, this paper proposed a new meta-heuristic biased instance based approximate backbone (BI-AB), whose basic idea is as follows: firstly, construct a new biased instance for every QAP instance (the optimal solution of the new instance is also optimal for the original one); secondly, the approximate backbone is obtained by intersection of multiple local optimal solutions computed by some existing algorithm; finally, search for the optimal solutions in the reduced space by fixing the approximate backbone Work of the paper enhanced the research area of theoretical analysis of backbone The meta-heuristic proposed in this paper provided a new way for general algorithm design of NP-hard problems as well quadratic assignment problem, NP-hard, backbone analysis, biased instance, meta-heuristic Introduction NP-hard problems [] have been the focus of computer science research for many years, with extensively wide applications arising in industrial control, transportation scheduling, network layout, very large scale integration (VLSI), etc According to the computational complexity theory [], there exists no exact algorithms to solve NP-hard problems in polynomial time unless P=NP Received October 7, 006; accepted July, 007 doi: 0007/s Corresponding author ( jianghe@dluteducn) Supported by the National Natural Science Foundation of China (Grant Nos and ), the Natural Science Foundation of LiaoNing Province (Grant No 00508), and the Gifted Young Foundation of Dalian University of Technology Sci China Ser F-Inf Sci May 008 vol 5 no

2 Therefore, exact algorithms are usually applied to small instances of NP-hard problems For larger instances, many heuristics offering very good near optimal solutions in a reasonable time have been developed Known heuristics can be classified into the following categories: ) simple heuristics, including greedy algorithm, local search, and tabu search [3] ; ) meta-heuristics, including ant colony algorithms [4,5], genetic algorithms [6], simulated annealing [7], neural networks [8], and quasi-physical and quasi-sociological algorithms [9] Backbone analysis [0] has been the hotline of research for NP-hard problems in recent years The backbone, the shared common part of all optimal solutions for an NP-hard problem instance, is important to evaluate the hardness and phase transition of NP-hard problems Monasson et al [] investigated the effect of the backbone on the hardness and phase transition of the satisfiability problem (SAT) in 998 Zhang [] analyzed the effect of the backbone size on the asymmetric traveling salesman problem (ATSP) Moreover, backbone analysis has been currently an important way to design heuristics for NP-hard problems Schneider et al [3] and Zou et al [4] proposed multilevel reduction algorithms for the TSP by using the intersection of local optimal solutions as the approximate backbone, respectively Zhang et al [5] presented a backbone guided LK algorithm for the TSP Zhang et al [6], Dubois et al [7], and Valnir et al [8] gave backbone guided local search algorithms for the SAT, respectively Zou et al [9] developed the approximate backbone-guided fant (ABFANT) for the quadratic assignment problem (QAP) However, backbone analysis is still at the beginning state by experimental statistic methods Also, there are few theoretical analysis results of the backbone As to our knowledge, there is no report on algorithms designed by theoretical analysis results of the backbone The only theoretical result is that Kilby et al [0] proved that it is NP-hard to obtain the backbone of the TSP For more complex NP-hard problems, such as the QAP, there still lacks theoretical analysis results of the backbone For the QAP, there exist some interesting and important problems as follows Firstly, is there a polynomial time algorithm to obtain the backbone of the QAP? If so, we can use the backbone other than the approximate backbone for algorithm design Secondly, a serious disadvantage exists to those widely used methods, which approximate the backbone by intersection of local optimal solutions: When the number of optimal solutions is not unique, the size of the backbone is very small and the benefit by using the approximate backbone would be obviously limited Is there any method to settle the deficiency due to small size of the backbone for NP-hard problems? This paper solves the problems mentioned above as follows Firstly, we proved that there is no polynomial time algorithm to obtain the backbone of the QAP under the assumption that P NP In the proof, a method constructing the biased QAP instance is given Every QAP instance can be transformed into its corresponding biased instance, whose optimal solution is also optimal to the original instance Secondly, a new meta-heuristic, the biased instance based approximate backbone (BI-AB) for the QAP, is proposed It consists of five steps: Step Given a QAP instance, construct its biased QAP instance with unique optimal solution (ie, backbone) Step Obtain multiple local optimal solutions to the new constructed QAP instance by calling some QAP heuristic Step 3 Obtain the approximate backbone by intersection of local optimal solutions Step 4 Reduce the search space by fixing the approximate backbone Step 5 Search for solutions in the reduced search space by calling some existing QAP heuristic JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

3 It can be drawn that we solve the above problem of small size backbone, due to existence of multiple optimal solutions We used FANT [] as the subordinate algorithm and tested the BI-AB on the standard QAP benchmark QAPLIB [] Experimental results indicated that the new heuristic outperforms ABFANT and Restart FANT in terms of solution quality Work of this paper extends the research of the QAP The method by construction of the biased instance provides a direction for backbone analysis of other NP-hard problems On the other hand, the new heuristic being presented in this paper, that constructs the biased instance and then obtains the approximate backbone by intersection of local optimal solutions, throws a light on algorithm design for other NP-hard problems Preliminaries In this section, we shall give some related definitions and notations Definition Given n facilities, represented by the set F = { f, f,, f n }, and n locations represented by the set L= { l, l,, l n }, the quadratic assignment problem (QAP) is to determine to which location each facility must be assigned at minimal cost Mathematically, let A= ( a ) nxn be a matrix where a represents the flow between facilities fi and f j Let B = ( ) be a matrix where instance b represents the distance between locations li and l j A feasible solution of a QAP n n j i i j) QAP( A, B ) is an assignment :{,,, n} {,,, n} whose cost function is defined as g ( A, B) = = = a b( ) ( In the QAP, the goal is to find a feasible solution * :{,,, n} {,,, n} that minimizes the cost, ie, g * ( A, B) = min({ g ( A, B) Π }), where Π is the set of all permutations of{,,, n} The QAP is a classic combinatorial optimization problem with extensively wide applications, such as hospital layout, keyboard arrangement, and VLSI Sahni et al [] have shown that the problem is NP-hard, so exact algorithms are currently only effective for instances of size n 30 Therefore, several heuristics have been proposed for finding near-optimum solutions to large QAP instances, including ant colony [3,4], genetic algorithm [5], simulated annealing [6], neural networks [7], tabu search [8,9], and GRASP [30] Definition Given the flow matrix, let m = max({ a i, j n} {}) Given the distance matrix A B = ( b ) nxn, let mb = max({ b i, j n} {}) The biased matrix of A is defined as ˆ in + jn A= ( a + ξ ) n n, where ξ = (3mB ), and the biased matrix of B is defined as ˆ in+ j B = ( b + δ ) n n where δ = (3mA ) In this paper, the biased QAP instance of QAP( A, B) is denoted as QAP( Aˆ, Bˆ ) Definition 3 Given a QAP instance optimal solutions to QAP( A, B), where backbone of QAP( A, B) P( ) = {( i, ( i)) i n} is defined as QAP( A, B), let q b nxn Π = {,,, } be the set of all * q = Π represents the number of optimal solutions The bone( A, B) = P( ) P( ) P ( * ), where q 478 JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

4 It is significant to obtain the backbone bone( A, B) for algorithm design According to Definition 3, a heuristic cannot obtain an optimal solution to a QAP instance, unless it obtains the whole backbone bone( A, B ) On the other hand, if the bone( A, B) is obtained, the search space could then be effectively reduced by assigning facilities in the backbone to some fixed locations Boese [3] observed that a local optimal solution tends to have 80% common edges shared with an optimal solution Such observations have also led to the big valley structure, which suggests that extensively many local optimal solutions form a single cluster around the optimal solutions Similar structures are also found in the graph partitioning problem [3], and the flow shop scheduling problem [33] Since it is hard to obtain the optimal solutions, many researchers usually approximate the backbone by intersection of local optimal solutions, according to the relationship between local optimal solutions and optimal solutions Definition 4 Given a QAP instance QAP( A, B), the approximate backbone is defined as a- bone(,,, k ) = P( ) P( ) P( k ), where,,, k represent local optimal solutions to the QAP( A, B) 3 Theoretical analysis for backbone Without loss of generality, we assume that the entries a, b in the matrices A and B are nonnegative integers, ie, a, b Z + { 0} for i, j n In Lemma and Lemma, we shall study the relationship between the solution of a QAP instance and that of the biased QAP instance Lemma Given the matrices A= ( a ) nxn ( a Z + {0} ) and B = ( b ) nxn ( b Z + {0} ), if g ( AB, ) < g ( AB, ) for any two distinct solutions, to the QAP instance QAP( A, B), then g ( AB ˆ, ˆ) < g ( AB ˆ, ˆ) for the biased QAP instance QAP( A ˆ, B ˆ ) Proof Firstly, we shall prove that max({ b i, j n}) 0 and max({ a i, j n}) 0 Otherwise, if max({ a i, j n }) = 0, then we have a = 0 for i, j n It follows that g ( A, B) = g ( A, B) = 0 which contradicts with the assumption of Lemma Similarly, we have max({ b i, j n}) 0 Thus, m = max({ a i, j n} {}) = max({ a i, j n}), A m = max({ b i, j n}) Secondly, we shall show that B g ( AB ˆ, ˆ) < g ( AB ˆ, ˆ) holds if g ( AB, ) < g ( AB, ) for any distinct feasible solutions, to the QAP instance QAP( A, B) + By the assumption of Lemma, we have g ( AB, ) g ( AB, ) Z By definition of the biased QAP instance, for any feasible solution, we have ˆ n n g ( A, Bˆ ) = ( a + ξ )( b + δ ) δ j= i= () i ( j) () i ( j) n n n n n n n n ab j i () i ( j) a δ j i () i ( j) ξ j i b() i ( j) ξ = = = = = = j= i= () i ( j) = JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

5 n n n n n n (, ) δ j i ( i) ( j) ξ j i ( i) ( j) ξ j i δ = = = = = = ( i) ( j) = g A B + a + b + Let Δ Then we have and n n n n n n = aδ() i ( j), Δ = ξb() i ( j), Δ = ξδ j= i= j= i= and 3 j= i= () i ( j) n n n n ( in ) + ( j) n n ( in ) + ( j) a δ j i () i ( j) a j i ma m j i A m = = = = = = A Δ = = ( ) ( ) n n () in+ ( j) n n in+ j =, 3 j= = i= 3 j= = i= 3 < n n + n 3 n n n n in + jn n n in + jn ξ b j i () i ( j) b j i () i ( j) mb m j i B m = = = = = = B Δ = = ( ) ( ) n n in + jn n n in+ j j= i= j= i= n n + n = < = <, n n n n in + jn + () i n+ ( j) ξδ j i () i ( j) mm = = j= i= A B Δ = = (9 ) n n in + jn + ( i) n+ ( j) n n ( i) n+ ( j) j= i= j i A B mm = = A B = < 9mm 9 n n in+ j = 9mm j= = i= 9 n n n A B mm + < A B 9 By the definition, we have in+ j 3 a 0, b 0, in + jn ξ = (3m ) > 0 and B 3 7 δ = (3mA ) > 0, so Δ 0, Δ 0, Δ 0, and 0 Δ +Δ +Δ 9 If g ( AB, ) < g ( AB, ) for any two distinct feasible solutions,, then we have 3 3 g ˆ ˆ ˆ ˆ ( A, B) g ( A, B) = ( g (, ) (, )) ( ) ( ) A B g A B + Δ +Δ +Δ Δ +Δ +Δ have if then Since g ( A, B) g ( A, B) g ( Aˆ, Bˆ) g ( Aˆ, Bˆ) > 0 and 7 Δ +Δ +Δ Δ +Δ +Δ ) 7, 3 3 ( ) ( 9 9 and so this lemma is proved Lemma Given the matrices A= ( a ) nxn ( a Z + {0} ) and B = ( b ) nxn ( b Z + {0} ), g ( AB, ) = g ( AB, ) for any two distinct solutions, to the QAP instance QAP( A, B), g ( AB ˆ, ˆ) g ( AB ˆ, ˆ) for the biased QAP instance QAP( A ˆ, B ˆ ) Proof By the assumption of Lemma, we have g ( A, B) = g ( A, B) Therefore, 3 3 g ˆ ˆ ˆ ˆ ( A, B) g ( A, B) = ( Δ +Δ +Δ ) ( Δ +Δ + Δ ) We need only to 3 3 prove that ( Δ +Δ +Δ ) ( Δ +Δ +Δ ) 0 Since, so there must exist i ( i n) subject to () i () i Let hi (, j, ) = in + jn + () in+ ( j), D = { hi (, j, ) hi (, j, ) hi (, j, ), i, j n}, and D = { hi (, j, ) hi (, j, ) hi (, j, ), i, j n}, then we have D D Let m = max( D D ) Therefore, we D D =, 480 JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

6 and 3 n n ( a δ () ( ) () ( ) () ( )) j i i j ξb i j ξδ = = i j = i, j n aδ() i ( j) + ξb() i ( j) + ξδ() i ( j) h (,, ) m Δ +Δ +Δ = + + ( ) + ( a δ + ξ b + ξ δ ), i, j n ( i) ( j) ( i) ( j) ( i) ( j) h (,, ) >m 3 n n ( a δ () ( ) () ( ) () ( )) j i i j ξb i j ξδ = = i j = i, j n aδ( i) ( j) + ξb( i) ( j) + ξδ( i) ( j) h (,, ) m Δ +Δ +Δ = + + By the definition of ( ) + ( a δ + ξ b + ξ δ ) i, j n () i ( j) () i ( j) () i ( j) h (,, ) >m, m for any i, j n, if hi (, j, ) = hi (, j, ), and ( i ) = ( i ), ( j ) = ( j ) Hence, Thus, we have > m (,, ) >, we have hi j m i, j n () i ( j) () i ( j) () i ( j) h (,, ) = ( a δ + ξ b + ξ δ ) i, j n () i ( j) () i ( j) () i ( j) h (,, ) > m 3 3 Δ +Δ +Δ ( Δ +Δ +Δ ) = ( a δ + ξ b + ξ δ ) i, j n ( i) ( j) ( i) ( j) ( i) ( j) h (,, ) m ( a δ + ξ b + ξ δ ) ( a δ + ξ b + ξ δ ) i, j n () i ( j) () i ( j) () i ( j) h (,, ) m () in+ ( j) in + jn in + jn + () i n+ ( j) i, j n ( a (3mA ) b() i ( j) (3mB ) + (9mAmB )) h (,, ) m = + = () in+ ( j) in + jn in+ jn+ () in+ ( j) ( a (3m ) b (3m ) (9m m ) ) + + i, j n A () i ( j) B A B h (,, ) m m m ( ( i) n+ ( j)) m ( in + jn ) m h( i, j, ) (9mAmB )) i, j n (3mBa + 3mb A () i ( j) + h (,, ) m m m ( ( i) n+ ( j)) m ( in + jn ) m (,, h i j ) i, j n (3mBa + 3mAb() i ( j) + ) h (,, ) m By the definition, we have m = max( D D ) D D, ie, m D or D that In the following proof, we only consider the case that m D, Assume that we can prove it in a similar way * 3 * m = h( i, j, ) = i n + j n + ( i ) n+ ( j ), 3 3 Δ +Δ +Δ ( Δ +Δ +Δ ) * 3 * in + jn B mab ( i ) ( j ) ( i ) n+ ( j ) m h( i, j, ) m D For the case m m ( ( i) n+ ( j)) m ( in + jn ) m (,, h i j ) = (9mAmB )) i, j n (3mBa + 3mAb() i ( j) + ) hi (, j, ) < m + (3m a + + ) then ) JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

7 m ( ( i) n+ ( j)) m ( in + jn ) m h( i, j, ) i, j n (3ma B + 3mb A () i ( j) + ) h (,, ) < m m m ( ( i) n+ ( j)) m ( in + jn ) m h( i, j, ) = (9mm A B )) i, j n (3ma B + 3mb A () i ( j) + ) h (,, ) < m * 3 * in+ jn B mb A ( i ) ( j ) ( i ) n+ ( j ) + (3ma + + ) m ( ( i) n+ ( j)) m ( in + jn ) m (,, h i j ) i, j n (3ma B + 3mb A () i ( j) + ) h (,, ) < m If hi (, j, ) < for i, j n,, then 3ma,, and m B m ( ( i) n+ ( j)) m h(, i j, ) 3mb If A () i ( j) m ( in + jn ) * 3 * in+ jn are even integers, so are 3ma B and + ( ) ( ) A ( i ) ( j ) 3 mb i n j ( ( ) ( )) hi (, j, ) < for i, j n, then m ma i n+ j m h(, i j, ),, and 3mb m m ( in + jn ) A are even integers Therefore, () i ( j) m ( ( i) n+ ( j)) m ( in + jn ) m h( i, j, ) i, j n (3ma B + 3mb A () i ( j) + ) h (,, ) < m * 3 * in+ jn ( i ) n+ ( j ) + (3ma B + 3mb A + ) ( i ) ( j ) ( in + jn ) m h( i, j, ) i, j + h (,, ) Hence, m ( ( i) n+ ( j)) m n (3ma B 3mb A () i ( j) < m 3 3 ( Δ +Δ +Δ ) ( Δ +Δ +Δ ) 0 B and so this lemma is proved + ) 0 Theorem Given the matrices A= ( a a Z + ) nxn ( { 0} ) and B = ( b ) nxn ( b Z + { 0}), there is a unique optimal solution to the biased QAP instance QAP( Aˆ, Bˆ ) Proof Otherwise, there must at least be two distinct optimal solutions to the QAP( Aˆ, Bˆ ) Let, be among these optimal solutions subject to We distinguish three cases, according to the relationship between g ( AB, ) and g (, ) AB Case g ( AB, ) = g ( AB, ) By Lemma, we have g ( ˆ, ˆ) ( ˆ, ˆ AB g AB), a contradiction Case g ( AB, ) < g ( AB, ) By Lemma, we have g ( Aˆ, Bˆ) < g ( Aˆ, Bˆ), a contradiction Case 3 g ( AB, ) < g ( AB, ) By Lemma, we have g ( ˆ, ˆ) ( ˆ, ˆ AB < g AB), a contradiction Thus, the theorem is proved Corollary Given the matrices A= ( a ) nxn ( a Z + {0}) and B = ( b ) nxn ( b Z + {0} ), the optimal solution to the biased QAP instance QAP( Aˆ, Bˆ ) is also optimal to 48 JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

8 the instance QAP( A, B) Proof By Theorem, there exists a unique optimal solution denoted by * to the biased QAP instance QAP( Aˆ, Bˆ ) Then, the solution * must be optimal to the instance QAP( A, B), * otherwise there exists a solution to the instance QAP( Aˆ, B ˆ ) which is subject to g ( AB, ) < g * ( AB, ) However, by Lemma, we have g ( AB ˆ, ˆ) < g * ( AB ˆ, ˆ) The corollary is proved Theorem There exists no polynomial time algorithm to obtain the backbone of the QAP unless P = NP Proof Otherwise, there must exist an algorithm denoted by Λ which is able to obtain the backbone of the QAP in polynomial time Given any arbitrary flow matrix A= ( a ) nxn ( a 0 ) and distance matrix B = ( b ) nxn b 0), the instance QAP( A, B) could be transformed to its biased QAP instance ( by a polynomial time algorithm denoted by Γ Let Ο ( ) be the time complexity of the algorithm Γ Since the QAP( Aˆ, Bˆ ) is also a QAP instance, its backbone can be obtained by Λ in polynomial time which is denoted by QAP( Aˆ, Bˆ ) Ο ( ) Also, by Lemma, the backbone is not only the unique optimal solution to the QAP ( Aˆ, Bˆ) itself, but also optimal to the QAP( A, B) Hence, any QAP instance can be exactly solved in Ο ( ) +Ο( ) time by Γ and Λ Obviously, such conclusion contradicts with the result that the QAP is NP-hard by Sahni [] Thus, this theorem is proved 4 Biased instance based approximate backbone algorithm In this section, we shall propose a new meta-heuristic for the QAP biased instance based approximate backbone (BI-AB), by constructing the biased QAP instance used in the backbone analysis Firstly, we review the background and disadvantages of the ABFANT After that, we give our algorithm BI-AB 4 Approximate backbone guided FANT Among those researchers who applied the backbone for algorithm design of NP-hard problem, Zou et al [9] proposed the ABFANT for the QAP in 005 Algorithm: ABFANT Input: QAP instance QAP(A, B) Output: Solution 0 Begin Find an initial solution 0 by using FANT for the QAP instance QAP(A, B) Do the following: Find k different solutions,,, k by FANT Obtain the approximate-backbone a- bone( 0,,, k ) of 0,,, k JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

9 3 Fix a- bone( 0,,, k ) to obtain a new search space 4 Run FANT to obtain the solution k 5 If g ( AB, ) < g ( AB, ), then set 0 = k k 0 S * ( QAP( A, B)) 3 Return 0 End The ABFANT achieved better performance than the FANT, in the way that reduces the search space by fixing the approximate backbone The theoretical basis of the AB-FANT is the big valley structure as mentioned above However, the effect of the approximate backbone on the QAP is not as prominent as that on the TSP, for which Zou et al [4] designed a multilevel reduction algorithm using the approximate backbone The reason for this is, that the backbone size is usually equal to the size of the TSP instances from the TSPLIB, ie, there exists unique optimal solution to such TSP instance Slaney et al [0] generated -D integer Euclidean TSP problems with 0 cities randomly placed in a square and found that the backbone size quickly approximates n as the length of the square is increased Zhang [] observed that the backbone of random generated asymmetric TSP instances rises to n quickly, as the precision of intercity distances increases However, the property of solutions to the QAP is more complex than that of the TSP Firstly, the backbone size of the QAP sharply varies from instance to instance For example, many instances from the QAPLIB have more than one optimal solution By enumeration method, we obtained the statistic results of instances with facilities from the QAPLIB (see Table ) We can find that only an optimal solution exists to every QAP instance of chrb, chrc, roua, and taia, whose backbone size is the same as the instance size here Whereas, the instances scra and nuga have more than one optimal solution, ie, the backbone size is 0 Here, the backbones of the instances scra and nuga are not significant for algorithm design Table Backbone size and optimal solutions of instances from the QAPLIB Instance Instance size Optimal solutions Backbone size chrb chrc roua scra taia nuga In the mean time, the size of the QAP approximate backbone obtained by methods of Zou et al is far smaller than that of the TSP, so is the average probability of the QAP approximate backbone appearing in the optimal solutions For typical instances from the QAPLIB, Zou et al found by statistic method that the proportion of the backbone in intersection of local optimal solutions 484 JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

10 to the intersection itself is comparatively low Moreover, the size of intersection of local optimal solutions decreases quickly as the number of local optimal solutions being intersected increases (When the number of local optimal solutions increases across 3, the intersection size of most QAP instances decreases to 0) Table gives the statistic results [9] of intersection of local optimal solutions to typical instances from the QAPLIB Table Statistic results of approximate backbone (AB) Instance Instance size Approximate Average probability of AB backbone size appearing in the optimal solutions (%) Iterations times chr5a tai30b tai40b lipa40a tai50b chrb tai80b Biased instance based approximate backbone algorithm In this subsection, we give the new meta-heuristic BI-AB under the analysis of subsection 4 The basic idea of the BI-AB is to transform a QAP instance to its biased QAP instance with unique optimal solution, ie, the backbone, by constructing biased matrices of the original matrices A= ( a ) nxn and B = ( b ) nxn The optimal solution to the new instance is also optimal to the original one, so we need only to solve the new instance After the transformation, we use some existing algorithm (denoted by Λ) as the subordinate algorithm to obtain k local optimal solutions to the new instance and fix the approximate backbone by intersecting such k solutions to obtain smaller search space which is further searched for new and better solutions Algorithm: BI-AB Input: QAP instance QAP(A, B), subordinate algorithm Λ Output: Solution 0 Begin Construct the biased QAP instance QAP( Aˆ, B ˆ) for the instance QAP( A, B) Find an initial solution 0 by using Λ for the QAP( Aˆ, Bˆ ) 3 Do the following: 3 Find k- different solutions,,, k to the QAP( Aˆ, Bˆ ) by Λ 3 Obtain the approximate-backbone a- bone( 0,,, k ) of 0,,, k * 33 Fix a- bone( 0,,, k ) to obtain a new search space S ( QAP( Aˆ, B)) ˆ 34 Run Λ to obtain the solution k 35 If g ( AB ˆ, ˆ) < g ( AB ˆ, ˆ), then set 0 = k k 4 Return 0 End The run time of the BI-AB can be estimated as follows: 0 JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

11 Step Takes Ο( n ) time to construct the biased QAP instance Step Let the run time of Λ to solve the QAP( Aˆ, B ˆ) be Ο ( ) Step 3 Takes ( k ) Ο ( ) time Step 3 Takes Ο ( kn) time to obtain the approximate backbone Step 33 Takes O(n) time Step 34 Takes Ο ( ) time Step 35 Takes O(n ) time Step 4 Takes O() time Summarizing the above, the time complexity of the BI-AB is Ο( n ) + kο ( ) + Ο( kn) There are many advantages of the BI-AB Firstly, the BI-AB provides a framework which can employ any existing algorithm as the subordinate algorithm Λ Secondly, it is really simple to construct a biased QAP instance, though the proof of unique optimal solution is overloaded with details Finally, the size of backbone is remarkably improved by constructing the biased QAP instance, and the effect of the approximate backbone is also enhanced 5 Experiments and analysis Sahni et al [] showed in 976 that there is no ε approximation algorithm for the QAP unless P=NP (ε > 0) Hence, it is the only way to evaluate heuristics by experimental methods The meta-heuristic BI-AB provides a framework, which could employ tabu or greedy algorithm as the subordinate algorithm Since we mainly investigate the application of the approximate backbone in algorithm design, the FANT is incorporated as the subordinate algorithm in order to compare the BI-AB with ABFANT Performance comparison results about FANT, genetic algorithms and tabu search can be found in ref [34] We implement the BI-AB (denoted by BI-ABFANT) in C++ language For performance comparison, we also implement ABFANT and random restart FANT (denoted by Re-FANT) In the experiments, we take the parameter k= The iteration times of the FANT is set to be 000 The Re-FANT runs 3 times FANT and chooses the best solution as the final result Our experiments are conducted on a PC with one P4 4G CPU and 5 MB of main memory running Redhat Linux 90 Experimental results for 0 runs of typical instances from the QAPLIB are given in Table 3, where g opt denotes cost of the optimal solution, t mean the average runtime in seconds, τ min the minimum percentage excess over the optimal solution, and τ mean the average percentage excess over the optimal solution, respectively Table 3 illustrates that the ABFANT, Re-FANT, and BI-ABFANT achieve good performance on most instances from the QAPLIB in nearly same time The ABFANT, Re-FANT, and BI-ABFANT can find optimal solutions to 8 instances out of the total 0 instances in 0 runs It means that the subordinate algorithm FANT is efficient and effective The BI-ABFANT outperforms the ABFANT on average solution quality for instances Conversely, the ABFANT outperforms the BI-ABFANT on average solution quality for instance The BI-ABFANT is better for 9 instances, worse for 3 instances than the Re-FANT on average solution quality, respectively For the remainder instances, both the BI-ABFANT and the Re-FANT always obtain the optimal solutions Summarizing the above, the BI-ABFANT significantly outperforms the ABFANT and the Re-FANT in terms of solution quality 486 JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

12 Table 3 Results of ABFANT, Re-FANT, BI-ABFANT on instances from the QAPLIB (0 runs) ABFANT Re_FANT BI-ABFANT Instance g opt τ min τ mean τ t min τ mean τ (%) (%) mean t min τ mean (%) (%) mean (%) (%) t mean bur6a chr5a chr5b chra chrb chr5a lip40a lip50a nug6a nug30a rou0a scra scr0a ste36b taia taib tai40b tai50a tai80b tho30a Conclusions In this paper, we performed a theoretical analysis for the backbone of the QAP By constructing the biased QAP instance, we proved that it is NP-hard to obtain the backbone of the QAP, so it is reasonable to obtain the approximate backbone by intersection of local optimal solutions The biased instance based approximate backbone algorithm is then proposed Work of this paper provides a case study of theoretical analysis for the backbone of NP-hard problems Moreover, the way of designing algorithms based on the biased instance, casts a light on algorithm design for other NP-hard problems by using the backbone Garey M R, Johnson D S Computers and Intractability: A Guide to the Theory of NP-completeness San Francisco: W H Freeman, Sahni S, Gonzalez T P-complete approximation problems J ACM, 976, 3(3): Glover F, Laguna M Tabu Search Boston: Kluwer Academic Publishers, Dorigo M, DiCaro G The ant colony optimization meta-heuristic In: Corne D, Dorigo M, Glover F, eds New Ideas in Optimization London: McGraw Hill, UK, Duan H B Ant Colony Algorithms: Theory and Applications (in Chinese) Being: Science Press, Chen G L, Wang X F, Zhuang Z Q Genetic Algorithm and its Application (in Chinese) Being: Posts & Telecom Press, Kang L S, Xie Y, You S Y, et al Non numeric Parallel Algorithms: Simulated Annealing (in Chinese) Being: Science Press, Zhou Z H, Cao C G Neutral Networks and Applications (in Chinese) Being: Press of Tsinghua University, Huang W Q, Xu R C Introduction to Modern Computational Theory: Background, Perspective, and Algorithms Research of NP-hard Problems (in Chinese) Being: Science Press, Slaney J, Walsh T Backbones in optimization and approximation In: Proc 7th Intern Joint Conf Artif Intell (IJCAI-0) San Francisco: Morgan Kaufmann Publishers, JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

13 Monasson R, Zecchina R, Kirkpatrick S, et al Determining computational complexity for characteristic phase transition Nature, 998, 400(8): Zhang W X Phase transition and backbones of the asymmetric traveling salesman problem J Artif Intell Res, 004, (): Schneider J Searching for backbones-a high-performance parallel algorithm for solving combinatorial optimization problems Future Gener Comput Syst, 003, 9(): 3[DOI] 4 Zou P, Zhou Z, Chen G L, et al A multilevel reduction algorithm to TSP J Software (in Chinese), 003, 4(): Zhang W X, Looks M A novel local search algorithm for the traveling salesman problem that exploit backbones In: Proc 9th Intern Joint Conf Artif Intell (IJCAI-05), San Francisco: Morgan Kaufmann Publishers, Zhang W X Configuration landscape analysis and backbone guided local search: Part I: Satisifiability and maximum satisfiability Artif Intell, 004, 58(): 6[DOI] 7 Dubois O, Seymour P A backbone-search heuristic for efficient solving of hard 3-SAT formula In: Proc 7th Intern Joint Conf Artif Intell (IJCAI-0) San Francisco: Morgan Kaufmann Publishers, Valnir F J Backbone guided dynamic local search for propositional satisfiability In: Proc 9th Intern Symposium on Artif Intell Math (AI & Math -06) New York: Springer, Zou P, Zhou Z, Chen G L, et al Approximate-backbone guided fast ant algorithms to QAP J Software, 005, 6(0): [DOI] 0 Kilby P, Slaney J, Walsh T The backbone of the traveling salesperson In: Proc 9th Intern Joint Conf Artif Intell (IJCAI05) San Francisco: Morgan Kaufmann Publishers, Taillard E D Fant: fast ant system Techn Report IDSIA-46-98, 998 Burkard R E, Karisch S, Rendl F Qaplib: a quadratic assignment problem library Eur J Oper Res, 99, 55(): 5 9[DOI] 3 Gambardella L, Taillard E, Dorigo M Ant colonies for the QAP J Oper Res Soc, 999, 50(): 67 76[DOI] 4 Middendorf M, Reischle F, Schmeck H Multi colony ant algorithms J Heuristics, 00, 8(3): [DOI] 5 Drezner Z A new genetic algorithm for the quadratic assignment problem Informs J Comput, 003, 5(3): [DOI] 6 Tsuchiya K, Nishiyama T, Tsujita K A deterministic annealing algorithm for a combinatorial optimization problem using replicator equations Phys D: Nonlin Phenom, 00, 49(3): 6 73[DOI] 7 Ishii S, Sato M Doubly constrained network for combinatorial optimization Neurocomputing, 00, 43(-4): 39 57[DOI] 8 Misevicius A A modification of tabu search and its applications to the quadratic assignment problem Inform Techn Control, 003, (7): 0 9 Drezner Z The extended concentric tabu for the quadratic assignment problem Eur J Oper Res, 005, 60(): 46 4[DOI] 30 Oliveira C A S, Pardalos P M, Resende M G C Grasp with path-relinking for the QAP In: Proc 5th Metaheuristics Intern Conf (MIC003) Boston: Kluwer Academic Publishers, Boese K D Cost versus distance in the traveling salesman problem Tech Report CSD-95008, Merz P, Freisleben B Fitness landscapes and memetic algorithms and greedy operators for graph bi-partitioning Evolut Comput, 000, 8(): 6 9[DOI] 33 Reeves C R Landscapes, operators and heuristic search Annals Oper Res, 999, 86(): [DOI] 34 Merz P, Freisleben B A comparison of memetic algorithms, tabu search, and ant colonies for the quadratic assignment problem In: Proc IEEE Congress on Evolut Computation (CEC 99) Piscataway: IEEE Press, JIANG He et al Sci China Ser F-Inf Sci May 008 vol 5 no

AN ANT APPROACH FOR STRUCTURED QUADRATIC ASSIGNMENT PROBLEMS

AN ANT APPROACH FOR STRUCTURED QUADRATIC ASSIGNMENT PROBLEMS AN ANT APPROACH FOR STRUCTURED QUADRATIC ASSIGNMENT PROBLEMS Éric D. Taillard, Luca M. Gambardella IDSIA, Corso Elvezia 36, CH-6900 Lugano, Switzerland. Extended abstract IDSIA-22-97 ABSTRACT. The paper

More information

Ruin and Recreate Principle Based Approach for the Quadratic Assignment Problem

Ruin and Recreate Principle Based Approach for the Quadratic Assignment Problem Ruin and Recreate Principle Based Approach for the Quadratic Assignment Problem Alfonsas Misevicius Kaunas University of Technology, Department of Practical Informatics, Studentu St. 50 400a, LT 3031 Kaunas,

More information

Distance Metrics and Fitness Distance Analysis for the Capacitated Vehicle Routing Problem

Distance Metrics and Fitness Distance Analysis for the Capacitated Vehicle Routing Problem MIC2005. The 6th Metaheuristics International Conference 603 Metrics and Analysis for the Capacitated Vehicle Routing Problem Marek Kubiak Institute of Computing Science, Poznan University of Technology

More information

Solving Multi-Criteria Optimization Problems with Population-Based ACO

Solving Multi-Criteria Optimization Problems with Population-Based ACO Solving Multi-Criteria Optimization Problems with Population-Based ACO Michael Guntsch 1 and Martin Middendorf 2 1 Institute for Applied Computer Science and Formal Description Methods University of Karlsruhe,

More information

Multi-objective Quadratic Assignment Problem instances generator with a known optimum solution

Multi-objective Quadratic Assignment Problem instances generator with a known optimum solution Multi-objective Quadratic Assignment Problem instances generator with a known optimum solution Mădălina M. Drugan Artificial Intelligence lab, Vrije Universiteit Brussel, Pleinlaan 2, B-1050 Brussels,

More information

New approaches to evaporation in ant colony optimization algorithms

New approaches to evaporation in ant colony optimization algorithms New approaches to evaporation in ant colony optimization algorithms E. Foundas A. Vlachos Department of Informatics University of Piraeus Piraeus 85 34 Greece Abstract In Ant Colony Algorithms (ACA), artificial

More information

Hamiltonian(t) - An Ant-Inspired Heuristic for Recognizing Hamiltonian Graphs

Hamiltonian(t) - An Ant-Inspired Heuristic for Recognizing Hamiltonian Graphs to appear in CEC 99 Hamiltonian(t) - An Ant-Inspired Heuristic for Recognizing Hamiltonian Graphs Israel A Wagner and Alfred M Bruckstein 3 IBM Haifa Research Lab, Matam, Haifa 395, Israel Department of

More information

Algorithms and Complexity theory

Algorithms and Complexity theory Algorithms and Complexity theory Thibaut Barthelemy Some slides kindly provided by Fabien Tricoire University of Vienna WS 2014 Outline 1 Algorithms Overview How to write an algorithm 2 Complexity theory

More information

Lin-Kernighan Heuristic. Simulated Annealing

Lin-Kernighan Heuristic. Simulated Annealing DM63 HEURISTICS FOR COMBINATORIAL OPTIMIZATION Lecture 6 Lin-Kernighan Heuristic. Simulated Annealing Marco Chiarandini Outline 1. Competition 2. Variable Depth Search 3. Simulated Annealing DM63 Heuristics

More information

Sensitive Ant Model for Combinatorial Optimization

Sensitive Ant Model for Combinatorial Optimization Sensitive Ant Model for Combinatorial Optimization CAMELIA CHIRA cchira@cs.ubbcluj.ro D. DUMITRESCU ddumitr@cs.ubbcluj.ro CAMELIA-MIHAELA PINTEA cmpintea@cs.ubbcluj.ro Abstract: A combinatorial optimization

More information

Metaheuristics and Local Search

Metaheuristics and Local Search Metaheuristics and Local Search 8000 Discrete optimization problems Variables x 1,..., x n. Variable domains D 1,..., D n, with D j Z. Constraints C 1,..., C m, with C i D 1 D n. Objective function f :

More information

Metaheuristics and Local Search. Discrete optimization problems. Solution approaches

Metaheuristics and Local Search. Discrete optimization problems. Solution approaches Discrete Mathematics for Bioinformatics WS 07/08, G. W. Klau, 31. Januar 2008, 11:55 1 Metaheuristics and Local Search Discrete optimization problems Variables x 1,...,x n. Variable domains D 1,...,D n,

More information

Motivation, Basic Concepts, Basic Methods, Travelling Salesperson Problem (TSP), Algorithms

Motivation, Basic Concepts, Basic Methods, Travelling Salesperson Problem (TSP), Algorithms Motivation, Basic Concepts, Basic Methods, Travelling Salesperson Problem (TSP), Algorithms 1 What is Combinatorial Optimization? Combinatorial Optimization deals with problems where we have to search

More information

Capacitor Placement for Economical Electrical Systems using Ant Colony Search Algorithm

Capacitor Placement for Economical Electrical Systems using Ant Colony Search Algorithm Capacitor Placement for Economical Electrical Systems using Ant Colony Search Algorithm Bharat Solanki Abstract The optimal capacitor placement problem involves determination of the location, number, type

More information

Solving the Homogeneous Probabilistic Traveling Salesman Problem by the ACO Metaheuristic

Solving the Homogeneous Probabilistic Traveling Salesman Problem by the ACO Metaheuristic Solving the Homogeneous Probabilistic Traveling Salesman Problem by the ACO Metaheuristic Leonora Bianchi 1, Luca Maria Gambardella 1 and Marco Dorigo 2 1 IDSIA, Strada Cantonale Galleria 2, CH-6928 Manno,

More information

A.I.: Beyond Classical Search

A.I.: Beyond Classical Search A.I.: Beyond Classical Search Random Sampling Trivial Algorithms Generate a state randomly Random Walk Randomly pick a neighbor of the current state Both algorithms asymptotically complete. Overview Previously

More information

A new set of hybrid heuristics for the quadratic assignment problem

A new set of hybrid heuristics for the quadratic assignment problem A new set of hybrid heuristics for the quadratic assignment problem André de São Thiago Moreira Valdir Agustinho de Melo Paulo Oswaldo Boaventura Netto Programa de Engenharia de Produção, COPPE/UFRJ CP

More information

ON THE DUALITY FEATURE OF P-CLASS PROBLEMS AND NP COMPLETE PROBLEMS

ON THE DUALITY FEATURE OF P-CLASS PROBLEMS AND NP COMPLETE PROBLEMS ON THE DUALITY FEATURE OF P-CLASS PROBLEMS AND NP COMPLETE PROBLEMS WenhongTian 1, 2 1 University of Electronic Science and Technology of China, Chengdu, China. 2 Chongqing Institute of Green and Intelligent

More information

An Interesting Perspective to the P versus NP Problem

An Interesting Perspective to the P versus NP Problem An Interesting Perspective to the P versus NP Problem Wenming Zhang School of Economics and Management, Northwest University, Xi an, 710127, China wenming@nwu.edu.cn Abstract. We discuss the P versus NP

More information

Ant Colony Optimization for Resource-Constrained Project Scheduling

Ant Colony Optimization for Resource-Constrained Project Scheduling Ant Colony Optimization for Resource-Constrained Project Scheduling Daniel Merkle, Martin Middendorf 2, Hartmut Schmeck 3 Institute for Applied Computer Science and Formal Description Methods University

More information

Ant Colony Optimization: an introduction. Daniel Chivilikhin

Ant Colony Optimization: an introduction. Daniel Chivilikhin Ant Colony Optimization: an introduction Daniel Chivilikhin 03.04.2013 Outline 1. Biological inspiration of ACO 2. Solving NP-hard combinatorial problems 3. The ACO metaheuristic 4. ACO for the Traveling

More information

New Instances for the Single Machine Total Weighted Tardiness Problem

New Instances for the Single Machine Total Weighted Tardiness Problem HELMUT-SCHMIDT-UNIVERSITÄT UNIVERSITÄT DER BUNDESWEHR HAMBURG LEHRSTUHL FÜR BETRIEBSWIRTSCHAFTSLEHRE, INSBES. LOGISTIK-MANAGEMENT Prof. Dr. M. J. Geiger Arbeitspapier / Research Report RR-10-03-01 March

More information

OPTIMIZED RESOURCE IN SATELLITE NETWORK BASED ON GENETIC ALGORITHM. Received June 2011; revised December 2011

OPTIMIZED RESOURCE IN SATELLITE NETWORK BASED ON GENETIC ALGORITHM. Received June 2011; revised December 2011 International Journal of Innovative Computing, Information and Control ICIC International c 2012 ISSN 1349-4198 Volume 8, Number 12, December 2012 pp. 8249 8256 OPTIMIZED RESOURCE IN SATELLITE NETWORK

More information

Minimum Linear Arrangements

Minimum Linear Arrangements Minimum Linear Arrangements Rafael Andrade, Tibérius Bonates, Manoel Câmpelo, Mardson Ferreira ParGO - Research team in Parallel computing, Graph theory and Optimization Department of Statistics and Applied

More information

A Note on the Density of the Multiple Subset Sum Problems

A Note on the Density of the Multiple Subset Sum Problems A Note on the Density of the Multiple Subset Sum Problems Yanbin Pan and Feng Zhang Key Laboratory of Mathematics Mechanization, Academy of Mathematics and Systems Science, Chinese Academy of Sciences,

More information

Meta heuristic algorithms for parallel identical machines scheduling problem with weighted late work criterion and common due date

Meta heuristic algorithms for parallel identical machines scheduling problem with weighted late work criterion and common due date DOI 10.1186/s40064-015-1559-5 RESEARCH Open Access Meta heuristic algorithms for parallel identical machines scheduling problem with weighted late work criterion and common due date Zhenzhen Xu, Yongxing

More information

Computational complexity theory

Computational complexity theory Computational complexity theory Introduction to computational complexity theory Complexity (computability) theory deals with two aspects: Algorithm s complexity. Problem s complexity. References S. Cook,

More information

2 Notation and Preliminaries

2 Notation and Preliminaries On Asymmetric TSP: Transformation to Symmetric TSP and Performance Bound Ratnesh Kumar Haomin Li epartment of Electrical Engineering University of Kentucky Lexington, KY 40506-0046 Abstract We show that

More information

XLVI Pesquisa Operacional na Gestão da Segurança Pública

XLVI Pesquisa Operacional na Gestão da Segurança Pública A linear formulation with O(n 2 ) variables for the quadratic assignment problem Serigne Gueye and Philippe Michelon Université d Avignon et des Pays du Vaucluse, Laboratoire d Informatique d Avignon (LIA),

More information

Self-Adaptive Ant Colony System for the Traveling Salesman Problem

Self-Adaptive Ant Colony System for the Traveling Salesman Problem Proceedings of the 29 IEEE International Conference on Systems, Man, and Cybernetics San Antonio, TX, USA - October 29 Self-Adaptive Ant Colony System for the Traveling Salesman Problem Wei-jie Yu, Xiao-min

More information

Unit 1A: Computational Complexity

Unit 1A: Computational Complexity Unit 1A: Computational Complexity Course contents: Computational complexity NP-completeness Algorithmic Paradigms Readings Chapters 3, 4, and 5 Unit 1A 1 O: Upper Bounding Function Def: f(n)= O(g(n)) if

More information

Scheduling Parallel Jobs with Linear Speedup

Scheduling Parallel Jobs with Linear Speedup Scheduling Parallel Jobs with Linear Speedup Alexander Grigoriev and Marc Uetz Maastricht University, Quantitative Economics, P.O.Box 616, 6200 MD Maastricht, The Netherlands. Email: {a.grigoriev, m.uetz}@ke.unimaas.nl

More information

Ant Algorithms. Ant Algorithms. Ant Algorithms. Ant Algorithms. G5BAIM Artificial Intelligence Methods. Finally. Ant Algorithms.

Ant Algorithms. Ant Algorithms. Ant Algorithms. Ant Algorithms. G5BAIM Artificial Intelligence Methods. Finally. Ant Algorithms. G5BAIM Genetic Algorithms G5BAIM Artificial Intelligence Methods Dr. Rong Qu Finally.. So that s why we ve been getting pictures of ants all this time!!!! Guy Theraulaz Ants are practically blind but they

More information

Permutation distance measures for memetic algorithms with population management

Permutation distance measures for memetic algorithms with population management MIC2005: The Sixth Metaheuristics International Conference??-1 Permutation distance measures for memetic algorithms with population management Marc Sevaux Kenneth Sörensen University of Valenciennes, CNRS,

More information

PROBLEM SOLVING AND SEARCH IN ARTIFICIAL INTELLIGENCE

PROBLEM SOLVING AND SEARCH IN ARTIFICIAL INTELLIGENCE Artificial Intelligence, Computational Logic PROBLEM SOLVING AND SEARCH IN ARTIFICIAL INTELLIGENCE Lecture 4 Metaheuristic Algorithms Sarah Gaggl Dresden, 5th May 2017 Agenda 1 Introduction 2 Constraint

More information

Implementation of Travelling Salesman Problem Using ant Colony Optimization

Implementation of Travelling Salesman Problem Using ant Colony Optimization RESEARCH ARTICLE OPEN ACCESS Implementation of Travelling Salesman Problem Using ant Colony Optimization Gaurav Singh, Rashi Mehta, Sonigoswami, Sapna Katiyar* ABES Institute of Technology, NH-24, Vay

More information

Long Range Frustration in Finite Connectivity Spin Glasses: Application to the random K-satisfiability problem

Long Range Frustration in Finite Connectivity Spin Glasses: Application to the random K-satisfiability problem arxiv:cond-mat/0411079v1 [cond-mat.dis-nn] 3 Nov 2004 Long Range Frustration in Finite Connectivity Spin Glasses: Application to the random K-satisfiability problem Haijun Zhou Max-Planck-Institute of

More information

Research Article An Auxiliary Function Method for Global Minimization in Integer Programming

Research Article An Auxiliary Function Method for Global Minimization in Integer Programming Mathematical Problems in Engineering Volume 2011, Article ID 402437, 13 pages doi:10.1155/2011/402437 Research Article An Auxiliary Function Method for Global Minimization in Integer Programming Hongwei

More information

The Multiple Traveling Salesman Problem with Time Windows: Bounds for the Minimum Number of Vehicles

The Multiple Traveling Salesman Problem with Time Windows: Bounds for the Minimum Number of Vehicles The Multiple Traveling Salesman Problem with Time Windows: Bounds for the Minimum Number of Vehicles Snežana Mitrović-Minić Ramesh Krishnamurti School of Computing Science, Simon Fraser University, Burnaby,

More information

Backbone Guided Tabu Search for Solving the UBQP Problem

Backbone Guided Tabu Search for Solving the UBQP Problem Noname manuscript No. (will be inserted by the editor) Backbone Guided Tabu Search for Solving the UBQP Problem Yang Wang Zhipeng Lü Fred Glover Jin-Kao Hao Received: date / Accepted: date Abstract We

More information

Theory of Computation Chapter 1: Introduction

Theory of Computation Chapter 1: Introduction Theory of Computation Chapter 1: Introduction Guan-Shieng Huang Sep. 20, 2006 Feb. 9, 2009 0-0 Text Book Computational Complexity, by C. H. Papadimitriou, Addison-Wesley, 1994. 1 References Garey, M.R.

More information

Computational complexity theory

Computational complexity theory Computational complexity theory Introduction to computational complexity theory Complexity (computability) theory deals with two aspects: Algorithm s complexity. Problem s complexity. References S. Cook,

More information

Generalization of Dominance Relation-Based Replacement Rules for Memetic EMO Algorithms

Generalization of Dominance Relation-Based Replacement Rules for Memetic EMO Algorithms Generalization of Dominance Relation-Based Replacement Rules for Memetic EMO Algorithms Tadahiko Murata 1, Shiori Kaige 2, and Hisao Ishibuchi 2 1 Department of Informatics, Kansai University 2-1-1 Ryozenji-cho,

More information

A pruning pattern list approach to the permutation flowshop scheduling problem

A pruning pattern list approach to the permutation flowshop scheduling problem A pruning pattern list approach to the permutation flowshop scheduling problem Takeshi Yamada NTT Communication Science Laboratories, 2-4 Hikaridai, Seika-cho, Soraku-gun, Kyoto 619-02, JAPAN E-mail :

More information

Approximating the minimum quadratic assignment problems

Approximating the minimum quadratic assignment problems Approximating the minimum quadratic assignment problems Refael Hassin Asaf Levin Maxim Sviridenko Abstract We consider the well-known minimum quadratic assignment problem. In this problem we are given

More information

Scheduling Linear Deteriorating Jobs with an Availability Constraint on a Single Machine 1

Scheduling Linear Deteriorating Jobs with an Availability Constraint on a Single Machine 1 Scheduling Linear Deteriorating Jobs with an Availability Constraint on a Single Machine 1 Min Ji a, b, 2 Yong He b, 3 T.C.E. Cheng c, 4 a College of Computer Science & Information Engineering, Zhejiang

More information

SIMU L TED ATED ANNEA L NG ING

SIMU L TED ATED ANNEA L NG ING SIMULATED ANNEALING Fundamental Concept Motivation by an analogy to the statistical mechanics of annealing in solids. => to coerce a solid (i.e., in a poor, unordered state) into a low energy thermodynamic

More information

Algorithms and Complexity Theory. Chapter 8: Introduction to Complexity. Computer Science - Durban - September 2005

Algorithms and Complexity Theory. Chapter 8: Introduction to Complexity. Computer Science - Durban - September 2005 Algorithms and Complexity Theory Chapter 8: Introduction to Complexity Jules-R Tapamo Computer Science - Durban - September 2005 Contents 1 Introduction 2 1.1 Dynamic programming...................................

More information

Artificial Intelligence Heuristic Search Methods

Artificial Intelligence Heuristic Search Methods Artificial Intelligence Heuristic Search Methods Chung-Ang University, Jaesung Lee The original version of this content is created by School of Mathematics, University of Birmingham professor Sandor Zoltan

More information

The minimum G c cut problem

The minimum G c cut problem The minimum G c cut problem Abstract In this paper we define and study the G c -cut problem. Given a complete undirected graph G = (V ; E) with V = n, edge weighted by w(v i, v j ) 0 and an undirected

More information

5. Simulated Annealing 5.1 Basic Concepts. Fall 2010 Instructor: Dr. Masoud Yaghini

5. Simulated Annealing 5.1 Basic Concepts. Fall 2010 Instructor: Dr. Masoud Yaghini 5. Simulated Annealing 5.1 Basic Concepts Fall 2010 Instructor: Dr. Masoud Yaghini Outline Introduction Real Annealing and Simulated Annealing Metropolis Algorithm Template of SA A Simple Example References

More information

Variable Objective Search

Variable Objective Search Variable Objective Search Sergiy Butenko, Oleksandra Yezerska, and Balabhaskar Balasundaram Abstract This paper introduces the variable objective search framework for combinatorial optimization. The method

More information

A New Approximation Algorithm for the Asymmetric TSP with Triangle Inequality By Markus Bläser

A New Approximation Algorithm for the Asymmetric TSP with Triangle Inequality By Markus Bläser A New Approximation Algorithm for the Asymmetric TSP with Triangle Inequality By Markus Bläser Presented By: Chris Standish chriss@cs.tamu.edu 23 November 2005 1 Outline Problem Definition Frieze s Generic

More information

Discrete evaluation and the particle swarm algorithm

Discrete evaluation and the particle swarm algorithm Volume 12 Discrete evaluation and the particle swarm algorithm Tim Hendtlass and Tom Rodgers Centre for Intelligent Systems and Complex Processes Swinburne University of Technology P. O. Box 218 Hawthorn

More information

Single Machine Scheduling with a Non-renewable Financial Resource

Single Machine Scheduling with a Non-renewable Financial Resource Single Machine Scheduling with a Non-renewable Financial Resource Evgeny R. Gafarov a, Alexander A. Lazarev b Institute of Control Sciences of the Russian Academy of Sciences, Profsoyuznaya st. 65, 117997

More information

Approximation complexity of min-max (regret) versions of shortest path, spanning tree, and knapsack

Approximation complexity of min-max (regret) versions of shortest path, spanning tree, and knapsack Approximation complexity of min-max (regret) versions of shortest path, spanning tree, and knapsack Hassene Aissi, Cristina Bazgan, and Daniel Vanderpooten LAMSADE, Université Paris-Dauphine, France {aissi,bazgan,vdp}@lamsade.dauphine.fr

More information

On a Polynomial Fractional Formulation for Independence Number of a Graph

On a Polynomial Fractional Formulation for Independence Number of a Graph On a Polynomial Fractional Formulation for Independence Number of a Graph Balabhaskar Balasundaram and Sergiy Butenko Department of Industrial Engineering, Texas A&M University, College Station, Texas

More information

R u t c o r Research R e p o r t. Uniform partitions and Erdös-Ko-Rado Theorem a. Vladimir Gurvich b. RRR , August, 2009

R u t c o r Research R e p o r t. Uniform partitions and Erdös-Ko-Rado Theorem a. Vladimir Gurvich b. RRR , August, 2009 R u t c o r Research R e p o r t Uniform partitions and Erdös-Ko-Rado Theorem a Vladimir Gurvich b RRR 16-2009, August, 2009 RUTCOR Rutgers Center for Operations Research Rutgers University 640 Bartholomew

More information

Totally unimodular matrices. Introduction to integer programming III: Network Flow, Interval Scheduling, and Vehicle Routing Problems

Totally unimodular matrices. Introduction to integer programming III: Network Flow, Interval Scheduling, and Vehicle Routing Problems Totally unimodular matrices Introduction to integer programming III: Network Flow, Interval Scheduling, and Vehicle Routing Problems Martin Branda Charles University in Prague Faculty of Mathematics and

More information

The Complexity of the Permanent and Related Problems

The Complexity of the Permanent and Related Problems The Complexity of the Permanent and Related Problems Tim Abbott and Alex Schwendner May 9, 2007 Contents 1 The Permanent 2 1.1 Variations on the Determinant...................... 2 1.2 Graph Interpretation...........................

More information

A new restart procedure for combinatorial optimization and its convergence

A new restart procedure for combinatorial optimization and its convergence A new restart procedure for combinatorial optimization and its convergence Davide Palmigiani and Giovanni Sebastiani arxiv:79.6449v [math.oc] 8 Sep 27 Abstract We propose a new iterative procedure to optimize

More information

GRASP with path-relinking for the weighted maximum satisfiability problem

GRASP with path-relinking for the weighted maximum satisfiability problem GRASP with path-relinking for the weighted maximum satisfiability problem Paola Festa 1, Panos M. Pardalos 2, Leonidas S. Pitsoulis 3, and Mauricio G. C. Resende 4 1 Department of Mathematics and Applications,

More information

The Complexity of Maximum. Matroid-Greedoid Intersection and. Weighted Greedoid Maximization

The Complexity of Maximum. Matroid-Greedoid Intersection and. Weighted Greedoid Maximization Department of Computer Science Series of Publications C Report C-2004-2 The Complexity of Maximum Matroid-Greedoid Intersection and Weighted Greedoid Maximization Taneli Mielikäinen Esko Ukkonen University

More information

MICROCANONICAL OPTIMIZATION APPLIED TO THE TRAVELING SALESMAN PROBLEM

MICROCANONICAL OPTIMIZATION APPLIED TO THE TRAVELING SALESMAN PROBLEM International Journal of Modern Physics C, Vol. 9, No. 1 (1998) 133 146 c World Scientific Publishing Company MICROCANONICAL OPTIMIZATION APPLIED TO THE TRAVELING SALESMAN PROBLEM ALEXANDRE LINHARES Computação

More information

Meta-heuristics for combinatorial optimisation

Meta-heuristics for combinatorial optimisation Meta-heuristics for combinatorial optimisation João Pedro Pedroso Centro de Investigação Operacional Faculdade de Ciências da Universidade de Lisboa and Departamento de Ciência de Computadores Faculdade

More information

HIGH-PERFORMANCE circuits consume a considerable

HIGH-PERFORMANCE circuits consume a considerable 1166 IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, VOL 17, NO 11, NOVEMBER 1998 A Matrix Synthesis Approach to Thermal Placement Chris C N Chu D F Wong Abstract In this

More information

A NEW MODELING AND SOLUTION APPROACH FOR THE NUMBER PARTITIONING PROBLEM

A NEW MODELING AND SOLUTION APPROACH FOR THE NUMBER PARTITIONING PROBLEM A NEW MODELING AND SOLUTION APPROACH FOR THE NUMBER PARTITIONING PROBLEM BAHRAM ALIDAEE, FRED GLOVER, GARY A. KOCHENBERGER, AND CESAR REGO Received 24 November 2003 and in revised form 9 May 2004 The number

More information

A Continuation Approach Using NCP Function for Solving Max-Cut Problem

A Continuation Approach Using NCP Function for Solving Max-Cut Problem A Continuation Approach Using NCP Function for Solving Max-Cut Problem Xu Fengmin Xu Chengxian Ren Jiuquan Abstract A continuous approach using NCP function for approximating the solution of the max-cut

More information

Introduction to Complexity Theory

Introduction to Complexity Theory Introduction to Complexity Theory Read K & S Chapter 6. Most computational problems you will face your life are solvable (decidable). We have yet to address whether a problem is easy or hard. Complexity

More information

Algorithms. Outline! Approximation Algorithms. The class APX. The intelligence behind the hardware. ! Based on

Algorithms. Outline! Approximation Algorithms. The class APX. The intelligence behind the hardware. ! Based on 6117CIT - Adv Topics in Computing Sci at Nathan 1 Algorithms The intelligence behind the hardware Outline! Approximation Algorithms The class APX! Some complexity classes, like PTAS and FPTAS! Illustration

More information

MAX-2-SAT: How Good is Tabu Search in the Worst-Case?

MAX-2-SAT: How Good is Tabu Search in the Worst-Case? MAX-2-SAT: How Good is Tabu Search in the Worst-Case? Monaldo Mastrolilli IDSIA Galleria 2, 6928 Manno, Switzerland monaldo@idsia.ch Luca Maria Gambardella IDSIA Galleria 2, 6928 Manno, Switzerland luca@idsia.ch

More information

Introduction to integer programming III:

Introduction to integer programming III: Introduction to integer programming III: Network Flow, Interval Scheduling, and Vehicle Routing Problems Martin Branda Charles University in Prague Faculty of Mathematics and Physics Department of Probability

More information

Methods for finding optimal configurations

Methods for finding optimal configurations CS 1571 Introduction to AI Lecture 9 Methods for finding optimal configurations Milos Hauskrecht milos@cs.pitt.edu 5329 Sennott Square Search for the optimal configuration Optimal configuration search:

More information

This article was published in an Elsevier journal. The attached copy is furnished to the author for non-commercial research and education use, including for instruction at the author s institution, sharing

More information

NP-Completeness. NP-Completeness 1

NP-Completeness. NP-Completeness 1 NP-Completeness Reference: Computers and Intractability: A Guide to the Theory of NP-Completeness by Garey and Johnson, W.H. Freeman and Company, 1979. NP-Completeness 1 General Problems, Input Size and

More information

Problem Complexity Classes

Problem Complexity Classes Problem Complexity Classes P, NP, NP-Completeness and Complexity of Approximation Joshua Knowles School of Computer Science The University of Manchester COMP60342 - Week 2 2.15, March 20th 2015 In This

More information

Solving Combinatorial Optimization Problems via Reformulation and Adaptive Memory Metaheuristics

Solving Combinatorial Optimization Problems via Reformulation and Adaptive Memory Metaheuristics Solving Combinatorial Optimization Problems via Reformulation and Adaptive Memory Metaheuristics Gary A. Kochenberger a, Fred Glover b, Bahram Alidaee c and Cesar Rego d a b c d School of Business, University

More information

Single Solution-based Metaheuristics

Single Solution-based Metaheuristics Parallel Cooperative Optimization Research Group Single Solution-based Metaheuristics E-G. Talbi Laboratoire d Informatique Fondamentale de Lille Single solution-based metaheuristics Improvement of a solution.

More information

Feasibility-Preserving Crossover for Maximum k-coverage Problem

Feasibility-Preserving Crossover for Maximum k-coverage Problem Feasibility-Preserving Crossover for Maximum -Coverage Problem Yourim Yoon School of Computer Science & Engineering Seoul National University Sillim-dong, Gwana-gu Seoul, 151-744, Korea yryoon@soar.snu.ac.r

More information

The Greedy Algorithm for the Symmetric TSP

The Greedy Algorithm for the Symmetric TSP The Greedy Algorithm for the Symmetric TSP Gregory Gutin Anders Yeo Abstract We corrected proofs of two results on the greedy algorithm for the Symmetric TSP and answered a question in Gutin and Yeo, Oper.

More information

Mining Spatial Trends by a Colony of Cooperative Ant Agents

Mining Spatial Trends by a Colony of Cooperative Ant Agents Mining Spatial Trends by a Colony of Cooperative Ant Agents Ashan Zarnani Masoud Rahgozar Abstract Large amounts of spatially referenced data has been aggregated in various application domains such as

More information

On-line Bin-Stretching. Yossi Azar y Oded Regev z. Abstract. We are given a sequence of items that can be packed into m unit size bins.

On-line Bin-Stretching. Yossi Azar y Oded Regev z. Abstract. We are given a sequence of items that can be packed into m unit size bins. On-line Bin-Stretching Yossi Azar y Oded Regev z Abstract We are given a sequence of items that can be packed into m unit size bins. In the classical bin packing problem we x the size of the bins and try

More information

Design and Analysis of Algorithms

Design and Analysis of Algorithms CSE 0, Winter 08 Design and Analysis of Algorithms Lecture 8: Consolidation # (DP, Greed, NP-C, Flow) Class URL: http://vlsicad.ucsd.edu/courses/cse0-w8/ Followup on IGO, Annealing Iterative Global Optimization

More information

GRASP heuristics for discrete & continuous global optimization

GRASP heuristics for discrete & continuous global optimization GRASP heuristics for discrete & continuous global optimization Tutorial given at Learning and Intelligent Optimization Conference (LION 8) Gainesville, Florida February 17, 2014 Copyright @2014 AT&T Intellectual

More information

A parallel metaheuristics for the single machine total weighted tardiness problem with sequence-dependent setup times

A parallel metaheuristics for the single machine total weighted tardiness problem with sequence-dependent setup times A parallel metaheuristics for the single machine total weighted tardiness problem with sequence-dependent setup times Wojciech Bożejko Wroc law University of Technology Institute of Computer Engineering,

More information

Computability and Complexity Theory

Computability and Complexity Theory Discrete Math for Bioinformatics WS 09/10:, by A Bockmayr/K Reinert, January 27, 2010, 18:39 9001 Computability and Complexity Theory Computability and complexity Computability theory What problems can

More information

P,NP, NP-Hard and NP-Complete

P,NP, NP-Hard and NP-Complete P,NP, NP-Hard and NP-Complete We can categorize the problem space into two parts Solvable Problems Unsolvable problems 7/11/2011 1 Halting Problem Given a description of a program and a finite input, decide

More information

12. LOCAL SEARCH. gradient descent Metropolis algorithm Hopfield neural networks maximum cut Nash equilibria

12. LOCAL SEARCH. gradient descent Metropolis algorithm Hopfield neural networks maximum cut Nash equilibria 12. LOCAL SEARCH gradient descent Metropolis algorithm Hopfield neural networks maximum cut Nash equilibria Lecture slides by Kevin Wayne Copyright 2005 Pearson-Addison Wesley h ttp://www.cs.princeton.edu/~wayne/kleinberg-tardos

More information

An ACO Algorithm for the Most Probable Explanation Problem

An ACO Algorithm for the Most Probable Explanation Problem An ACO Algorithm for the Most Probable Explanation Problem Haipeng Guo 1, Prashanth R. Boddhireddy 2, and William H. Hsu 3 1 Department of Computer Science, Hong Kong University of Science and Technology

More information

Multiple Sequence Alignment: Complexity, Gunnar Klau, January 12, 2006, 12:

Multiple Sequence Alignment: Complexity, Gunnar Klau, January 12, 2006, 12: Multiple Sequence Alignment: Complexity, Gunnar Klau, January 12, 2006, 12:23 6001 6.1 Computing MSAs So far, we have talked about how to score MSAs (including gaps and benchmarks). But: how do we compute

More information

LCPI method to find optimal solutions of nonlinear programming problems

LCPI method to find optimal solutions of nonlinear programming problems ISSN 1 746-7233, Engl, UK World Journal of Modelling Simulation Vol. 14 2018) No. 1, pp. 50-57 LCPI method to find optimal solutions of nonlinear programming problems M.H. Noori Skari, M. Ghaznavi * Faculty

More information

Verification Methods for Ill-Conditioned Relaxation Frameworks of Quadratic Assignment Problems.

Verification Methods for Ill-Conditioned Relaxation Frameworks of Quadratic Assignment Problems. Methods for Ill-Conditioned Relaxation Frameworks of Quadratic Assignment Problems. Hamburg University of Technology October 16, 2012 Outline Outline Outline Outline Introduction Applications Complexity

More information

Tightening a Discrete Formulation of the Quadratic Assignment Problem

Tightening a Discrete Formulation of the Quadratic Assignment Problem A publication of 1309 CHEMICAL ENGINEERING TRANSACTIONS VOL. 32, 2013 Chief Editors: Sauro Pierucci, Jiří J. Klemeš Copyright 2013, AIDIC Servizi S.r.l., ISBN 978-88-95608-23-5; ISSN 1974-9791 The Italian

More information

Research Article A Novel Differential Evolution Invasive Weed Optimization Algorithm for Solving Nonlinear Equations Systems

Research Article A Novel Differential Evolution Invasive Weed Optimization Algorithm for Solving Nonlinear Equations Systems Journal of Applied Mathematics Volume 2013, Article ID 757391, 18 pages http://dx.doi.org/10.1155/2013/757391 Research Article A Novel Differential Evolution Invasive Weed Optimization for Solving Nonlinear

More information

A reactive framework for Ant Colony Optimization

A reactive framework for Ant Colony Optimization A reactive framework for Ant Colony Optimization Madjid Khichane 1,2, Patrick Albert 1, and Christine Solnon 2 1 ILOG SA, 9 rue de Verdun, 94253 Gentilly cedex, France {mkhichane,palbert}@ilog.fr 2 LIRIS

More information

Informatik-Bericht Nr

Informatik-Bericht Nr F Informatik-Bericht Nr. 2007-4 Schriftenreihe Fachbereich Informatik, Fachhochschule Trier First Come, First Served Tour Scheduling with Priorities Heinz Schmitz and Sebastian Niemann Fachhochschule Trier,

More information

A Mixed-Integer Linear Program for the Traveling Salesman Problem with Structured Time Windows

A Mixed-Integer Linear Program for the Traveling Salesman Problem with Structured Time Windows A Mixed-Integer Linear Program for the Traveling Salesman Problem with Structured Time Windows Philipp Hungerländer Christian Truden 5th January 2017 Abstract In this extended abstract we introduce the

More information

Variable Neighborhood Decomposition Search

Variable Neighborhood Decomposition Search Journal of Heuristics, 7: 335 350, 2001 c 2001 Kluwer Academic Publishers Variable Neighborhood Decomposition Search PIERRE HANSEN AND NENAD MLADENOVIĆ Gerad, Ecole des Hautes Etudes Commerciales, Montreal,

More information

Consecutive ones Block for Symmetric Matrices

Consecutive ones Block for Symmetric Matrices Consecutive ones Block for Symmetric Matrices Rui Wang, FrancisCM Lau Department of Computer Science and Information Systems The University of Hong Kong, Hong Kong, PR China Abstract We show that a cubic

More information

Polynomial-Solvability of N P-class Problems

Polynomial-Solvability of N P-class Problems Polynomial-Solability of N P-class Problems Anatoly Panyuko paniukoa@susu.ru arxi:1409.03755 [cs.cc] 24 Aug 2018 Abstract LetHamiltoniancomplement ofthegraphg = (V(G), E(G))betheminimal cardinality set

More information