rebalancing problem - arXiv · 2016-05-05 · rebalancing problem F abio Cruz, Anand Subramanian...

A heuristic algorithm for a single vehicle static bike sharing

rebalancing problem

Fabio Cruz Anand SubramanianCentro de Informatica

Universidade Federal da Paraıba CEP 58059-900 Joao Pessoa Brazilfabiocbadiufpbbr anandctufpbbr

Bruno P Bruck Manuel IoriDISMI University of Modena and Reggio Emilia 42122 Reggio Emilia Italy

brunopbruckgmailcom manuelioriunimoreit

Working Paper UFPB ndash May 2016

Abstract

The static bike rebalancing problem (SBRP) concerns the task of repositioning bikes among stationsin self-service bike-sharing systems This problem can be seen as a variant of the one-commoditypickup and delivery vehicle routing problem where multiple visits are allowed to be performed ateach station ie the demand of a station is allowed to be split Moreover a vehicle may temporarilydrop its load at a station leaving it in excess or alternatively collect more bikes from a station(even all of them) thus leaving it in default Both cases require further visits in order to meet theactual demands of such station This paper deals with a particular case of the SBRP in which onlya single vehicle is available and the objective is to find a least-cost route that meets the demand ofall stations and does not violate the minimum (zero) and maximum (vehicle capacity) load limitsalong the tour Therefore the number of bikes to be collected or delivered at each station should beappropriately determined in order to respect such constraints We propose an iterated local search(ILS) based heuristic to solve the problem The ILS algorithm was tested on 980 benchmark instancesfrom the literature and the results obtained are quite competitive when compared to other existingmethods Moreover our heuristic was capable of finding most of the known optimal solutions andalso of improving the results on a number of open instances

Keywords Bike-sharing Vehicle Routing Split pickup and delivery Iterated local search

1 Introduction

The task of repositioning a commodity from one location to another is a well-known problem arising in

different contexts such as logistics transportation and various disciplines notably industrial engineering

and operations management A practical application arises in self-service bike sharing systems (BSS)

which are becoming increasingly popular in recent years Users rent bikes and return them at stations

distributed over a region A vehicle with limited load capacity then periodically collects and delivers

bikes across different stations so as to rebalance the system

Alternatives to the street traffic are important not only because of its impact in urban congestion

but also in the environment commuting and so on The emerging worldwide BSS are proving to be

an effective solution to mitigate the effects of traffic issues in large urban centers Up to 2009 there

were about 120 bike sharing programs around the world and according to DeMaio (2009) they have

a favorable impact on decreasing traffic congestion improving public health and helping reducing the

1

arX

iv1

605

0070

2v2

[cs

AI]

4 M

ay 2

016

Working Paper

level of CO2 emissions One of the most famous systems is the Velibrsquo system in Paris with 1800 stations

and more than 20000 bikes

In such systems each station has an inventory with a load capacity an initial number of bikes and

consequently a number of free slots where users can return bikes to the system Throughout the day

some stations may have no bike to be rented or free slots to store returned bikes Therefore an attempt

to avoid this scenario which is unpleasant for users is to determine an initial acceptable number of bikes

(and free slots) at each station This task can be done based on demand history and peaks at each

station

The activity of repositioning bikes among stations on a regular basis is called rebalancing and this is

done by one or more vehicles that move bikes from one station to another in order to restore its inventory

to the initial desired configuration As per DeMaio (2009) good rebalancing systems are present in

successful bike sharing programs and since the vehicles move back and forth across an urban area a

vehicle routing optimization can be utilized

The rebalancing is either static performed when nearly no bikes are being used or dynamic which

is done while the system is still in use The static bike rebalancing problem (SBRP) is motivated by

the fact that very few bikes are being used at night Indeed according to Chemla et al (2013a) and

DellrsquoAmico et al (2014) there are many systems that are even closed during the night

In this work we consider the single vehicle SBRP which is clearly NP-hard because it includes

among others the classical traveling salesman problem (TSP) as a special case The SBRP can be seen

as a variant of the one-commodity pickup and delivery TSP (Hernandez-Perez and Salazar-Gonzalez

2004ab) where multiple visits are allowed to be performed at each station ie the demand of a station

is allowed to be split Moreover a vehicle may arbitrarily drop its load at a station leaving it in excess

or alternatively collect more bikes from a station (even all of them) thus leaving it in default Both

cases require further visits in order to meet the actual demands of such station This strategy of allowing

a station to act as a buffer or a temporary depot is denoted as preemption or temporary operation (ie

temporary pickup and temporary dropoff) Finally visits to balanced stations are optional for the SBRP

Salazar-Gonzalez and Santos-Hernandez (2015) considered a similar yet different problem in which an

upper limit is imposed on the number of visits to the customers and to the depot and the single vehicle

that performs the rebalancing is not forced to leave the depot with an empty load

An increasing number of works regarding bike sharing systems and related issues such as the balancing

of their stations has been published over the last years Exact approaches for multiple vehicle SBRPs

were suggested by DellrsquoAmico et al (2014) Di Gaspero et al (2013ab 2015) Kloimullner et al (2015)

Raviv et al (2013) Moreover Alvarez-Valdes et al (2016) DellrsquoAmico et al (2016) Forma et al (2015)

Papazek et al (2014 2013) Raidl et al (2013) Rainer-Harbach et al (2013 2014) also addressed different

types of multiple vehicle SBRPs but with heuristics

Several exact Benchimol et al (2011) Chemla et al (2013a) Erdogan et al (2015 2014) Salazar-

Gonzalez and Santos-Hernandez (2015) and heuristic (Ho and Szeto 2014 Pal and Zhang 2015) algo-

rithms were proposed for single vehicle SBRPs Furthermore in contrast to static rebalancing there

are relatively few works related to dynamic rebalancing Caggiani and Ottomanelli (2013) Chemla et al

(2013b) Contardo et al (2012) Kloimullner et al (2014) Schuijbroek et al (2013)

The works of Chemla et al (2013a) and Erdogan et al (2015) were the only ones to consider the

same variant dealt in the present paper Chemla et al (2013a) proposed a mathematical formulation

over an extended graph where each station is replicated according to an upper bound on the number of

visits Due to its visible intractability two relaxations were developed The authors also presented among

other contributions a polynomial algorithm to compute optimal bike displacements for a given sequence

of vertices which is useful to determine if a route is feasible or not as well as tabu search heuristics

and a branch-and-cut algorithm that solves a relaxation of the problem Erdogan et al (2015) proposed

2

Working Paper

the first exact method for the problem which consists of a branch-and-cut algorithm that makes use of

no-good cuts and they reported optimal solutions for instances with up to 60 stations

Despite the advances on the development of efficient exact approaches for SBRPs heuristic methods

still appear to be more suitable for dealing with medium and large size instances of this challenging

class of problems In addition high quality heuristic solutions may be crucial to improve the runtime

performance of exact algorithms

This work proposes a hybrid iterated local search (ILS) based heuristic for the single vehicle SBRP

considered in Chemla et al (2013a) and Erdogan et al (2015) Hybridized ILS algorithms especially

when combined with randomized variable neighborhood descent (RVND) revealed to be very effective

when solving a large variety of vehicle routing problems DellrsquoAmico et al (2016) Penna et al (2013)

Silva et al (2015) Subramanian (2012) Subramanian et al (2010) Vidal et al (2015) including those

involving only a single vehicle (Blum and Roli 2003 Subramanian and Battarra 2013)

The algorithm that was developed combines successful ingredients from previous works with some

problem-specific procedures suggested in Chemla et al (2013a) to improve the solution quality as well

as to check if a solution is infeasible We also implemented several perturbation mechanisms and the

impact of each possible combination on the solution quality and CPU time are measured by extensive

computational experiments on a subset of challenging test-problems The results obtained on 980 bench-

mark instances from the literature show that our algorithm is quite competitive when compared to other

methods and a number of new best known solutions is reported We also conduct an analysis on how the

performance of the algorithms in terms of solution quality and CPU time varies according to the number

of stations and the vehicle capacity

The remainder of the paper is organized as follows Section 2 presents a formal problem definition

Section 3 describes the proposed heuristic algorithm Section 4 reports and discusses the computational

results and Section 5 contains the concluding remarks

2 Problem description

Let n be the number of stations V = 1 n be the set of vertices representing their locations (station 0

represents the depot) and A be the set of arcs in a complete and directed graph G = (V cup0 A) For each

arc a(ij) isin A there is a cost ca satisfying the triangular inequality (c(ij) + c(jk) ge c(ik)foralli j k isin V )

For each i isin V let pi isin Z be the amount of bikes initially stored pprimei isin Z be the number of bikes

requested by i after the service is performed and di = pprimei minus pi be the total demand When di gt 0 and

di lt 0 we assume that i isin V is a delivery and a pickup station respectively A station i isin V may have

no demand (pi = pprimei) and in this case the visit becomes optional Each station i isin V has a capacity qi isin Zand the depot is assumed to have no bikes ie q0 = p0 = pprime0 = 0 Finally let Q isin Z be the vehicle

capacity

The objective is to find a least-cost route that starts and ends at the depot visits each station with

non-zero demand at least once meets the demands of all stations (ie the initial load pi is changed

to the target demand pprimeiforalli isin V ) and does not violate the minimum (zero) and maximum (Q) load

limits Therefore the number of bikes to be collected or delivered at each visit to a station should be

appropriately determined in order to respect such constraints

Finally stations may serve to perform temporary operations (preemption) either as a temporary

depot or a temporary buffer ie supply more bikes than their initial demand or hold more bikes (without

exceeding its inventory load capacity) and in both cases have their demand satisfied in later visits

Figure 1 shows a graphical representation of an optimal solution for the benchmark instance n20q10D

(n = 20 and Q = 10) The nodes are distributed according to the spatial coordinates of the stations

The positive and negative values next to the nodes are the number of bikes collected and delivered

3

Working Paper

respectively The arcs and their associated values represent the vehicle traveling to the next station in

the sequence and the vehicle load respectively For example the vehicle delivers 2 bikes in the first visit

to station 12 collects 10 at station 10 returns to 12 to deliver 6 more (meeting the demand of 8) and

then travels to station 14 with a load of 4 bikes

0

1

minus5+1

2

minus1

3

minus2

4+7

5 minus8

6

+1

7

+7

8

+9

9+7

10+10

12 minus6minus2

13

+4

14

+6

15 minus5

16 minus9

17

minus9

18

minus5

19

+8

20

minus8

1

10

20

7

6

10

2

0

10

4

10

1

290

7

2

10

5

Figure 1 Representation of optimal solution with value 5989 for instance n20q10D

3 Proposed heuristic

ILS iteratively alternates between local search (intensification) and perturbation (diversification) mech-

anisms with a view of finding high quality solutions In our case we embed a variable neighborhood

descent (VND) (Mladenovic and Hansen 1997) based procedure in the local search phase of the meta-

heuristic As in previous works (eg Penna et al 2013 and Silva et al 2015) the neighborhoods of our

algorithm are examined in a random ordering during the search (RVND)

As opposed to most of the former ILS implementations cited in Section 1 infeasible solutions are

temporary accepted after the application of perturbation moves not only for the sake of diversification

but also as an attempt to escape from local optimal solutions This modification was crucial for the

favorable performance of the heuristic when dealing with the single vehicle SBRP considered here which

appears to be more challenging to solve than other VRPs where ILS was successfully applied to obtain

high quality solutions by only considering the feasible search space

The proposed hybrid heuristic called ILSSBRP combines multiple restarts local search perturbations

mechanisms and a repair phase Figure 2 illustrates the flowchart of ILSSBRP For each of the IR

restarts a feasible initial solution is generated using a simple greedy randomized constructive algorithm

(see Section 33) Next local search perturbation and repair procedures are successively applied until the

4

Working Paper

ILSSBRP

iterR = 0

iterR lt IR

iterILS = 0

Stop

Generate initial solution s

iterILS lt IILS

Is s feasiblef (sprime) lt f

s = RVND(s)

s = AddUnbalanced(s)

f (s) lt f (sprime)

sprime = s

s = Perturb(sprime)

iterILS = 0

iterILS = iterILS + 1

s = sprime

iterR = iterR + 1

No

Yes

YesNo

No

Yes

No

Yes

No

Figure 2 ILSSBRP flowchart

stopping criterion is met that is when the number of consecutive attempts to escape from a local optimal

solution reaches IILS trials Because perturbation moves are allowed to produce infeasible solutions we

implemented a procedure called AddUnbalanced (Chemla et al 2013a) (see Section 32) with the aim of

repairing such solutions Nevertheless there is no guarantee that a solution will be feasible after applying

this procedure When an infeasible solution is not totally repaired and the local search does not find a

move that leads to a feasible solution then the infeasible solution is disregarded Note that perturbation

is always applied over the best solution of the current multi-start iteration Finally ILSSBRP returns the

best solution found among all restarts

5

Working Paper

31 Solution representation

A solution for the single vehicle SBRP considered in the present work can be represented as a sequence of

visits to stations starting and ending at the depot along with the amount of bikes collected or delivered

at each visit

Three vectors are used as data structures to store (i) the route where the first and last element are

fixed at 0 ie the depot (ii) the operation performed by the vehicle during a visit where negative and

positive values indicate the amount of bikes delivered and collected respectively and (iii) the vehicle

load during the route

As in Chemla et al (2013a) a flow network is used to check in polynomial time whether or not a

solution is feasible with respect to bike displacements and vehicle capacity given a sequence of vertices

representing visits to stations A detailed explanation can be found in Appendix A

We also use another data structure which consists of a key-value map composed by n + 1 elements

that store the number of visits performed at each station This is useful for example to check whether

a solution includes all stations with non-zero demand Note that information hold in (ii) is extracted

from the computed bike displacements when solving the max-flow problem (see Appendix A) From such

it is possible to derive in linear time the vehicle loads in (iii) by the adding or subtracting the bike

displacements at each visit

32 Repairing infeasible solutions

As already mentioned infeasible solutions are allowed after perturbations We therefore implemented

a procedure called AddUnbalanced (Chemla et al 2013a) which tries to repair a solution by adding

stations to the route More precisely both the most unbalanced station in excess and in default i and j

respectively are selected and three moves are possible (i) adding arcs (j i) and (i j) after the existing

visit to j (ii) adding arcs (i j) and (j i) after the existing visit to i (iii) if both i and j are not in the

sequence adding (i j) at the end of the sequence before returning to the depot

For example let us consider a scenario where stations i = 12 and j = 14 are the most unbalanced

More precisely i has initially 20 bikes and a demand of minus10 ie a pickup station while j is initially

holding 3 out of 10 (target) bikes ie a delivery station with demand 7 An infeasible solution is

presented in Figure 3a where the referred stations are not balanced that is their demands are not met

since only 4 bikes were collected in station 12 and 4 bikes were delivered at station 14 Figure 3b shows a

modified solution where after the addition of arcs (14 12) and (12 14) a new and feasible configuration

of bike displacements were determined by means of the maximum flow check (see Appendix A) We can

observe that the second visit complements the first one meeting the demand of both stations the vehicle

deliveries 1 bike at station 14 collects the remaining 7 at station 12 now balanced and finally meets the

demand of station 14 by delivering 6 more bikes

It is worth emphasizing that the AddUnbalanced procedure does not necessary lead to a feasible

solution However in general experimental results showed that such procedure has a high level of

success in fully repairing infeasible solutions

33 Constructive Procedure

The pseudocode of the greedy randomized constructive procedure is presented in Alg 1 The algorithm

stores and maintains a list of open vertices (OV ) corresponding to stations whose demands are still not

fully met Stations without demand are also included in this list In order to ensure a level of diversity

during the process of generating an initial solution OV is sorted in random order (line 4)

The algorithm follows a greedy procedure by selecting the first vertex to be inserted at the end of

6

Working Paper

3

+8

6

+8

11 minus2

12

+4

14minus4

15

minus5

20 minus9

0

4 2

10

1

94

0

(a) Infeasible solution

3+8

6

+8

11 minus2

12

+3+7

14 minus1minus6

15

minus5

20

minus9

0

31

9

0

8

3

2

9

3

(b) Feasible solution after additional visits to unbalanced stations 12 and 14

Figure 3 Handling an infeasible solution by considering additional visits to unbalanced stations

the route (before the depot) whose demand is completely met by a single visit without violating the

load limits ([0 Q]) Next the vehicle load is updated and the station that was inserted into the partial

solution is removed from OV (lines 8-12)

However it may come to a point where no station can be fully served in a single visit either because

the vehicle has not enough bikes to deliver or the residual capacity is not sufficient to collect the required

bikes at once Hence a split becomes necessary The second part of the constructive procedure (lines

13-17) iterates over OV searching for a station whose demand maximizes the number of bikes that can be

delivered or collected Ties are broken according to the nearest insertion criterion The station demand

and vehicle load are updated after the insertion Next the algorithm restarts from line 5 and the entire

insertion procedure is repeated until OV becomes empty Note that the generated initial solution is

always feasible

34 Local search

Initial and perturbed solutions are possibly improved by means of an RVND based procedure during the

local search RVND consists of systematically examining different types of neighborhoods in a random

ordering In particular if the best neighbor consists of an improving move then the search may continue

from any of the existing neighborhoods (including the last one to be explored) at random Otherwise a

different neighborhood other than those that did not succeed in finding an improved move is randomly

selected The procedure ends when all neighborhoods fail to refine the current solution Only feasible

moves are accepted

The following six neighborhood structures were implemented

7

Working Paper

Algorithm 1 Initial Solution Constructive Procedure

1 Procedure GenerateInitialSolution2 Qprime larr Q3 Solutionlarr empty4 OV larr List sorted in random order with all stations where di 6= 0 + random ones with di = 05 repeat6 inserted larr false7 for all i isin OV do8 if di le Qprime or QminusQprime ge di then9 Solutionlarr Solution cup i

10 Update vehicle capacity and remove i from OV11 inserted larr true12 break13 if not inserted then14 for all j isin OV do15 compute exchangej16 ilarr maxexchangej | j isin exchange17 Solutionlarr Solution cup i18 update OV19 update Qprime

20 until OV 6= empty21 return Solution22 end GenerateInitialSolution

bull Reinsertion mdash N (1) A station is removed and then reinserted in another position of the sequence

bull Or-opt2 mdash N (2) Two consecutive stations are removed and then inserted in another position

bull Or-opt3 mdash N (3) Three consecutive stations are removed and then inserted in another position

bull 2-opt mdash N (4) Two non-adjacent arcs are removed from the sequence and then two new ones are

inserted In other words a subsequence of the tour is reversed

bull Swap mdash N (5) Permutation of two stations

bull Suppression mdash N (6) Given a sequence L = i0 i1 ik a suppression list is composed of visits to

stations ij forallj isin 1 kminus1 such that pprimeij = pij (zero demand) or pprimeij 6= pij and ij is visited more

than once in the tour The best move if any consists in selecting one station to be removed from

L so that the solution cost is minimized and the resulting new sequence Lprime is feasible For example

Figure 4b shows the removal of an additional visit to station 2 thus modifying the subsequence

2 6 2 9 0 to 2 6 9 0 This neighborhood was originally proposed by Chemla et al (2013a) but

the authors considered all stations

The first five are well-known TSP neighborhood structures while the last is a problem-specific neigh-

borhood Figure 4a depicts an initial solution and Figures 4b to 4g illustrate modified solutions that

were obtained after changing the previous one by means of one of the neighborhoods described above

For example Figure 4d shows a solution in which a 2-opt move was applied over the solution shown in

Figure 4c For ease of presentation values of pickupdelivery operations as well as the vehicle load are

omitted

35 Perturbation mechanisms

One of the four mechanisms described below is selected at random whenever the algorithm enters the

perturbation phase

8

Working Paper

01

2

3

4

5

6

7

89

(a) Initial solution

01

2

3

4

5

6

7

89

(b) Suppression of second visitto station 2 after visiting 6

01

2

3

4

5

6

7

89

(c) Or-Opt3 of three consecutivestations 8 7 and 1

01

2

3

4

5

6

7

89

(d) 2-opt on subsequence5 8 7 1 3

01

2

3

4

5

6

7

89

(e) Swap between station 1 andstation 3

01

2

3

4

5

6

7

89

(f) Or-Opt2 of two consecutivestations 1 and 3

01

2

3

4

5

6

7

89

(g) Reinsertion of station 9 beforestation 7

Figure 4 Example regarding the application of neighborhood structures

bull AddBuffer mdash P (1) An additional visit to a station is included expecting to act as buffer using the

cheapest insertion criterion Unrouted stations are inserted twice using the same criterion (Chemla

et al 2013a)

bull AddStations mdash P (2) This perturbation mechanism generalizes the previous one in the sense of

allowing multiple visits to be added in the solution but with a different insertion criterion More

precisely an additional visit (or two in the case of unrouted stations) to up to three random stations

are included towards the end of the route Here we only consider stations that are visited at most

once Adjacent visits to the same station are forbidden

bull Double-Bridge mdash P (3) Introduced by Martin et al (1991) for the TSP this perturbation consists

9

Working Paper

of a permutation of two subsequences As a result four arcs are removed and four new ones are

added so as to generate a new sequence

bull Suppression mdash P (4) A suppression move (see Section 34) is applied at random but in this case

the resulting modified sequence is allowed to be infeasible

Figure 5 shows an example of perturbations applied over a (supposedly) local optimal solution In

Figure 5d a Double-Bridge move is applied by interchanging subsequence 6 4 1 with subsequence 7 9

Figure 5b shows the AddBuffer perturbation when an additional visit to station 7 is performed expecting

it to act as a buffer In Figure 5c the perturbation AddStations is applied by adding two random visits

one to station 8 and another one to station 6

01

2

3

4

5

6

7

89

(a) Solution before perturbation

01

2

3

4

5

6

7

89

(b) AddBuffer

01

2

3

4

5

6

7

89

(c) AddStations

01

2

3

4

5

6

7

89

(d) Double-Bridge

Figure 5 Example regarding the application of perturbation mechanisms

4 Computational experiments

The ILSSBRP algorithm was coded in C++ (g++ 464) and the computational tests were carried on an

Intel RcopyCoreTM i7-3770 with 340 GHz and 16 GB of RAM running Ubuntu 1404 Only a a single thread

was used during the experiments

41 Instances

The benchmark instances used to test the proposed algorithm are those suggested by Hernandez-Perez

and Salazar-Gonzalez (2004a) which were originally created for the one-commodity pickup and delivery

traveling salesman problem The benchmark contains instances ranging from 20 to 500 customers (sta-

tions) and vehicle capacities ranging from 10 to 1000 For each pair of problem size and vehicle capacity

10

Working Paper

there are 10 instances named from A to J and for each vertex i there is a demand di isin [minus10 10] Chemla

et al (2013a) and Erdogan et al (2015) only reported results for a subset of instances of the referred

benchmark Therefore in order to compare our results with theirs we tested ILSSBRP for all instances

considered in at least one of the two works (see Section 44) Furthermore to compute the initial and

final targets as well as the load capacity for each station the same procedures adopted by such authors

were employed for each vertex i pi = α times 10 pprimei = α times (10 + di) qi = α times 20 where α is an input

parameter and experiments were conducted with α = 1 and α = 3 In order to properly compare our

results with those in Chemla et al (2013a) and Erdogan et al (2015) we adopted their same convention

of rounding down the values of the cost matrix to the nearest integer (floor) although we noticed that

this can cause slight violations of the triangle inequality

42 Impact of the perturbation mechanisms

In this section we are interested in evaluating the impact of the perturbation mechanisms described in

Section 35 that is AddBuffer (P (1)) AddStations (P (2)) Double-Bridge (P (3)) and Suppression (P (4))

In view of this we selected a subset of 30 challenging instances for performing the experiments These

instances were chosen according to the largest gap values with respect to the lower bounds reported in

Erdogan et al (2015) We ran ILSSBRP 10 times for each of the 30 instances considering all possible

combinations of perturbations For this testing we arbitrarily adopted IR = 10 and IILS = n

Table 1 shows the impact of each combination over the average gaps (computed as (UB minus LB)LB)

and CPU times required by ILSSBRP to run to completion From the results obtained we believe that

P (2) + P (3) + P (4) seems to offer a good compromise between solution quality (average gap lt 1) and

CPU time Hence we herein decided to adopt this configuration for the perturbation mechanisms

Table 1 Impact of different combinations of the perturbation mechanisms

Perturbations Avg Timeused gap () (s)

P (1) 243 24440P (2) 108 77349P (3) 104 49824P (4) 156 24988

P (1) + P (2) 127 50645P (1) + P (2) + P (3) 105 51441

P (1) + P (2) + P (3) + P (4) 111 43143P (1) + P (2) + P (4) 125 41730

P (1) + P (3) 121 37144P (1) + P (3) + P (4) 130 33170

P (1) + P (4) 178 24869P (2) + P (3) 091 63029

P (2) + P (3) + P (4) 099 50570P (2) + P (4) 108 50375P (3) + P (4) 129 34132

43 Parameter tuning

The main ILSSBRP parameters to be calibrated are the number of restarts (IR) and the maximum number

of consecutive ILS iterations without improvement over the current local optimal solution (IILS) Here

we set IR = 10 as in Silva et al (2015) where the authors put forward a multi-start ILS that was capable

of obtaining state-of-the-art results for the split-delivery VRP

11

Working Paper

In previous works such as those mentioned in Section 1 the value of IILS was tuned based on the

instance size In VRPs (eg Penna et al (2013) Silva et al (2015) Subramanian (2012) and Vidal et al

(2015)) this parameter is usually set as a function of the number of customers and vehicles In TSP-like

(or single machine scheduling) problems (eg Silva et al 2012 Subramanian and Battarra 2013 and

Subramanian et al 2014) IILS was set only as a function of the number of customers (or jobs) We

decided to use the same rationale as in Silva et al (2012) by setting IILS = maxImin β times n where

Imin and β are input parameters For the latter we set β = 4 as in Subramanian et al (2014) Note that

Imin is more important for small size instances and its role is to prevent low values for IILS which in

this case may lead to an insufficient number of ILS iterations required for obtaining high quality (or even

optimal) solutions We then tested several values for Imin more specifically 100 120 140 160 and 180

For each of them we ran ILSSBRP 10 times for all instances containing 20 and 30 stations The average

results obtained suggest that Imin = 160 seems to provide a good compromise between solution quality

and CPU time since the algorithm managed to find almost all best known solutions in a relatively short

amount of time when using this value Therefore we set IILS = max160 4times n

44 Comparison with the literature

ILSSBRP was executed 10 times for each instance with a time limit of 1 hour The results found by

our algorithm are compared with the upper bounds determined by two versions of the tabu search

heuristic of Chemla et al (2013a) The first one (TS1) starts from an initial solution generated by a

greedy procedure while the second version (TS2) receives the solution produced by their branch-and-cut

algorithm (over a relaxation of the problem) as initial solution Detailed results of TS1 and TS2 are

available at httpcermicsenpcfr~meuniefrSVOCPDPhtml According to the authors it should

be noted that UB1 is the best solution found by TS1 and UB2 is the best solution found considering

both TS1 and TS2 A comparison is also performed with the lower and upper bounds obtained by the

exact Branch-and-cut algorithm of Erdogan et al (2015) Regarding the benchmark instances Chemla

et al (2013a) considered n isin 20 40 60 100 and Q isin 10 30 45 1000 whereas Erdogan et al (2015)

considered n isin 20 30 40 50 60 and Q isin 10 15 20 25 30 35 40 45 1000Chemla et al (2013a) ran their experiments on an AMD Athlon 5600+ 28 GHz with 16 GB of RAM

while Erdogan et al (2015) performed their testing on an Intel i7 360 GHz and 8 GB of RAM On the

one hand because the hardware performance of the first appears to be quite inferior to the second as

well as to our intel i7 340 GHz we decided to estimate an approximation factor based on the single

thread rating values reported in httpswwwcpubenchmarknetcomparephpcmp[]=86ampcmp[]=896

so as to better compare the runtime performance of the methods According to the referred website

the AMD Athlon 5600+ 28 GHz is roughly 243 times slower than our processor We thus report the

original CPU time values of Chemla et al (2013a) divided by a factor of 243 On the other hand since

the machine used in Erdogan et al (2015) is rather equivalent to ours perhaps even slightly faster we

decided to consider the original runtime values reported by the authors

441 Results for instances with up to 60 stations

The aggregate average results for instances containing 20 30 40 50 and 60 stations are reported in Tables

2 and 3 where Instance group denotes the set of 10 instances of a particular group (for example group

n20q10 contains 10 instances with n = 20 and Q = 10) UB1 Gap () UB2 Gap () and Gap

() correspond to the gap between UB1 UB2 and the upper bound found by Erdogan et al (2015)

respectively and the lower bound reported in Erdogan et al (2015) Time (s) UB1 Time (s) and

UB2 Time (s) indicate respectively the CPU time in seconds spent by Erdogan et al (2015) TS1 and

TS2 where the last two are scaled to our processor as mentioned above Avg Gap () and Best Gap

12

Working Paper

() are the gaps of the average solution and the best solution respectively found by ILSSBRP over the

10 runs with respect to the lower bounds in Erdogan et al (2015) Avg Time (s) is the average CPU

time in seconds spent by ILSSBRP to completion over the 10 runs Avg TTUB2 (s) denotes the average

time over the 10 runs to find or improve the best heuristic solution found in Chemla et al (2013a) (UB2)

and Avg NV is the average number of visits of the final solutions found by ILSSBRP Detailed results

are reported in Appendix B including the best solution found when considering all experiments (Best

of all exp)

Table 2 Aggregate average results per instance group for n isin 20 30 40 50 60 and α = 1

Erdogan Chemlaet al 2015 et al 2013b

ILSSBRP

Instance UB1 UB1 UB2 UB2 Avg Best Avg Avggroup

Gap TimeGap Time Gap Time Gap Gap Time TTUB2

Avg() (s)

() (s) () (s) () () (s) (s)NV

n20q10 000 035 257 275 000 353 006 000 584 038 2010n20q15 000 030 - - - - 000 000 213 - 1919n20q20 000 013 - - - - 000 000 094 - 1865n20q25 000 015 - - - - 000 000 063 - 1862n20q30 000 196 247 222 039 238 000 000 062 001 1866n20q35 000 112 - - - - 000 000 046 - 1870n20q40 000 122 - - - - 000 000 046 - 1865n20q45 000 113 152 271 001 283 000 000 045 001 1862n20q1000 000 083 243 292 000 304 000 000 045 001 1862

Avg 000 080 225 265 010 295 001 000 133 010 1887n30q10 000 622 - - - - 002 000 2808 - 3014n30q15 000 387 - - - - 012 000 1029 - 2849n30q20 000 16359 - - - - 002 002 506 - 2784n30q25 000 561 - - - - 000 000 236 - 2753n30q30 000 8220 - - - - 000 000 185 - 2753n30q35 000 29327 - - - - 002 000 189 - 2770n30q40 000 58462 - - - - 000 000 161 - 2762n30q45 000 22169 - - - - 001 000 147 - 2762n30q1000 000 19020 - - - - 000 000 142 - 2764

Avg 000 17236 - - - - 002 000 600 - 2801n40q10 000 12480 356 15192 009 175275 009 004 5689 1402 3976n40q15 000 2555 - - - - 001 000 2178 - 3753n40q20 000 1472 - - - - 001 000 933 - 3701n40q25 003 72388 - - - - 004 003 579 - 3683n40q30 000 3656 405 9272 000 9387 000 000 443 006 3679n40q35 000 3866 - - - - 000 000 336 - 3677n40q40 000 7065 - - - - 000 000 313 - 3679n40q45 000 7428 469 9363 061 9482 000 000 299 004 3683n40q1000 000 7017 507 11248 077 11454 000 000 295 003 3682

Avg 029 207552 876 41321 167 159699 019 017 8744 963 5597

From Table 2 it can be observed that the quality of the solutions found by ILSSBRP as well as those

obtained by the algorithm of Erdogan et al (2015) are visibly superior than the ones determined by the

13

Working Paper

tabu searches of Chemla et al (2013a) especially TS1 Such superiority becomes even more prominent

for α = 3 as presented in Table 3 Also the average CPU times spent by ILSSBRP to find or improve the

best solutions reported by Chemla et al (2013a) (UB2) are rather small in most cases sometimes only a

matter of relatively few seconds except for the instance group n40q10 when α = 3 where the proposed

algorithm required more CPU time

In addition assuming the same values for the demands the smaller the vehicle capacity the larger

the relative number of visits This increases the size of the tour thus affecting the number of operations

performed during the local search and possibly the number of ILS iterations as more moves are required

to be evaluated

Figures 6 and 7 show how the average gaps and CPU times of each method vary according to the value

of Q We omit the results of TS1 because the associated gaps are quite inferior when compared to those

obtained by the other algorithms While the average gaps of ILSSBRP tend to be larger for very small

values of Q the average CPU time decreases as the value of Q increases both for α = 1 and α = 3 A

similar behavior regarding the CPU time performance can be observed for the heuristic of Chemla et al

(2013a) as opposed to the algorithm of Erdogan et al (2015) which does not seem to have a consistent

pattern when considering this aspect

0

01

02

03

04

05

06

07

08

09

10 15 20 25 30 35 40 45 1000

Aver

age

gap

()

Q

Erdogan et alChemla et al (UB2)

ILSSBRP

(a) Average gap ()

0

500

1000

1500

2000

10 15 20 25 30 35 40 45 1000

Aver

age

CPU

tim

e (s

)

Q

ILSSBRP

(b) Average CPU time (s)

Figure 6 Average gap () and average CPU time (s) per Q and α = 1

0

05

1

15

2

10 15 20 25 30 35 40 45 1000

Aver

age

gap

()

Q

ILSSBRP

(a) Average gap ()

0

500

1000

1500

2000

2500

10 15 20 25 30 35 40 45 1000

Aver

age

CPU

tim

e (s

)

Q

ILSSBRP

Figures 8 and 9 illustrate the behavior of the average gaps and CPU times of the algorithms as

the number of stations increases Overall the quality of the solutions found by ILSSBRP and by the

algorithm of Erdogan et al (2015) are equivalent except for n = 60 and α = 3 where the former clearly

outperforms the latter Moreover there is a considerable increase on the CPU time for both methods

from the literature for n gt 40 in contrast to ILSSBRP whose increase appears to be more moderate

14

Working Paper

ILSSBRP

Avg() (s)

() (s) () (s) () () (s) (s)NV

Avg 000 042 288 1233 000 1560 003 001 3637 478 2388

n30q10 000 15385 - - - - 004 000 111536 - 5739n30q15 000 6536 - - - - 003 000 20632 - 4247n30q20 000 816 - - - - 003 000 13346 - 3938n30q25 000 1027 - - - - 010 003 6547 - 3397n30q30 000 991 - - - - 002 000 2878 - 3012n30q35 000 937 - - - - 022 000 2429 - 2983n30q40 000 592 - - - - 008 005 1185 - 2832n30q45 000 296 - - - - 014 000 996 - 2849n30q1000 000 22130 - - - - 002 000 139 - 2759

Avg 000 5412 - - - - 008 001 17743 - 3528

Avg 000 11613 478 17850 017 145586 005 001 40558 11991 4584

Avg 055 251170 - - - - 036 027 84933 - 5819

Avg 164 361681 2059 41062 375 334542 048 036 124842 1333 7070

However this was somewhat expected since the CPU effort of the algorithms of Erdogan et al (2015)

and of Chemla et al (2013a) tend to increase exponentially with increasing values of n

Finally Table 4 shows a summary of the best solutions found by the proposed algorithm where Opt

15

Working Paper

0

02

04

06

08

1

12

14

16

18

20 30 40 50 60

Aver

age

gap

()

n

ILSSBRP

(a) Average gap ()

0

500

1000

1500

2000

20 30 40 50 60

Aver

age

CPU

tim

e (s

)

n

ILSSBRP

Figure 8 Average gap () and average CPU time (s) per n and α = 1

0

05

1

15

2

25

3

35

20 30 40 50 60

Aver

age

gap

()

n

ILSSBRP

(a) Average gap ()

0

500

1000

1500

2000

2500

3000

3500

20 30 40 50 60

Aver

age

CPU

tim

e (s

)

n

ILSSBRP

denotes the number of optimal solutions Impr corresponds to the number of improved solutions

Equal indicates the number of equal solutions and Worse is the number of worse solutions The

results of Chemla et al (2013a) were not included in these tables because they are dominated by those

obtained by our heuristic for all instances For α = 1 ILSSBRP found 419 of the 424 known optimal

solutions and improved the result of 17 of the 26 instances that remain open As for α = 3 ILSSBRP

achieved 377 out of 399 proven optimal solutions and improved the best known solution of 44 out of 51

open instances

Table 4 Summary of the performance of the best solutions aggregated by n

Erdoganet al 2015

ILSSBRP

nOpt Opt Impr Equal Worse

α = 1

20 90 90 0 90 030 90 89 0 89 140 89 87 0 88 250 83 83 5 85 060 72 70 12 76 2

Total 424 419 17 428 5

α = 3

20 90 89 0 89 130 90 87 0 87 340 90 83 0 83 750 69 63 17 67 660 60 55 27 58 5

Total 399 377 44 384 22

16

Working Paper

442 Results for instances with 100 stations

Tables 5 and 6 illustrate the detailed results found by our algorithm and the tabu searches of Chemla

et al (2013a) for every instance containing 100 stations In this case the gaps are computed with respect

to the lower bound reported in Chemla et al (2013a) The results obtained show that ILSSBRP clearly

outperforms the methods from the literature both in terms of solution quality and CPU time In general

the proposed heuristic was capable of significantly improving the best known solution of all instances

In addition when considering α = 1 ILSSBRP required on average at most 8 seconds to find or

improve the best results of Chemla et al (2013a) In some cases such as those involving Q ge 30 our

algorithm spent on average only a fraction of a second to achieve a superior solution than the best one

from the literature For α = 3 more time was required on average to accomplish the same purpose but

mostly for Q = 10

5 Concluding Remarks

In this work we proposed a hybrid ILS algorithm that was especially designed to solve a challenging

single-vehicle SBRP variant Extensive computational experiments were conducted on 980 instances

from the literature ranging from 20 to 100 stations The results were compared with those reported in

Chemla et al (2013a) and Erdogan et al (2015) For the 900 instances containing up to 60 stations

the proposed heuristic called ILSSBRP was capable of finding 796 out of 823 known optimal solutions

(97) and improving the result of 61 out of 77 open instances (79) Our algorithm only failed to

be at least equal to the best known solution in 27 instances (3) In addition the average gap of the

average solutions found by ILSSBRP and the lower bound reported in Erdogan et al (2015) for each

instance group was always smaller than 07 thus ratifying the robustness of our heuristic As for the

80 instances involving 100 stations ILSSBRP outperformed the best heuristics available for the problem

by considerably improving the best known solution for all instances

Future work may include the development of an enhanced procedure to verify whether or not the

solution is feasible Currently this is the most time consuming part of the algorithm where we use a

relatively costly max-flow based procedure (Chemla et al 2013a) for performing this task Hence any

improvement on this procedure could possibly lead to an improvement on the CPU time Also other

type of hybridizations could be experimented by combining for example efficient exact algorithms with

the heuristic suggested in this work

Acknowledgments

This research was partially supported by the Conselho Nacional de Desenvolvimento Cientıfico e Tec-

nologico (CNPq) grant 3052232015-1 and by the Comissao de Aperfeicoamento de Pessoal de Nıvel

Superior (CAPES) grant PVE A007 2013

17

Working Paper

Table 5 Detailed results for n = 100 and α = 1

Chemla

et al 2013bILSSBRP

Instance UB1 UB1 UB2 UB2 Avg Best Avg Avg

UB1 Gap Time UB2 Gap TimeAvg

GapBest

Gap Time TTUB2Avg

() (s) () (s)Sol

()Sol

() (s) (s)NV

n100q10A 14921 4558 68115 13273 2950 480544 1128320 1009 11258 984 355839 323 9730

n100q10B 17658 5929 47081 14981 3514 458893 1266920 1429 12609 1375 360031 528 10220

n100q10C 17138 4475 61706 15636 3207 475860 1325100 1192 13224 1169 360047 472 10070

n100q10D 19278 5705 48929 16586 3512 462098 1383280 1269 13783 1229 360054 323 9840

n100q10E 16867 6973 36851 12513 2591 448581 1097490 1044 10954 1023 338598 610 10570

n100q10F 14759 4688 48642 12621 2560 460372 1122620 1172 11191 1137 360031 526 10130

n100q10G 16772 6348 38001 13820 3471 449896 1118660 904 11160 878 334766 240 10010

n100q10H 15941 4521 39316 14863 3539 451170 1233970 1241 12308 1212 360040 402 10670

n100q10I 17799 5046 48519 16602 4034 460372 1354070 1446 13469 1386 360081 415 10460

n100q10J 20044 7598 45971 14988 3159 457866 1249130 967 12462 941 360054 372 9690

Avg - 5584 48313 - 3254 460565 - 1167 - 1133 354954 421 10139

n100q30A 12175 6241 51846 8033 716 463494 782000 431 7820 431 22557 147 9100

n100q30B 11066 4870 53777 10223 3737 465754 809450 877 8094 876 57056 026 9350

n100q30C 12106 5014 43096 9149 1347 454908 850570 549 8503 546 48066 070 9180

n100q30D 11317 4552 58995 9690 2460 470643 833990 724 8336 719 60254 052 9110

n100q30E 10446 3542 45191 8479 992 456921 799290 362 7986 353 40374 103 9590

n100q30F 11960 6114 37385 8281 1157 449773 802860 817 8020 805 31618 124 9340

n100q30G 11290 4631 49094 8872 1497 460988 807500 464 8075 464 24669 024 9000

n100q30H 11144 4219 38782 8944 1412 450389 825700 535 8257 535 63006 052 9700

n100q30I 12112 4532 45068 9189 1025 456880 867460 408 8652 381 54799 161 9650

n100q30J 10636 4470 56037 9014 2263 467890 792300 779 7923 779 47518 036 9040

Avg - 4819 47927 - 1661 459764 - 595 - 589 44992 080 9306

n100q45A 8494 1812 75756 8103 1269 487610 763200 614 7632 614 17988 012 9100

n100q45B 8838 2733 39193 8020 1554 450882 766000 1036 7660 1036 26068 036 9250

n100q45C 10056 3021 43507 8270 709 455196 799300 350 7993 350 18928 097 9040

n100q45D 9442 2715 61419 8535 1494 472985 791470 658 7900 638 31862 027 8820

n100q45E 10258 3947 61377 7864 692 472985 783500 653 7835 653 17133 782 9410

n100q45F 10348 3736 48560 7817 376 460290 773100 262 7731 262 17523 107 9330

n100q45G 9856 3159 66800 8286 1063 479188 786400 499 7864 499 13925 010 9000

n100q45H 9506 2594 48149 7796 328 459756 774000 254 7740 254 26665 506 9400

n100q45I 10334 3223 42685 8667 1090 454333 804220 290 8037 284 28897 021 9610

n100q45J 9021 2753 53695 7860 1112 465631 758850 728 7566 696 26060 041 8970

Avg - 2969 54114 - 969 465886 - 534 - 529 22505 164 9193

n100q1000A 8447 1881 68238 8199 1532 480338 745300 483 7453 483 13276 010 9200

n100q1000B 8669 2276 54352 8183 1588 466411 749100 608 7491 608 17213 009 9330

n100q1000C 8692 1581 64705 8673 1556 476682 789850 524 7895 519 11831 009 9010

n100q1000D 10116 4108 48971 8363 1663 460578 757280 561 7565 550 16368 011 8880

n100q1000E 8922 1762 35578 8071 640 447226 777100 244 7771 244 12579 009 9430

n100q1000F 10348 4247 48354 8053 1087 460249 764800 530 7648 530 12992 017 9310

n100q1000G 10954 4512 43219 8516 1282 454949 781750 357 7813 351 9681 006 9010

n100q1000H 9275 2286 39768 7786 314 337494 759300 058 7593 058 18562 036 9400

n100q1000I 10145 3158 39932 8461 974 451745 797500 344 7975 344 11825 013 9600

n100q1000J 9178 3027 44657 8655 2284 456593 731500 382 7315 382 11766 006 8940

Avg - 2884 48777 - 1292 449227 - 409 - 407 13609 013 9211

18

Working Paper

Chemla

et al 2013bILSSBRP

GapBest

Gap Time TTUB2Avg

() (s) () (s)Sol

()Sol

() (s) (s)NV

n100q10A 36057 6084 513328 28277 2614 930111 2412160 760 24014 712 361042 2211 9730

n100q10B 47107 7199 720056 35199 2851 1143453 2970960 847 29438 748 363188 4910 10220

n100q10C 50606 7267 48601 35779 2208 468260 3180250 851 31540 762 362325 6737 10070

n100q10D 47489 5118 776873 37972 2088 1201051 3384670 775 33654 714 364122 9256 9840

n100q10E 41002 7589 77605 30222 2965 493855 2509260 764 24917 689 361353 2373 10570

n100q10F 43544 8670 45930 28488 2214 466042 2530710 851 25176 794 361996 5033 10130

n100q10G 38539 6926 45437 28822 2658 462919 2487720 926 24642 823 361624 3489 10010

n100q10H 44411 6769 58789 33853 2783 481694 2903970 965 28794 872 361554 6392 10670

n100q10I 48727 6576 518792 37199 2654 942395 3238000 1015 32023 894 364319 6252 10460

n100q10J 45590 6676 35578 34086 2468 454333 2937350 744 29147 661 363050 7132 9690

Avg - 6887 284099 - 2550 704411 - 850 - 767 362457 5379 10139

n100q30A 16110 5573 44164 13366 2920 456058 1127800 902 11258 883 355025 240 9100

n100q30B 18739 6861 34674 15537 3980 446404 1265070 1383 12605 1341 360072 428 9350

n100q30C 18871 6146 43753 16116 3789 457784 1323150 1321 13224 1314 360042 355 9180

n100q30D 18262 4967 72223 16419 3456 486501 1378600 1298 13783 1296 360072 361 9110

n100q30E 16867 6983 36810 13118 3208 448623 1098410 1059 10954 1029 319465 351 9590

n100q30F 14838 4649 45355 13674 3500 456962 1124580 1102 11191 1048 360031 221 9340

n100q30G 16772 6937 38166 13262 3392 450512 1118330 1293 11160 1270 340420 392 9000

n100q30H 15609 4053 54352 14495 3050 466000 1233280 1103 12296 1070 360052 403 9700

n100q30I 19159 5785 26663 16620 3693 441022 1351920 1138 13469 1097 360051 348 9650

n100q30J 20044 7822 45971 15423 3713 457866 1251300 1126 12462 1081 360043 317 9040

Avg - 5978 44213 - 3470 456773 - 1173 - 1143 353527 342 9306

n100q45A 11372 3643 71771 10694 2830 483378 922940 1073 9192 1028 159741 178 9100

n100q45B 16114 7468 40959 14520 5740 452608 1023370 1093 10209 1067 295564 099 9250

n100q45C 14817 5009 51600 13243 3415 463330 1087150 1012 10815 955 309303 155 9040

n100q45D 15845 6258 42479 15845 6258 454087 1114390 1434 11103 1392 356247 103 8820

n100q45E 11628 3252 60268 11400 2992 471834 952170 851 9498 824 168703 160 9410

n100q45F 12821 5090 44616 12243 4410 474217 943770 1108 9398 1062 166330 109 9330

n100q45G 14829 7160 43630 10827 2529 455360 944500 930 9445 930 121565 209 9000

n100q45H 13072 4235 65280 12319 3415 476887 1022660 1137 10206 1114 239188 209 9400

n100q45I 14366 4689 38700 14366 4689 450307 1086410 1108 10841 1085 306606 168 9610

n100q45J 13989 5375 49464 11850 3024 461112 1013810 1143 10131 1135 246949 209 8970

Avg - 5218 50877 - 3930 464312 - 1089 - 1059 237020 160 9193

n100q1000A 9402 2943 46957 8017 1036 459016 745720 266 7453 260 11751 008 9200

n100q1000B 8793 2568 50696 7595 855 462714 749100 707 7491 707 18214 082 9330

n100q1000C 9312 2143 42274 8554 1155 454169 790430 308 7895 296 11819 007 9010

n100q1000D 9832 3373 53161 7595 331 464973 757400 302 7565 290 15408 2625 8880

n100q1000E 8922 1587 35578 8071 482 70785 777100 092 7771 092 11950 030 9430

n100q1000F 9371 2857 72141 8783 2050 483830 764800 493 7648 493 12409 006 9310

n100q1000G 10954 4616 43178 8219 967 454908 781650 430 7813 425 10384 018 9010

n100q1000H 8829 2019 54928 8488 1555 466617 759300 337 7593 337 17474 008 9400

n100q1000I 10664 4010 30607 8149 706 442173 797500 478 7975 478 12487 038 9600

n100q1000J 8311 2004 53449 7976 1520 465056 731500 565 7315 565 11168 012 8940

Avg - 2812 48297 - 1066 422424 - 398 - 394 13306 283 9211

19

Working Paper

References

Alvarez-Valdes R Belenguer JM Benavent E Bermudez JD Munoz F Vercher E and Verdejo

F (2016) Optimizing the level of service quality of a bike-sharing system Omega v 62 p 163 ndash 175

Benchimol M Benchimol P Chappert B de la Taille A Laroche F Meunier F and Robinet L

(2011) Balancing the stations of a self service ldquobike hirerdquo system RAIRO Operations Research v 45

n 1 p 33ndash61

Blum C and Roli A (2003) Metaheuristics in combinatorial optimization Overview and conceptual

comparison ACM Computing Systems v 35 p 268ndash308

Caggiani L and Ottomanelli M (2013) A dynamic simulation based model for optimal fleet reposition-

ing in bike-sharing systems Procedia-Social and Behavioral Sciences v 87 p 203ndash210

Chemla D Meunier F and Wolfer Calvo R (2013a) Bike sharing systems Solving the static rebal-

ancing problem Discrete Optimization v 10 n 2 p 120ndash146

Chemla D Meunier F Pradeau T Wolfler Calvo R and Yahiaoui H Self-service bike sharing sys-

tems simulation repositioning pricing URL httpshalarchives-ouvertesfrhal-00824078

Working paper 2013b

Contardo C Morency C and Rousseau LM Balancing a dynamic public bike-sharing system Tech-

nical report CIRRELT-2012-09 CIRRELT Montreal Canada 2012

DellrsquoAmico M Hadjicostantinou E Iori M and Novellani S (2014) The bike sharing rebalancing

problem Mathematical formulations and benchmark instances Omega v 45 p 7ndash19

DellrsquoAmico M Iori M Novellani S and Stutzle T (2016) A Destroy and Repair Algorithm for the

Bike sharing Rebalancing Problem Computers amp Operations Research v 71 p 149 ndash 162

DeMaio P (2009) Bike-sharing History impacts models of provision and future Journal of Public

Transportation v 12 n 4 p 41ndash56

Di Gaspero L Rendl A and Urli T Constraint-based approaches for balancing bike sharing systems

Principles and Practice of Constraint Programming volume 8124 of Lecture Notes in Computer Science

p 758ndash773 Springer 2013a

Di Gaspero L Rendl A and Urli T A hybrid ACO+CP for balancing bicycle sharing systems Hybrid

Metaheuristics volume 7919 of Lecture Notes in Computer Science p 198ndash212 Springer 2013b

Di Gaspero L Rendl A and Urli T (2015) Balancing bike sharing systems with constraint program-

ming Constraints v 21 p 1ndash31

Erdogan G Battarra M and Wolfler Calvo R (2015) An exact algorithm for the static rebalancing

problem arising in bicycle sharing systems European Journal of Operational Research v 245 n 3 p

667ndash679

Erdogan G Laporte G and Wolfler Calvo R (2014) The static bicycle relocation problem with

demand intervals European Journal of Operational Research v 238 n 2 p 451ndash457

Forma IA Raviv T and Tzur M (2015) A 3-step math heuristic for the static repositioning problem

in bike-sharing systems Transportation research part B methodological v 71 p 230ndash247

20

Working Paper

Hernandez-Perez H and Salazar-Gonzalez JJ (2004a) A branch-and-cut algorithm for a traveling

salesman problem with pickup and delivery Discrete Applied Mathematics v 145 p 126ndash139

Hernandez-Perez H and Salazar-Gonzalez JJ (2004b) Heuristics for the one-commodity pickup-and-

delivery traveling salesman problem Transportation Science v 38 p 245ndash255

Ho SC and Szeto W (2014) Solving a static repositioning problem in bike-sharing systems using

iterated tabu search Transportation Research Part E Logistics and Transportation Review v 69 p

180ndash198

Kloimullner C Papazek P Hu B and Raidl GR Balancing bicycle sharing systems An approach for

the dynamic case Evolutionary Computation in Combinatorial Optimisation volume 7832 of Lecture

Notes in Computer Science p 73ndash84 Springer 2014

Kloimullner C Papazek P Hu B and Raidl GR A cluster-first route-second approach for balancing

bicycle sharing systems Computer Aided Systems Theory mdash EUROCAST 2015 volume 9520 of Lecture

Martin O Otto SW and Felten EW (1991) Large-step markov chains for the traveling salesman

problem Complex Systems v 5 p 299ndash326

Mladenovic N and Hansen P (1997) Variable neighborhood search Computers amp Operations Research

v 24 n 11 p 1097ndash1100

Pal A and Zhang Y Free-floating bike sharing Solving real-life large-scale static rebalancing problems

Technical report University of South Florida 2015

Papazek P Kloimullner C Hu B and Raidl GR Balancing bicycle sharing systems an analysis

of path relinking and recombination within a grasp hybrid Parallel Problem Solving from Nature mdash

PPSN XIII volume 8672 of Lecture Notes in Computer Science p 792ndash801 Springer 2014

Papazek P Raidl GR Rainer-Harbach M and Hu B A PILOT VND GRASP hybrid for the

static balancing of public bicycle sharing systems Computer Aided Systems Theory mdash EUROCAST

2013 volume 8111 of Lecture Notes in Computer Science p 372ndash379 Springer 2013

Penna PHV Subramanian A and Ochi LS (2013) An iterated local search heuristic for the hetero-

geneous fleet vehicle routing problem Journal of Heuristics v 19 p 201ndash232

Raidl GR Hu B Rainer-Harbach M and Papazek P Balancing bicycle sharing systems Improving

a vns by efficiently determining optimal loading operations Hybrid Metaheuristics volume 7919 of

Lecture Notes in Computer Science p 130ndash143 Springer 2013

Rainer-Harbach M Papazek P Hu B and Raidl G Balancing bicycle sharing systems A variable

neighborhood search approach Evolutionary Computation in Combinatorial Optimization volume

7832 of Lecture Notes in Computer Science p 121ndash132 Springer 2013

Rainer-Harbach M Papazek P Raidl GR Hu B and Kloimullner C (2014) PILOT GRASP and

VNS approaches for the static balancing of bicycle sharing systems Journal of Global Optimization

v 63 n 3 p 597ndash629

Raviv T Tzur M and Forma IA (2013) Static repositioning in a bike-sharing system models and

solution approaches EURO Journal on Transportation and Logistics v 2 n 3 p 187ndash229

Salazar-Gonzalez JJ and Santos-Hernandez B (2015) The split-demand one-commodity pickup-and-

delivery travelling salesman problem Transportation Research Part B Methodological v 75 p 58ndash73

21

Working Paper

Schuijbroek J Hampshire R and van Hoeve WJ Inventory rebalancing and vehicle routing in bike

sharing systems Technical report Carnegie Mellon University 2013

Silva MM Subramanian A and Ochi LS (2015) An iterated local search heuristic for the split

delivery vehicle routing problem Computers amp Operations Research v 53 p 234ndash249

Silva M Subramanian A Vidal T and Ochi LS (2012) A simple and effective metaheuristic for the

minimum latency problem European Journal of Operational Research v 221 n 3 p 513ndash520

Subramanian A and Battarra M (2013) An iterated local search algorithm for the travelling salesman

problem with pickups and deliveries Journal of the Operational Research Society v 64 n 3 p

402ndash409

Subramanian A Heuristic exact and hybrid approaches for vehicle routing problems PhD thesis

Universidade Federal Fluminense Niteroi Brazil 2012

Subramanian A Battarra M and Potts CN (2014) An iterated local search heuristic for the single

machine total weighted tardiness scheduling problem with sequence-dependent setup times Interna-

tional Journal of Production Research v 52 n 9 p 2729ndash2742

Subramanian A Drummond LMdA Bentes C Ochi LS and Farias R (2010) A parallel heuristic

for the vehicle routing problem with simultaneous pickup and delivery Computers amp Operations

Research v 37 n 11 p 1899ndash1911

Vidal T Battarra M Subramanian A and Erdogan G (2015) Hybrid metaheuristics for the clustered

vehicle routing problem Computers amp Operations Research v 58 p 87 ndash 99

22

Working Paper

Appendix A Checking feasibility

Let L = i1 i2 ik be a sequence of vertices where i1 = ik = 0 A directed graph can be built using

pi pprimei and qi for each i in the sequence as follows

bull Let s be the source of the flow network and for each vertex i representing the first occurrence of

each station in the sequence let us define a set of arcs ui with capacity pi

bull Let t be the sink of the flow network and for each iprime representing the last occurrence of each station

in the sequence let us define a set of arcs wiprime with capacity pprimei

bull For each j = 2 k minus 1 let us define an arc bjj+1 with capacity Q and

bull If a station i is visited more than once let us define an arc dee+1 with capacity qi between the eth

and (e+ 1)th visits to i

By computing an sndasht maximum flow one can find optimal bike displacements along the sequence L

For each station i let us define pi and pprimei respectively as the resulting sndasht flow on arcs ui and wi Flow

on arcs bjj+1 indicates the number of bikes from ij to ij+1 and flow on arcs dee+1 denotes the quantity

of bikes remaining in a station i after the eth visit and before the (e+ 1)th visit Figure 10 depicts a flow

network for a sequence L = 0 1 4 2 3 5 2 4 1 0 where s and t correspond to the depot

s

1

4

5

3

2 2rsquo

1rsquo

4rsquo

t

p1

p2

p3

p4

p5

Q

p1

q2

Q

pprime3

Q

q4

Q

pprime2Q

pprime4

pprime1

Q

pprime5

Figure 10 Flow network for feasibility checking

Chemla et al (2013a) states that sequence L induces a feasible solution when pi = pi for each station

i in the sequence Also if a vertex iprime is not in L then piprime = pprimeiprime = 0

Appendix B Detailed results

This appendix presents the detailed results found by ILSSBRB as well as those obtained by the exact

algorithm of Erdogan et al (2015) for the instances with up to 60 stations

23

Working Paper

Table 7 Detailed results for n isin 20 30 40 50 60 and α = 1

Erdoganet al 2015

ILSSBRP

Instance UB UB Avg Best Avg BestLB UB Gap Time

AvgGap Best Gap Time

Avgof all

() (s)Sol

() () (s)NV

expn20q10A 4702 4702 000 026 470200 000 4702 000 293 1800 4702n20q10B 4769 4769 000 011 476900 000 4769 000 404 1900 4769n20q10C 6012 6012 000 068 601200 000 6012 000 1864 2400 6012n20q10D 5989 5989 000 020 598900 000 5989 000 777 2000 5989n20q10E 6245 6245 000 131 624500 000 6245 000 881 2000 6245n20q10F 4717 4717 000 025 471700 000 4717 000 349 2200 4717n20q10G 5070 5070 000 007 507000 000 5070 000 274 1910 5070n20q10H 5542 5542 000 030 557520 060 5542 000 277 1990 5542n20q10I 4576 4576 000 014 457600 000 4576 000 507 2100 4576n20q10J 4078 4078 000 016 407800 000 4078 000 212 1800 4078n20q15A 3948 3948 000 021 394800 000 3948 000 097 1700 3948n20q15B 4174 4174 000 005 417400 000 4174 000 131 1900 4174n20q15C 5137 5137 000 120 513700 000 5137 000 412 2100 5137n20q15D 5446 5446 000 044 544600 000 5446 000 531 2000 5446n20q15E 5560 5560 000 060 556080 001 5560 000 373 1890 5560n20q15F 4290 4290 000 004 429000 000 4290 000 160 2100 4290n20q15G 4489 4489 000 008 448900 000 4489 000 103 1900 4489n20q15H 4559 4559 000 011 455900 000 4559 000 127 1900 4559n20q15I 4051 4051 000 008 405100 000 4051 000 088 1900 4051n20q15J 3905 3905 000 022 390500 000 3905 000 113 1800 3905n20q20A 3585 3585 000 006 358500 000 3585 000 053 1700 3585n20q20B 4073 4073 000 003 407300 000 4073 000 110 1800 4073n20q20C 4486 4486 000 015 448600 000 4486 000 220 2000 4486n20q20D 4674 4674 000 035 467400 000 4674 000 173 1900 4674n20q20E 4599 4599 000 029 459900 000 4599 000 104 1800 4599n20q20F 4108 4108 000 009 410800 000 4108 000 044 2000 4108n20q20G 4351 4351 000 010 435100 000 4351 000 049 1900 4351n20q20H 4143 4143 000 006 414300 000 4143 000 060 1900 4143n20q20I 4051 4051 000 011 405100 000 4051 000 069 1900 4051n20q20J 3674 3674 000 008 367400 000 3674 000 057 1750 3674n20q25A 3583 3583 000 006 358300 000 3583 000 040 1700 3583n20q25B 3792 3792 000 004 379200 000 3792 000 041 1800 3792n20q25C 4189 4189 000 007 418900 000 4189 000 143 2000 4189n20q25D 4226 4226 000 041 422600 000 4226 000 085 1900 4226n20q25E 4556 4556 000 032 455600 000 4556 000 083 1800 4556n20q25F 4108 4108 000 011 410800 000 4108 000 040 2000 4108n20q25G 4216 4216 000 024 421600 000 4216 000 045 1900 4216n20q25H 3992 3992 000 007 399200 000 3992 000 048 1900 3992n20q25I 3960 3960 000 008 396000 000 3960 000 060 1900 3960n20q25J 3650 3650 000 010 365000 000 3650 000 045 1720 3650n20q30A 3583 3583 000 003 358300 000 3583 000 036 1700 3583n20q30B 3792 3792 000 003 379200 000 3792 000 037 1800 3792n20q30C 4184 4184 000 1648 418400 000 4184 000 119 2000 4184n20q30D 4089 4089 000 014 408900 000 4089 000 120 1900 4089n20q30E 4388 4388 000 228 438800 000 4388 000 090 1800 4388n20q30F 4108 4108 000 018 410800 000 4108 000 037 2000 4108n20q30G 4216 4216 000 012 421600 000 4216 000 039 1900 4216n20q30H 3992 3992 000 013 399200 000 3992 000 042 1900 3992n20q30I 3960 3960 000 007 396000 000 3960 000 055 1900 3960n20q30J 3650 3650 000 012 365000 000 3650 000 042 1760 3650n20q35A 3583 3583 000 003 358300 000 3583 000 037 1730 3583n20q35B 3792 3792 000 004 379200 000 3792 000 035 1800 3792n20q35C 4045 4045 000 782 404500 000 4045 000 071 2000 4045n20q35D 3806 3806 000 183 380600 000 3806 000 048 1900 3806n20q35E 4277 4277 000 087 427700 000 4277 000 056 1800 4277n20q35F 4108 4108 000 018 410800 000 4108 000 037 2000 4108n20q35G 4216 4216 000 014 421600 000 4216 000 037 1900 4216n20q35H 3992 3992 000 014 399200 000 3992 000 042 1900 3992n20q35I 3960 3960 000 007 396000 000 3960 000 052 1900 3960n20q35J 3650 3650 000 010 365000 000 3650 000 040 1770 3650n20q40A 3583 3583 000 003 358300 000 3583 000 036 1700 3583n20q40B 3792 3792 000 003 379200 000 3792 000 036 1800 3792n20q40C 4042 4042 000 882 404200 000 4042 000 074 2020 4042n20q40D 3806 3806 000 191 380600 000 3806 000 045 1900 3806n20q40E 4277 4277 000 085 427700 000 4277 000 054 1800 4277n20q40F 4108 4108 000 011 410800 000 4108 000 038 2000 4108

Continued on next page

24

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn20q40G 4216 4216 000 013 421600 000 4216 000 037 1900 4216n20q40H 3992 3992 000 013 399200 000 3992 000 041 1900 3992n20q40I 3960 3960 000 007 396000 000 3960 000 053 1900 3960n20q40J 3650 3650 000 012 365000 000 3650 000 041 1730 3650n20q45A 3583 3583 000 002 358300 000 3583 000 037 1700 3583n20q45B 3792 3792 000 003 379200 000 3792 000 036 1800 3792n20q45C 4042 4042 000 788 404200 000 4042 000 069 2000 4042n20q45D 3806 3806 000 193 380600 000 3806 000 044 1900 3806n20q45E 4277 4277 000 086 427700 000 4277 000 054 1810 4277n20q45F 4108 4108 000 011 410800 000 4108 000 037 2000 4108n20q45G 4216 4216 000 014 421600 000 4216 000 038 1900 4216n20q45H 3992 3992 000 013 399200 000 3992 000 040 1900 3992n20q45I 3960 3960 000 007 396000 000 3960 000 055 1900 3960n20q45J 3650 3650 000 013 365000 000 3650 000 039 1710 3650n20q1000A 3583 3583 000 003 358300 000 3583 000 037 1700 3583n20q1000B 3792 3792 000 004 379200 000 3792 000 036 1800 3792n20q1000C 4042 4042 000 497 404200 000 4042 000 066 2000 4042n20q1000D 3806 3806 000 199 380600 000 3806 000 044 1900 3806n20q1000E 4277 4277 000 069 427700 000 4277 000 053 1800 4277n20q1000F 4108 4108 000 011 410800 000 4108 000 038 2000 4108n20q1000G 4216 4216 000 013 421600 000 4216 000 037 1900 4216n20q1000H 3992 3992 000 013 399200 000 3992 000 042 1900 3992n20q1000I 3960 3960 000 008 396000 000 3960 000 054 1900 3960n20q1000J 3650 3650 000 013 365000 000 3650 000 040 1720 3650n30q10A 6236 6236 000 703 623600 000 6236 000 1985 3000 6236n30q10B 6308 6308 000 256 630800 000 6308 000 1298 2600 6308n30q10C 6335 6335 000 1245 633920 007 6335 000 2986 3090 6335n30q10D 6076 6076 000 198 607920 005 6076 000 4372 3380 6076n30q10E 5877 5877 000 096 588120 007 5877 000 1578 2960 5877n30q10F 5695 5695 000 178 569500 000 5695 000 1551 3000 5695n30q10G 8891 8891 000 2634 889100 000 8891 000 7994 3100 8891n30q10H 5884 5884 000 229 588400 000 5884 000 2181 2910 5884n30q10I 5430 5430 000 471 543000 000 5430 000 2109 3100 5430n30q10J 5764 5764 000 210 576400 000 5764 000 2021 3000 5764n30q15A 5442 5442 000 303 544200 000 5442 000 1001 3000 5442n30q15B 5364 5364 000 218 536400 000 5364 000 686 2620 5364n30q15C 5170 5170 000 206 517000 000 5170 000 621 2820 5170n30q15D 5142 5142 000 227 514200 000 5142 000 1061 3000 5142n30q15E 5121 5121 000 168 512100 000 5121 000 686 2700 5121n30q15F 4895 4895 000 119 489500 000 4895 000 536 2900 4895n30q15G 7146 7146 000 1515 720600 084 7146 000 3121 2960 7146n30q15H 5103 5103 000 232 512090 035 5103 000 909 2900 5103n30q15I 4622 4622 000 298 462200 000 4622 000 428 2700 4622n30q15J 5242 5242 000 588 524390 004 5242 000 1243 2890 5242n30q20A 4894 4894 000 141 489400 000 4894 000 373 2900 4894n30q20B 4831 4831 000 274 483100 000 4831 000 371 2610 4831n30q20C 4821 4821 000 098 483100 021 4831 021 437 2800 4831n30q20D 4871 4871 000 202 487100 000 4871 000 641 2900 4871n30q20E 4547 4547 000 021 454700 000 4547 000 208 2700 4547n30q20F 4439 4439 000 152 443900 000 4439 000 199 2910 4439n30q20G 6399 6399 000 162581 639900 000 6399 000 1971 2900 6399n30q20H 4275 4275 000 041 427500 000 4275 000 346 2700 4275n30q20I 4325 4325 000 047 432500 000 4325 000 179 2710 4325n30q20J 4536 4536 000 033 453600 000 4536 000 335 2710 4536n30q25A 4825 4825 000 2649 482500 000 4825 000 240 2910 4825n30q25B 4530 4530 000 066 453000 000 4530 000 212 2610 4530n30q25C 4586 4586 000 1579 458600 000 4586 000 211 2800 4586n30q25D 4723 4723 000 271 472300 000 4723 000 286 2800 4723n30q25E 4466 4466 000 025 446600 000 4466 000 105 2700 4466n30q25F 4439 4439 000 151 443900 000 4439 000 149 2910 4439n30q25G 5731 5731 000 642 573100 000 5731 000 563 2700 5731n30q25H 4142 4142 000 115 414200 000 4142 000 243 2700 4142n30q25I 4272 4272 000 024 427200 000 4272 000 120 2700 4272n30q25J 4465 4465 000 083 446500 000 4465 000 235 2700 4465n30q30A 4603 4603 000 012 460300 000 4603 000 169 2900 4603n30q30B 4496 4496 000 75394 449600 000 4496 000 207 2620 4496n30q30C 4565 4565 000 5005 456500 000 4565 000 222 2800 4565n30q30D 4701 4701 000 299 470100 000 4701 000 226 2800 4701

25

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn30q30E 4466 4466 000 016 446600 000 4466 000 090 2700 4466n30q30F 4384 4384 000 195 438400 000 4384 000 154 2900 4384n30q30G 5402 5402 000 1020 540200 000 5402 000 378 2700 5402n30q30H 4030 4030 000 046 403000 000 4030 000 138 2700 4030n30q30I 4272 4272 000 182 427200 000 4272 000 109 2700 4272n30q30J 4300 4300 000 034 430000 000 4300 000 160 2710 4300n30q35A 4603 4603 000 031 460300 000 4603 000 150 2900 4603n30q35B 4434 4434 000 270658 443400 000 4434 000 252 2720 4434n30q35C 4525 4525 000 21196 452500 000 4525 000 219 2810 4525n30q35D 4666 4666 000 618 467500 019 4666 000 207 2870 4666n30q35E 4466 4466 000 018 446600 000 4466 000 087 2700 4466n30q35F 4384 4384 000 203 438400 000 4384 000 130 2900 4384n30q35G 5211 5211 000 242 521100 000 5211 000 464 2700 5211n30q35H 4030 4030 000 060 403000 000 4030 000 123 2700 4030n30q35I 4272 4272 000 192 427200 000 4272 000 104 2700 4272n30q35J 4300 4300 000 056 430000 000 4300 000 153 2700 4300n30q40A 4603 4603 000 074 460300 000 4603 000 145 2900 4603n30q40B 4340 4340 000 39541 434000 000 4340 000 128 2600 4340n30q40C 4475 4475 000 116415 447500 000 4475 000 223 2800 4475n30q40D 4666 4666 000 30923 466600 000 4666 000 219 2900 4666n30q40E 4466 4466 000 016 446600 000 4466 000 087 2700 4466n30q40F 4384 4384 000 202 438400 000 4384 000 126 2910 4384n30q40G 5031 5031 000 397065 503100 000 5031 000 328 2700 5031n30q40H 4030 4030 000 038 403000 000 4030 000 120 2700 4030n30q40I 4272 4272 000 099 427200 000 4272 000 102 2700 4272n30q40J 4300 4300 000 247 430000 000 4300 000 133 2710 4300n30q45A 4603 4603 000 132 460300 000 4603 000 142 2910 4603n30q45B 4340 4340 000 40972 434000 000 4340 000 128 2600 4340n30q45C 4475 4475 000 147963 447500 000 4475 000 211 2800 4475n30q45D 4666 4666 000 31656 467200 013 4666 000 204 2880 4666n30q45E 4466 4466 000 017 446600 000 4466 000 088 2700 4466n30q45F 4384 4384 000 208 438400 000 4384 000 127 2920 4384n30q45G 4827 4827 000 225 482700 000 4827 000 217 2700 4827n30q45H 4030 4030 000 025 403000 000 4030 000 118 2700 4030n30q45I 4272 4272 000 254 427200 000 4272 000 101 2710 4272n30q45J 4300 4300 000 241 430000 000 4300 000 135 2700 4300n30q1000A 4603 4603 000 132 460300 000 4603 000 142 2900 4603n30q1000B 4340 4340 000 45347 434000 000 4340 000 124 2600 4340n30q1000C 4475 4475 000 106421 447500 000 4475 000 202 2810 4475n30q1000D 4666 4666 000 32479 466600 000 4666 000 224 2900 4666n30q1000E 4466 4466 000 017 446600 000 4466 000 087 2700 4466n30q1000F 4384 4384 000 213 438400 000 4384 000 125 2910 4384n30q1000G 4781 4781 000 5113 478100 000 4781 000 157 2700 4781n30q1000H 4030 4030 000 066 403000 000 4030 000 117 2700 4030n30q1000I 4272 4272 000 229 427200 000 4272 000 103 2710 4272n30q1000J 4300 4300 000 179 430000 000 4300 000 137 2710 4300n40q10A 6949 6949 000 4196 696540 024 6949 000 6078 4090 6949n40q10B 5949 5949 000 816 594900 000 5949 000 2750 3900 5949n40q10C 7237 7237 000 1590 723700 000 7237 000 5939 3650 7237n40q10D 7692 7692 000 1323 769200 000 7692 000 11557 3940 7692n40q10E 6424 6424 000 2752 644100 026 6441 026 2483 3410 6441n40q10F 7074 7074 000 88639 709460 029 7085 016 7032 4080 7074n40q10G 7375 7375 000 16215 737550 001 7375 000 6076 4320 7375n40q10H 6556 6556 000 1337 656000 006 6556 000 5663 4240 6556n40q10I 6901 6901 000 1244 690200 001 6901 000 5459 4110 6901n40q10J 6267 6267 000 6684 627060 006 6267 000 3855 4020 6267n40q15A 6059 6059 000 4921 605900 000 6059 000 2228 3800 6059n40q15B 5319 5319 000 312 531900 000 5319 000 1035 3700 5319n40q15C 5912 5912 000 1459 591240 001 5912 000 2293 3570 5912n40q15D 6388 6388 000 886 638800 000 6388 000 3794 3770 6388n40q15E 5671 5671 000 2390 567100 000 5671 000 1181 3400 5671n40q15F 5847 5847 000 5785 584700 000 5847 000 2052 3820 5847n40q15G 6409 6409 000 2897 641410 008 6409 000 3149 4040 6409n40q15H 5575 5575 000 662 557500 000 5575 000 2046 3920 5575n40q15I 5732 5732 000 871 573200 000 5732 000 1675 3730 5732n40q15J 5630 5630 000 5371 563020 000 5630 000 2326 3780 5630n40q20A 5435 5435 000 1781 543500 000 5435 000 989 3800 5435n40q20B 5150 5150 000 391 515000 000 5150 000 440 3700 5150

26

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn40q20C 5331 5331 000 931 533620 010 5331 000 1902 3660 5331n40q20D 5850 5850 000 7970 585000 000 5850 000 1171 3700 5850n40q20E 5271 5271 000 553 527100 000 5271 000 497 3300 5271n40q20F 5289 5289 000 996 528900 000 5289 000 752 3700 5289n40q20G 5476 5476 000 497 547600 000 5476 000 1095 3920 5476n40q20H 5117 5117 000 349 511700 000 5117 000 782 3830 5117n40q20I 4984 4984 000 046 498400 000 4984 000 661 3700 4984n40q20J 4996 4996 000 1205 499600 000 4996 000 1038 3700 4996n40q25A 4949 4949 000 092 494900 000 4949 000 428 3800 4949n40q25B 5110 5110 000 540 511000 000 5110 000 309 3700 5110n40q25C 506967 5087 034 718269 509220 044 5087 034 1405 3670 5087n40q25D 5748 5748 000 3139 574800 000 5748 000 706 3600 5748n40q25E 5237 5237 000 637 523700 000 5237 000 262 3300 5237n40q25F 5048 5048 000 332 504800 000 5048 000 575 3720 5048n40q25G 5361 5361 000 231 536100 000 5361 000 574 3930 5361n40q25H 4886 4886 000 100 488600 000 4886 000 498 3800 4886n40q25I 4801 4801 000 041 480100 000 4801 000 500 3700 4801n40q25J 4811 4811 000 498 481100 000 4811 000 536 3610 4811n40q30A 4949 4949 000 123 494900 000 4949 000 306 3800 4949n40q30B 5110 5110 000 210 511000 000 5110 000 270 3700 5110n40q30C 4692 4692 000 33372 469200 000 4692 000 948 3660 4692n40q30D 5378 5378 000 577 537800 000 5378 000 867 3600 5378n40q30E 5154 5154 000 890 515400 000 5154 000 315 3320 5154n40q30F 4921 4921 000 066 492100 000 4921 000 299 3700 4921n40q30G 5312 5312 000 986 531200 000 5312 000 396 3910 5312n40q30H 4840 4840 000 140 484000 000 4840 000 403 3800 4840n40q30I 4713 4713 000 050 471300 000 4713 000 308 3700 4713n40q30J 4753 4753 000 141 475300 000 4753 000 322 3600 4753n40q35A 4949 4949 000 179 494900 000 4949 000 274 3800 4949n40q35B 5110 5110 000 266 511000 000 5110 000 252 3700 5110n40q35C 4570 4570 000 32678 457000 000 4570 000 650 3660 4570n40q35D 5238 5238 000 185 523800 000 5238 000 459 3600 5238n40q35E 5069 5069 000 124 506900 000 5069 000 185 3310 5069n40q35F 4921 4921 000 204 492100 000 4921 000 270 3700 4921n40q35G 5253 5253 000 304 525300 000 5253 000 385 3900 5253n40q35H 4840 4840 000 4244 484000 000 4840 000 344 3800 4840n40q35I 4713 4713 000 044 471300 000 4713 000 283 3700 4713n40q35J 4753 4753 000 432 475300 000 4753 000 252 3600 4753n40q40A 4949 4949 000 208 494900 000 4949 000 267 3800 4949n40q40B 5110 5110 000 654 511000 000 5110 000 259 3700 5110n40q40C 4570 4570 000 65051 457000 000 4570 000 591 3670 4570n40q40D 5202 5202 000 271 520200 000 5202 000 414 3600 5202n40q40E 5069 5069 000 122 506900 000 5069 000 179 3320 5069n40q40F 4816 4816 000 250 481600 000 4816 000 232 3700 4816n40q40G 5253 5253 000 564 525300 000 5253 000 358 3900 5253n40q40H 4804 4804 000 3059 480400 000 4804 000 312 3800 4804n40q40I 4713 4713 000 044 471300 000 4713 000 270 3700 4713n40q40J 4753 4753 000 431 475300 000 4753 000 246 3600 4753n40q45A 4949 4949 000 211 494900 000 4949 000 261 3800 4949n40q45B 5110 5110 000 640 511000 000 5110 000 251 3700 5110n40q45C 4570 4570 000 68360 457000 000 4570 000 569 3680 4570n40q45D 5202 5202 000 638 520200 000 5202 000 367 3600 5202n40q45E 5069 5069 000 335 506900 000 5069 000 178 3330 5069n40q45F 4816 4816 000 250 481600 000 4816 000 227 3700 4816n40q45G 5253 5253 000 566 525300 000 5253 000 349 3910 5253n40q45H 4804 4804 000 2801 480400 000 4804 000 287 3810 4804n40q45I 4713 4713 000 045 471300 000 4713 000 260 3700 4713n40q45J 4753 4753 000 434 475300 000 4753 000 241 3600 4753n40q1000A 4949 4949 000 171 494900 000 4949 000 262 3800 4949n40q1000B 5110 5110 000 653 511000 000 5110 000 256 3700 5110n40q1000C 4570 4570 000 65774 457000 000 4570 000 532 3670 4570n40q1000D 5202 5202 000 567 520200 000 5202 000 353 3600 5202n40q1000E 5069 5069 000 337 506900 000 5069 000 177 3320 5069n40q1000F 4816 4816 000 358 481600 000 4816 000 224 3700 4816n40q1000G 5253 5253 000 826 525300 000 5253 000 347 3920 5253n40q1000H 4804 4804 000 1279 480400 000 4804 000 289 3810 4804n40q1000I 4713 4713 000 049 471300 000 4713 000 266 3700 4713n40q1000J 4753 4753 000 152 475300 000 4753 000 241 3600 4753

27

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn50q10A 6520 6520 000 2099 652750 012 6520 000 5031 4350 6520n50q10B 9025 9025 000 25540 902660 002 9025 000 24697 5020 9025n50q10C 8151 8151 000 84926 815100 000 8151 000 27258 4850 8151n50q10D 9804 9804 000 68226 980670 003 9804 000 43355 5090 9804n50q10E 9086 9086 000 163688 909320 008 9086 000 25379 5020 9086n50q10F 7661 7661 000 19268 766470 005 7661 000 20099 5090 7661n50q10G 6986 6986 000 6450 698760 002 6986 000 6576 4940 6986n50q10H 836953 9031 790 718324 856320 231 8563 231 19083 4900 8563n50q10I 7964 7964 000 76112 799640 041 7964 000 23332 5300 7964n50q10J 8054 8054 000 33846 805400 000 8054 000 16171 5080 8054n50q15A 5975 5975 000 1925 597500 000 5975 000 1277 4120 5975n50q15B 7697 7697 000 304500 773750 053 7697 000 13153 4720 7697n50q15C 6693 6693 000 18362 669300 000 6693 000 13303 4800 6693n50q15D 804229 8256 266 718315 814530 128 8136 117 11309 4920 8136n50q15E 7649 7773 162 718320 773200 109 7732 109 12023 4710 7732n50q15F 6464 6464 000 7524 646400 000 6464 000 5551 4650 6464n50q15G 6273 6273 000 2375 627300 000 6273 000 3409 4810 6273n50q15H 7005 7005 000 167901 700500 000 7005 000 5344 4700 7005n50q15I 6423 6423 000 4673 642300 000 6423 000 4480 4700 6423n50q15J 6601 6601 000 26220 660100 000 6601 000 5219 4580 6601n50q20A 5728 5728 000 575 572800 000 5728 000 959 4100 5728n50q20B 6888 6888 000 19034 688800 000 6888 000 7069 4700 6888n50q20C 6165 6165 000 41062 618940 040 6165 000 6735 4760 6165n50q20D 7157 7157 000 44711 715700 000 7157 000 3844 4600 7157n50q20E 6905 6905 000 121713 690570 001 6905 000 4058 4600 6905n50q20F 5921 5921 000 12953 592100 000 5921 000 3665 4800 5921n50q20G 5905 5905 000 15799 590500 000 5905 000 2002 4800 5905n50q20H 6326 6326 000 27542 632600 000 6326 000 2909 4500 6326n50q20I 6023 6023 000 7653 602570 004 6023 000 2211 4730 6023n50q20J 6050 6050 000 4410 605000 000 6050 000 2398 4520 6050n50q25A 5728 5728 000 2351 572800 000 5728 000 735 4100 5728n50q25B 6687 6687 000 87146 668910 003 6687 000 4285 4700 6687n50q25C 5772 5772 000 6148 577200 000 5772 000 5478 4700 5772n50q25D 6817 6817 000 28986 681700 000 6817 000 3507 4600 6817n50q25E 6381 6381 000 8105 638100 000 6381 000 2760 4600 6381n50q25F 5673 5673 000 2976 567300 000 5673 000 2647 4600 5673n50q25G 5885 5885 000 108572 588500 000 5885 000 1597 4800 5885n50q25H 5942 5942 000 7160 594200 000 5942 000 1738 4510 5942n50q25I 5615 5615 000 1342 561500 000 5615 000 1576 4700 5615n50q25J 5833 5833 000 20035 583300 000 5833 000 1196 4400 5833n50q30A 5634 5634 000 1847 563400 000 5634 000 494 4100 5634n50q30B 6285 6285 000 5961 628500 000 6285 000 1940 4620 6285n50q30C 5610 5610 000 1350 561000 000 5610 000 2469 4600 5610n50q30D 6420 6420 000 3688 642180 003 6420 000 1691 4630 6420n50q30E 6219 6219 000 2678 621900 000 6219 000 1161 4500 6219n50q30F 5424 5424 000 297 542400 000 5424 000 1715 4600 5424n50q30G 5864 5864 000 127554 586400 000 5864 000 1045 4800 5864n50q30H 5804 5804 000 23377 580400 000 5804 000 1004 4500 5804n50q30I 5491 5491 000 1881 549100 000 5491 000 846 4700 5491n50q30J 5747 5747 000 8770 574700 000 5747 000 817 4420 5747n50q35A 5634 5634 000 2565 563400 000 5634 000 414 4120 5634n50q35B 6192 6192 000 9365 619590 006 6192 000 1777 4650 6192n50q35C 5503 5503 000 1205 550300 000 5503 000 1545 4600 5503n50q35D 6277 6277 000 2694 627700 000 6277 000 1224 4620 6277n50q35E 6168 6168 000 1305 616800 000 6168 000 937 4500 6168n50q35F 527638 5351 141 718269 535100 141 5351 141 1388 4700 5351n50q35G 585457 5913 100 718330 586400 016 5864 016 967 4800 5864n50q35H 5684 5684 000 1772 568400 000 5684 000 873 4500 5684n50q35I 5462 5462 000 1159 546200 000 5462 000 770 4700 5462n50q35J 5675 5675 000 4421 567500 000 5675 000 718 4440 5675n50q40A 5634 5634 000 3073 563400 000 5634 000 396 4120 5634n50q40B 6105 6105 000 7960 610500 000 6105 000 1886 4620 6105n50q40C 5398 5398 000 1684 539800 000 5398 000 1031 4600 5398n50q40D 6220 6220 000 1581 622000 000 6220 000 970 4620 6220n50q40E 6089 6089 000 2165 608900 000 6089 000 755 4500 6089n50q40F 525667 5262 010 718330 526200 010 5262 010 1002 4600 5262n50q40G 5864 5864 000 662636 586400 000 5864 000 842 4800 5864n50q40H 5559 5559 000 1495 555900 000 5559 000 641 4500 5559

28

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn50q40I 5462 5462 000 740 546200 000 5462 000 670 4700 5462n50q40J 5675 5675 000 9278 567500 000 5675 000 636 4420 5675n50q45A 5634 5634 000 2644 563400 000 5634 000 398 4100 5634n50q45B 6042 6042 000 28074 604200 000 6042 000 1257 4610 6042n50q45C 5289 5289 000 1043 528900 000 5289 000 1280 4600 5289n50q45D 6184 6184 000 1976 618400 000 6184 000 1100 4600 6184n50q45E 6080 6080 000 4246 608000 000 6080 000 789 4500 6080n50q45F 5262 5262 000 527394 526200 000 5262 000 867 4600 5262n50q45G 5864 5864 000 643822 586400 000 5864 000 771 4800 5864n50q45H 5559 5559 000 965 555900 000 5559 000 603 4500 5559n50q45I 5462 5462 000 1138 546200 000 5462 000 651 4700 5462n50q45J 5675 5675 000 10028 567500 000 5675 000 632 4400 5675n50q1000A 5634 5634 000 3010 563400 000 5634 000 411 4130 5634n50q1000B 5901 5967 112 718250 595500 092 5955 092 828 4600 5955n50q1000C 5139 5139 000 3240 513900 000 5139 000 744 4600 5139n50q1000D 5836 5836 000 095 583600 000 5836 000 692 4600 5836n50q1000E 6023 6023 000 212532 602300 000 6023 000 685 4500 6023n50q1000F 5262 5262 000 600096 526200 000 5262 000 731 4600 5262n50q1000G 5864 5864 000 262617 586400 000 5864 000 783 4800 5864n50q1000H 5559 5559 000 99630 555900 000 5559 000 593 4500 5559n50q1000I 5462 5462 000 1257 546200 000 5462 000 591 4700 5462n50q1000J 5675 5675 000 9028 567500 000 5675 000 641 4410 5675n60q10A 7810 7810 000 516079 784460 044 7843 042 31430 5900 7843n60q10B 8156 8156 000 273044 818480 035 8156 000 25498 6070 8156n60q10C 893140 9214 316 718258 909000 178 9090 178 58659 6140 9090n60q10D 10159 10159 000 49859 1015900 000 10159 000 65644 5530 10159n60q10E 885506 9339 547 718263 897400 134 8949 106 42348 5980 8949n60q10F 8163 8163 000 55276 817920 020 8163 000 42350 6480 8163n60q10G 858588 8691 122 718294 866660 094 8665 092 39232 6320 8665n60q10H 8114 8114 000 125622 812030 008 8114 000 22646 5670 8114n60q10I 907809 9313 259 718323 918850 122 9188 121 67018 6220 9188n60q10J 8050 8050 000 31600 807800 035 8078 035 25128 5900 8050n60q15A 6730 6730 000 116723 673000 000 6730 000 12085 5400 6730n60q15B 6990 6990 000 38869 699000 000 6990 000 8968 5500 6990n60q15C 751748 7720 269 718305 763000 150 7630 150 22101 6060 7630n60q15D 8280 8477 238 718233 840500 151 8380 121 23182 5520 8380n60q15E 7446 7446 000 24049 744600 000 7446 000 12235 5600 7446n60q15F 7075 7075 000 68804 707650 002 7075 000 13739 5800 7075n60q15G 7344 7344 000 183521 734400 000 7344 000 15718 5830 7344n60q15H 6743 6743 000 7802 674300 000 6743 000 7541 5430 6743n60q15I 7577 7577 000 54646 757700 000 7577 000 17632 5920 7577n60q15J 7043 7043 000 26550 704300 000 7043 000 7088 5420 7043n60q20A 6298 6298 000 30920 629800 000 6298 000 6895 5440 6298n60q20B 6363 6363 000 37983 636300 000 6363 000 3816 5510 6363n60q20C 6962 6962 000 196536 696200 000 6962 000 9242 5850 6962n60q20D 7570 7570 000 715315 757120 002 7570 000 11311 5190 7570n60q20E 6980 6980 000 70054 698000 000 6980 000 4878 5400 6980n60q20F 6285 6285 000 6098 628500 000 6285 000 6126 5700 6285n60q20G 6802 6802 000 92004 680250 001 6802 000 9492 5800 6802n60q20H 6248 6248 000 3604 624800 000 6248 000 5055 5500 6248n60q20I 6880 6880 000 108416 688070 001 6880 000 12576 5900 6880n60q20J 6727 6727 000 24095 672700 000 6727 000 3598 5360 6727n60q25A 6069 6069 000 18660 606900 000 6069 000 4073 5400 6069n60q25B 6195 6195 000 10388 619500 000 6195 000 2298 5500 6195n60q25C 6620 6620 000 37477 662000 000 6620 000 5053 5840 6620n60q25D 7051 7051 000 14417 705100 000 7051 000 6310 5100 7051n60q25E 6633 6633 000 5981 663300 000 6633 000 3679 5400 6633n60q25F 601977 6060 067 718247 606000 067 6060 067 5322 5620 6060n60q25G 6447 6447 000 14427 644700 000 6447 000 4923 5710 6447n60q25H 6030 6030 000 18069 603000 000 6030 000 3714 5420 6030n60q25I 6563 6563 000 98703 656300 000 6563 000 6055 5800 6563n60q25J 6496 6496 000 7055 649600 000 6496 000 2896 5340 6496n60q30A 5924 5924 000 9835 592400 000 5924 000 2716 5430 5924n60q30B 6195 6195 000 13421 619500 000 6195 000 1856 5500 6195n60q30C 6383 6383 000 5745 638580 004 6383 000 3713 5810 6383n60q30D 6596 6596 000 12698 659600 000 6596 000 3222 5100 6596n60q30E 6494 6494 000 7177 649400 000 6494 000 3039 5400 6494n60q30F 594867 6028 133 718220 598800 066 5988 066 2922 5600 5988

29

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn60q30G 6360 6360 000 8760 636000 000 6360 000 2580 5700 6360n60q30H 5994 5994 000 30020 599400 000 5994 000 2756 5550 5994n60q30I 6419 6419 000 443943 642700 012 6419 000 5129 5800 6419n60q30J 6389 6389 000 2828 638900 000 6389 000 1693 5330 6389n60q35A 5773 5773 000 700 577300 000 5773 000 1961 5420 5773n60q35B 6132 6132 000 9562 613200 000 6132 000 1784 5500 6132n60q35C 6170 6170 000 3707 617000 000 6170 000 2181 5730 6170n60q35D 6491 6491 000 11582 649100 000 6491 000 1952 5100 6491n60q35E 6346 6346 000 6229 634600 000 6346 000 2276 5400 6346n60q35F 592833 6004 128 718322 596000 053 5960 053 2836 5600 5960n60q35G 6308 6308 000 14447 630800 000 6308 000 2378 5700 6308n60q35H 5936 5936 000 179478 593790 003 5936 000 1886 5440 5936n60q35I 6242 6242 000 148801 624200 000 6242 000 4030 5810 6242n60q35J 6374 6374 000 4149 637400 000 6374 000 1234 5310 6374n60q40A 5773 5773 000 4825 577300 000 5773 000 1695 5400 5773n60q40B 6114 6137 038 718224 613200 029 6132 029 1504 5500 6132n60q40C 6135 6135 000 1967 613500 000 6135 000 2338 5840 6135n60q40D 6369 6369 000 3133 636900 000 6369 000 2267 5100 6369n60q40E 6159 6159 000 21401 615900 000 6159 000 1749 5400 6159n60q40F 587117 5925 092 718192 591520 075 5912 070 3315 5800 5912n60q40G 6273 6273 000 186208 627300 000 6273 000 2361 5700 6273n60q40H 588467 5936 087 718214 593790 090 5936 087 1676 5490 5936n60q40I 6072 6072 000 233579 607200 000 6072 000 3113 5800 6072n60q40J 6374 6374 000 2163 637400 000 6374 000 1078 5300 6374n60q45A 5773 5773 000 7334 577300 000 5773 000 1592 5470 5773n60q45B 6125 6132 011 718223 613340 014 6132 011 1465 5500 6132n60q45C 6135 6135 000 3472 613500 000 6135 000 2322 5800 6135n60q45D 6269 6269 000 10105 626900 000 6269 000 2294 5200 6269n60q45E 6131 6131 000 2150 613100 000 6131 000 1475 5400 6131n60q45F 581644 5888 123 718329 588600 120 5886 120 3351 5800 5886n60q45G 6248 6248 000 9676 624800 000 6248 000 1851 5700 6248n60q45H 5878 5878 000 605207 587800 000 5878 000 2228 5440 5878n60q45I 5982 5993 018 718338 599540 022 5993 018 2240 5800 5993n60q45J 6374 6374 000 2367 637400 000 6374 000 1035 5310 6374n60q1000A 5773 5773 000 10651 577300 000 5773 000 1544 5450 5773n60q1000B 612450 6132 012 718333 613200 012 6132 012 1464 5500 6132n60q1000C 6135 6135 000 69903 613500 000 6135 000 2108 5800 6135n60q1000D 6223 6223 000 994 622300 000 6223 000 1171 5100 6223n60q1000E 6131 6131 000 13215 613100 000 6131 000 1356 5400 6131n60q1000F 580714 5879 124 718332 587800 122 5878 122 2541 5800 5878n60q1000G 6169 6169 000 5757 616900 000 6169 000 1526 5700 6169n60q1000H 5878 5878 000 557665 587800 000 5878 000 1631 5420 5878n60q1000I 5941 5970 049 718302 597000 049 5970 049 1930 5800 5970n60q1000J 6374 6374 000 3263 637400 000 6374 000 1009 5340 6374

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn20q10A 9084 9084 000 013 908400 000 9084 000 11611 3500 9084n20q10B 9883 9883 000 014 988300 000 9883 000 12978 3430 9883n20q10C 14039 14039 000 162 1403900 000 14039 000 38484 4240 14039n20q10D 14925 14925 000 015 1492500 000 14925 000 50100 4600 14925n20q10E 14379 14379 000 082 1437900 000 14379 000 31986 4300 14379n20q10F 9752 9752 000 062 975200 000 9752 000 12542 3600 9752n20q10G 10600 10600 000 019 1060000 000 10600 000 11654 3700 10600n20q10H 11895 11895 000 011 1189500 000 11895 000 22847 4200 11895n20q10I 9229 9229 000 086 924350 016 9229 000 16351 4070 9229n20q10J 9230 9230 000 015 923000 000 9230 000 11386 3740 9230n20q15A 6613 6613 000 011 661300 000 6613 000 2001 2400 6613n20q15B 7404 7404 000 017 740400 000 7404 000 2977 2600 7404n20q15C 9478 9478 000 013 947800 000 9478 000 7469 3110 9478

30

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn20q15D 10376 10376 000 026 1037600 000 10376 000 8349 3300 10376n20q15E 10102 10102 000 026 1010200 000 10102 000 9831 3300 10102n20q15F 7169 7169 000 009 716900 000 7169 000 2815 2900 7169n20q15G 7796 7796 000 062 779600 000 7796 000 2918 2700 7796n20q15Hlowast 8475 8475 000 040 847990 006 8474 -001 4119 3160 8474n20q15I 6652 6652 000 094 665200 000 6652 000 2320 2900 6652n20q15J 6730 6730 000 017 673000 000 6730 000 4063 2900 6730n20q20A 5748 5748 000 041 574800 000 5748 000 1373 2300 5748n20q20B 6296 6296 000 011 629600 000 6296 000 1790 2400 6296n20q20C 8433 8433 000 118 843300 000 8433 000 7778 3000 8433n20q20D 8647 8647 000 030 865220 006 8647 000 7340 3290 8647n20q20E 8623 8623 000 028 862300 000 8623 000 4654 3000 8623n20q20F 6159 6159 000 006 615900 000 6159 000 1138 2500 6159n20q20G 6694 6694 000 010 669400 000 6694 000 1727 2520 6694n20q20H 7208 7208 000 017 720800 000 7208 000 2804 2800 7208n20q20I 5939 5939 000 068 593900 000 5939 000 2014 2700 5939n20q20J 6056 6056 000 037 605600 000 6056 000 2341 2650 6056n20q25A 5083 5083 000 020 508300 000 5083 000 365 1900 5083n20q25B 5387 5387 000 018 538700 000 5387 000 500 1900 5387n20q25C 7229 7229 000 115 722970 001 7229 000 2586 2490 7229n20q25D 7182 7182 000 054 718200 000 7182 000 3412 2600 7182n20q25E 7148 7148 000 088 714800 000 7148 000 1610 2300 7148n20q25F 5121 5121 000 010 512100 000 5121 000 644 2300 5121n20q25G 5501 5501 000 021 550100 000 5501 000 638 2200 5501n20q25H 6027 6027 000 010 602700 000 6027 000 645 2100 6027n20q25I 5082 5082 000 008 508200 000 5082 000 943 2400 5082n20q25J 5150 5150 000 045 515000 000 5150 000 761 2100 5150n20q30A 4702 4702 000 025 470200 000 4702 000 283 1800 4702n20q30B 4769 4769 000 011 476900 000 4769 000 396 1900 4769n20q30C 6012 6012 000 075 601200 000 6012 000 1800 2400 6012n20q30D 5989 5989 000 020 598900 000 5989 000 783 2000 5989n20q30E 6245 6245 000 134 624500 000 6245 000 828 2000 6245n20q30F 4717 4717 000 025 471700 000 4717 000 345 2200 4717n20q30G 5070 5070 000 007 507000 000 5070 000 275 1900 5070n20q30H 5542 5542 000 026 557980 068 5542 000 274 1930 5542n20q30I 4576 4576 000 012 457600 000 4576 000 488 2100 4576n20q30J 4078 4078 000 016 407800 000 4078 000 215 1800 4078n20q35A 4387 4387 000 014 438700 000 4387 000 165 1800 4387n20q35B 4688 4688 000 016 468800 000 4688 000 186 1820 4688n20q35C 5813 5813 000 081 586740 094 5813 000 1087 2170 5813n20q35D 5983 5983 000 029 598300 000 5983 000 810 2000 5983n20q35E 5808 5808 000 042 583600 048 5836 048 753 2020 5836n20q35F 4691 4691 000 032 469100 000 4691 000 216 2100 4691n20q35G 5051 5051 000 038 505100 000 5051 000 288 1910 5051n20q35H 5320 5320 000 038 533090 020 5320 000 413 2090 5320n20q35I 4382 4382 000 017 438200 000 4382 000 341 2000 4382n20q35J 4078 4078 000 018 407800 000 4078 000 197 1800 4078n20q40A 4129 4129 000 009 412900 000 4129 000 180 1900 4129n20q40B 4295 4295 000 008 429500 000 4295 000 145 1930 4295n20q40C 5302 5302 000 038 530200 000 5302 000 530 2100 5302n20q40D 5648 5648 000 079 564800 000 5648 000 647 2010 5648n20q40E 5568 5568 000 020 556800 000 5568 000 331 1800 5568n20q40F 4614 4614 000 019 461400 000 4614 000 144 2100 4614n20q40G 4733 4733 000 029 473300 000 4733 000 189 1900 4733n20q40H 4954 4954 000 039 495400 000 4954 000 261 1900 4954n20q40I 4108 4108 000 012 410800 000 4108 000 110 1900 4108n20q40J 4049 4049 000 022 404900 000 4049 000 185 1800 4049n20q45A 3948 3948 000 012 394800 000 3948 000 097 1700 3948n20q45B 4174 4174 000 006 417400 000 4174 000 135 1900 4174n20q45C 5137 5137 000 148 513700 000 5137 000 402 2100 5137n20q45D 5446 5446 000 042 544600 000 5446 000 526 2000 5446n20q45E 5560 5560 000 046 556000 000 5560 000 374 1900 5560n20q45F 4290 4290 000 004 429000 000 4290 000 161 2100 4290n20q45G 4489 4489 000 007 448900 000 4489 000 102 1900 4489n20q45H 4559 4559 000 012 455900 000 4559 000 125 1900 4559n20q45I 4051 4051 000 007 405100 000 4051 000 088 1900 4051n20q45J 3905 3905 000 021 390530 001 3905 000 108 1770 3905n20q1000A 3583 3583 000 002 358300 000 3583 000 037 1700 3583

31

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn20q1000B 3792 3792 000 003 379200 000 3792 000 035 1800 3792n20q1000C 4042 4042 000 581 404200 000 4042 000 067 2000 4042n20q1000D 3806 3806 000 206 380600 000 3806 000 043 1900 3806n20q1000E 4277 4277 000 074 427700 000 4277 000 054 1800 4277n20q1000F 4108 4108 000 011 410800 000 4108 000 038 2000 4108n20q1000G 4216 4216 000 024 421600 000 4216 000 037 1900 4216n20q1000H 3992 3992 000 013 399200 000 3992 000 041 1900 3992n20q1000I 3960 3960 000 007 396000 000 3960 000 056 1900 3960n20q1000J 3650 3650 000 013 365000 000 3650 000 041 1700 3650n30q10A 14470 14470 000 14107 1447640 004 14470 000 146234 5920 14470n30q10B 15092 15092 000 135916 1509970 005 15092 000 126052 5330 15092n30q10C 12746 12746 000 1600 1278410 030 12746 000 108372 6040 12746n30q10D 11818 11818 000 225 1181800 000 11818 000 77770 5740 11818n30q10E 11704 11704 000 126 1170400 000 11704 000 45125 5410 11704n30q10F 12628 12628 000 090 1263300 004 12628 000 45821 5590 12628n30q10G 21203 21203 000 642 2120300 000 21203 000 325523 6410 21203n30q10H 12656 12656 000 412 1265600 000 12656 000 65712 5380 12656n30q10I 10902 10902 000 454 1090200 000 10902 000 79106 5810 10902n30q10J 12639 12639 000 273 1263900 000 12639 000 95648 5760 12639n30q15A 10099 10099 000 61789 1011060 011 10099 000 18579 4100 10099n30q15B 10234 10234 000 467 1023400 000 10234 000 17059 3840 10234n30q15C 9316 9316 000 286 931600 000 9316 000 20599 4420 9316n30q15D 8662 8662 000 1081 866550 004 8662 000 11944 4320 8662n30q15E 9026 9026 000 526 902710 001 9026 000 15012 4190 9026n30q15F 9028 9028 000 271 902920 001 9028 000 9954 4170 9028n30q15G 14752 14752 000 604 1476370 008 14752 000 71398 4800 14752n30q15H 8993 8993 000 113 899300 000 8993 000 21654 4430 8993n30q15I 7792 7792 000 138 779200 000 7792 000 9720 4200 7792n30q15J 8706 8706 000 086 870600 000 8706 000 10399 4000 8706n30q20A 8794 8794 000 2767 880840 016 8794 000 21920 4290 8794n30q20B 8820 8820 000 3522 882000 000 8820 000 13948 3740 8820n30q20C 8022 8022 000 137 802200 000 8022 000 12062 4140 8022n30q20D 7544 7544 000 540 754400 000 7544 000 8883 3800 7544n30q20E 7360 7360 000 039 736000 000 7360 000 5086 3600 7360n30q20F 8061 8061 000 212 806840 009 8061 000 8829 3880 8061n30q20G 11978 11978 000 408 1197800 000 11978 000 33229 4400 11978n30q20H 7723 7723 000 080 772300 000 7723 000 11112 3910 7723n30q20I 6803 6803 000 153 680300 000 6803 000 7682 3900 6803n30q20J 7788 7788 000 306 778800 000 7788 000 10704 3720 7788n30q25A 7571 7571 000 5565 757420 004 7571 000 7223 3380 7571n30q25B 7055 7055 000 087 705500 000 7055 000 9717 3340 7055n30q25C 6821 6821 000 330 682100 000 6821 000 5355 3460 6821n30q25D 6464 6464 000 093 646400 000 6464 000 5331 3570 6464n30q25E 6474 6474 000 617 647400 000 6474 000 5299 3500 6474n30q25F 6721 6721 000 131 672100 000 6721 000 5487 3700 6721n30q25G 10120 10120 000 1673 1021940 098 10149 029 13183 3270 10142n30q25H 7087 7087 000 178 708700 000 7087 000 4797 3410 7087n30q25I 6207 6207 000 264 620700 000 6207 000 2443 3100 6207n30q25J 6937 6937 000 1329 693700 000 6937 000 6631 3240 6937n30q30A 6236 6236 000 692 623600 000 6236 000 2008 3000 6236n30q30B 6308 6308 000 286 630800 000 6308 000 1279 2600 6308n30q30C 6335 6335 000 3391 633500 000 6335 000 3219 3110 6335n30q30D 6076 6076 000 505 607920 005 6076 000 4247 3380 6076n30q30E 5877 5877 000 106 588240 009 5877 000 1604 2920 5877n30q30F 5695 5695 000 203 569500 000 5695 000 1552 3000 5695n30q30G 8891 8891 000 3810 889100 000 8891 000 8626 3100 8891n30q30H 5884 5884 000 189 588400 000 5884 000 2294 2910 5884n30q30I 5430 5430 000 523 543400 007 5430 000 2016 3100 5430n30q30J 5764 5764 000 200 576400 000 5764 000 1931 3000 5764n30q35A 6024 6024 000 1209 602500 002 6025 002 1537 2900 6024n30q35B 6108 6108 000 996 611050 004 6108 000 2441 2910 6108n30q35C 5962 5962 000 1510 596200 000 5962 000 2167 2940 5962n30q35D 5826 5826 000 368 587280 080 5826 000 2617 3050 5826n30q35E 5788 5788 000 151 578830 001 5788 000 1636 2970 5788n30q35F 5230 5230 000 054 523000 000 5230 000 963 3000 5230n30q35G 8275 8275 000 3596 835320 095 8275 000 7200 3140 8275n30q35H 5678 5678 000 286 567800 000 5678 000 2314 2900 5678n30q35I 5186 5186 000 269 520490 036 5186 000 1304 3020 5186

32

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn30q35J 5622 5622 000 927 562200 000 5622 000 2109 3000 5622n30q40A 5745 5745 000 1671 574500 000 5745 000 1084 3100 5745n30q40B 5570 5570 000 585 557000 000 5570 000 1039 2630 5570n30q40C 5209 5209 000 041 520900 000 5209 000 775 2810 5209n30q40D 5614 5614 000 1702 561400 000 5614 000 1424 2800 5614n30q40E 5127 5127 000 106 512700 000 5127 000 691 2700 5127n30q40F 5221 5221 000 079 522100 000 5221 000 1001 3000 5221n30q40G 7468 7468 000 747 752680 079 7507 052 2647 2830 7468n30q40H 5300 5300 000 078 530000 000 5300 000 1191 2800 5300n30q40I 4775 4775 000 449 477500 000 4775 000 678 2830 4775n30q40J 5469 5469 000 457 546900 000 5469 000 1318 2820 5469n30q45A 5442 5442 000 347 544200 000 5442 000 950 3000 5442n30q45B 5364 5364 000 194 536400 000 5364 000 678 2610 5364n30q45C 5170 5170 000 149 517000 000 5170 000 618 2810 5170n30q45D 5142 5142 000 156 515760 030 5142 000 950 2990 5142n30q45E 5121 5121 000 190 512100 000 5121 000 703 2700 5121n30q45F 4895 4895 000 121 489500 000 4895 000 537 2900 4895n30q45G 7146 7146 000 892 720600 084 7146 000 3023 2960 7146n30q45H 5103 5103 000 206 511280 019 5103 000 901 2920 5103n30q45I 4622 4622 000 204 462200 000 4622 000 416 2700 4622n30q45J 5242 5242 000 498 524290 002 5242 000 1187 2900 5242n30q1000A 4603 4603 000 118 460300 000 4603 000 143 2910 4603n30q1000B 4340 4340 000 41414 434000 000 4340 000 129 2600 4340n30q1000C 4475 4475 000 141499 447500 000 4475 000 200 2800 4475n30q1000D 4666 4666 000 32467 467500 019 4666 000 194 2880 4666n30q1000E 4466 4466 000 017 446600 000 4466 000 087 2700 4466n30q1000F 4384 4384 000 246 438400 000 4384 000 127 2900 4384n30q1000G 4781 4781 000 5242 478100 000 4781 000 157 2700 4781n30q1000H 4030 4030 000 039 403000 000 4030 000 120 2700 4030n30q1000I 4272 4272 000 106 427200 000 4272 000 101 2700 4272n30q1000J 4300 4300 000 155 430000 000 4300 000 135 2700 4300n40q10A 15039 15039 000 20768 1506380 016 15039 000 331023 7860 15039n40q10B 11832 11832 000 1495 1183570 003 11832 000 139533 7130 11832n40q10C 16963 16963 000 7138 1702580 037 16964 001 360077 7690 16964n40q10D 17104 17104 000 2302 1710400 000 17104 000 358680 7110 17104n40q10E 13159 13159 000 1455 1315900 000 13159 000 118301 6980 13159n40q10F 15019 15019 000 185816 1502290 003 15019 000 260658 7450 15019n40q10Glowast 16600 16600 000 4255 1660540 003 16599 -001 338735 7540 16559n40q10H 13578 13578 000 2219 1357800 000 13578 000 297606 7420 13578n40q10I 14849 14849 000 8655 1485330 003 14849 000 277227 7490 14849n40q10J 12646 12646 000 1315 1264600 000 12646 000 137842 7000 12646n40q15A 10487 10487 000 3921 1048700 000 10487 000 46721 5400 10487n40q15B 8491 8491 000 427 849100 000 8491 000 14584 5000 8491n40q15C 11406 11406 000 606 1140600 000 11406 000 47637 5340 11406n40q15D 12166 12166 000 814 1216600 000 12166 000 72755 5120 12166n40q15E 9651 9651 000 3392 965100 000 9651 000 21392 5080 9651n40q15F 10590 10590 000 10928 1063690 044 10603 012 44037 5700 10590n40q15G 11932 11932 000 4994 1193630 004 11932 000 54403 5590 11932n40q15H 9670 9670 000 814 967010 000 9670 000 37756 5550 9670n40q15I 10654 10654 000 646 1065400 000 10654 000 30728 5430 10654n40q15J 9241 9241 000 2290 924100 000 9241 000 20626 5240 9241n40q20A 9051 9051 000 8913 905250 002 9051 000 30629 5000 9051n40q20B 7523 7523 000 2711 752300 000 7523 000 17570 4940 7523n40q20C 9895 9895 000 511 989500 000 9895 000 52216 5090 9895n40q20D 10390 10390 000 7567 1039000 000 10390 000 54125 4850 10390n40q20E 8387 8387 000 12240 838700 000 8387 000 17019 4620 8387n40q20F 9310 9310 000 6789 931000 000 9310 000 37520 5110 9310n40q20G 10085 10085 000 2398 1008500 000 10085 000 37755 5270 10085n40q20H 8685 8685 000 20022 869020 006 8685 000 35773 5320 8685n40q20I 8805 8805 000 218 880500 000 8805 000 20850 4940 8805n40q20J 7760 7760 000 851 776000 000 7760 000 27923 5150 7760n40q25A 7880 7880 000 9005 788000 000 7880 000 13448 4400 7880n40q25B 6607 6607 000 1857 660700 000 6607 000 4766 4100 6607n40q25C 8084 8084 000 822 808400 000 8084 000 15479 4180 8084n40q25D 9049 9049 000 1428 905300 004 9049 000 21924 4460 9049n40q25E 7119 7119 000 1206 711900 000 7119 000 5057 3920 7119n40q25F 7620 7620 000 4460 763800 024 7620 000 12740 4480 7620n40q25G 9041 9041 000 150870 907090 033 9047 007 21353 4550 9041

33

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn40q25H 7488 7488 000 326 748800 000 7488 000 12616 4600 7488n40q25I 7913 7913 000 4290 791300 000 7913 000 12100 4330 7913n40q25J 7226 7226 000 3125 722600 000 7226 000 16790 4550 7226n40q30A 6949 6949 000 3589 696140 018 6949 000 6293 4090 6949n40q30B 5949 5949 000 756 594900 000 5949 000 2764 3900 5949n40q30C 7237 7237 000 1541 723700 000 7237 000 6097 3650 7237n40q30D 7692 7692 000 1021 769200 000 7692 000 11788 3910 7692n40q30E 6424 6424 000 1643 644100 026 6441 026 2552 3400 6441n40q30F 7074 7074 000 61597 709590 031 7085 016 6710 4090 7074n40q30G 7375 7375 000 15415 737650 002 7375 000 6349 4270 7375n40q30H 6556 6556 000 1683 657060 022 6556 000 5540 4190 6556n40q30I 6901 6901 000 1937 690100 000 6901 000 5797 4110 6901n40q30J 6267 6267 000 19132 626910 003 6267 000 3971 4020 6267n40q35A 6780 6780 000 13302 678100 001 6780 000 6187 4000 6780n40q35B 5829 5829 000 3696 583900 017 5829 000 2670 3960 5829n40q35C 6978 6978 000 24670 697800 000 6978 000 7278 3910 6978n40q35D 7341 7341 000 5782 737840 051 7344 004 9366 3920 7341n40q35E 6364 6364 000 22764 638100 027 6381 027 2857 3400 6381n40q35F 6825 6825 000 183526 682870 005 6825 000 6696 4000 6825n40q35G 7146 7146 000 6881 714600 000 7146 000 6432 4230 7146n40q35H 6335 6335 000 6080 635410 030 6335 000 4162 3990 6335n40q35I 6377 6377 000 884 637700 000 6377 000 5264 3910 6377n40q35J 6118 6118 000 37113 611820 000 6118 000 4571 4030 6118n40q40A 6119 6119 000 1246 611900 000 6119 000 2233 3820 6119n40q40B 5615 5615 000 1252 561500 000 5615 000 1981 3800 5615n40q40C 6398 6398 000 1159 639800 000 6398 000 6405 3880 6398n40q40D 6898 6898 000 4303 690000 003 6898 000 6006 3900 6898n40q40E 5969 5969 000 1184 597110 004 5969 000 1700 3470 5969n40q40F 6145 6145 000 1300 614500 000 6145 000 2438 3730 6145n40q40G 6532 6532 000 1866 653200 000 6532 000 4151 4000 6532n40q40H 6050 6050 000 3000 605000 000 6050 000 2962 3910 6050n40q40I 6089 6089 000 2298 609160 004 6089 000 2965 3960 6089n40q40J 5801 5801 000 4067 580100 000 5801 000 3695 3830 5801n40q45A 6059 6059 000 4628 605900 000 6059 000 2077 3800 6059n40q45B 5319 5319 000 324 531900 000 5319 000 1041 3700 5319n40q45C 5912 5912 000 1425 591200 000 5912 000 2298 3570 5912n40q45D 6388 6388 000 929 638800 000 6388 000 3507 3740 6388n40q45E 5671 5671 000 2887 567100 000 5671 000 1194 3400 5671n40q45F 5847 5847 000 5583 584700 000 5847 000 2173 3830 5847n40q45G 6409 6409 000 3426 641240 005 6409 000 2974 4040 6409n40q45H 5575 5575 000 465 557500 000 5575 000 2109 3920 5575n40q45I 5732 5732 000 902 573200 000 5732 000 1674 3720 5732n40q45J 5630 5630 000 5174 563030 001 5630 000 2364 3770 5630n40q1000A 4949 4949 000 105 494900 000 4949 000 266 3800 4949n40q1000B 5110 5110 000 281 511000 000 5110 000 258 3700 5110n40q1000C 4570 4570 000 75952 457000 000 4570 000 534 3680 4570n40q1000D 5202 5202 000 328 520200 000 5202 000 362 3600 5202n40q1000E 5069 5069 000 449 506900 000 5069 000 175 3330 5069n40q1000F 4816 4816 000 399 481600 000 4816 000 226 3700 4816n40q1000G 5253 5253 000 480 525300 000 5253 000 346 3900 5253n40q1000H 4804 4804 000 2707 480400 000 4804 000 285 3800 4804n40q1000I 4713 4713 000 048 471300 000 4713 000 271 3700 4713n40q1000J 4753 4753 000 161 475300 000 4753 000 245 3600 4753n50q10A 13136 13136 000 24592 1313900 002 13139 002 338610 8330 13139n50q10B 2137443 21474 047 718299 2142980 026 21396 010 360270 9810 21396n50q10C 1807606 19623 856 718318 1837560 166 18356 155 360222 9610 18356n50q10D 22824 22824 000 313187 2283230 004 22824 000 360300 9810 22824n50q10Elowast 21223 21223 000 33313 2122280 000 21222 000 360188 9860 21222n50q10F 16566 16566 000 123828 1671970 093 16576 006 360193 9850 16576n50q10G 14018 14018 000 4628 1401800 000 14018 000 359248 9530 14018n50q10H 1864815 18815 089 718262 1868600 020 18672 013 360181 8950 18672n50q10I 16436 16436 000 15735 1646460 017 16440 002 360122 9470 16440n50q10J 18123 18123 000 23243 1812300 000 18123 000 360154 10080 18123n50q15A 9254 9254 000 2067 925400 000 9254 000 25658 5600 9254n50q15B 14697 14697 000 17859 1473070 023 14697 000 197848 7100 14697n50q15C 12630 12630 000 151251 1263000 000 12630 000 193207 6530 12630n50q15D 16449 16449 000 644476 1650490 034 16503 033 282767 7320 16503n50q15E 14812 14812 000 182790 1481290 001 14812 000 235785 7530 14812

34

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn50q15Flowast 11859 11859 000 95482 1185800 -001 11858 -001 135571 7340 11858n50q15G 10226 10226 000 827 1022600 000 10226 000 43505 7030 10226n50q15H 13137 13137 000 216821 1313700 000 13137 000 114834 6420 13137n50q15I 11907 11907 000 3915 1190700 000 11907 000 124520 7330 11907n50q15J 13049 13049 000 386679 1304900 000 13049 000 149872 7240 13049n50q20A 8431 8431 000 7225 843170 001 8431 000 28518 5390 8431n50q20B 1241531 12662 199 718250 1256200 118 12562 118 206570 6500 12562n50q20C 1081135 11220 378 718239 1103860 210 11031 203 214952 6340 11031n50q20D 13139 13139 000 107340 1319560 043 13139 000 205614 6570 13139n50q20E 12532 12532 000 309914 1254890 013 12541 007 255110 6690 12533n50q20F 10075 10075 000 279906 1008060 006 10075 000 102847 6790 10075n50q20G 8870 8870 000 2719 887000 000 8870 000 56102 6550 8870n50q20H 11195 11551 318 718328 1134190 131 11297 091 94361 6030 11297n50q20I 10042 10042 000 37957 1004500 003 10042 000 85678 6370 10042n50q20J 11132 11132 000 185516 1113200 000 11132 000 124375 6810 11132n50q25A 7279 7279 000 5407 727900 000 7279 000 10740 4500 7279n50q25B 10656 10656 000 211639 1065600 000 10656 000 94350 5750 10656n50q25C 912921 9490 395 718325 923020 111 9217 096 68599 5410 9217n50q25D 11364 11364 000 53681 1136960 005 11364 000 101026 5980 11364n50q25E 10241 10241 000 87712 1024100 000 10241 000 94617 5690 10241n50q25F 8685 8685 000 119890 870720 026 8685 000 52943 5750 8685n50q25G 7536 7536 000 1523 755940 031 7536 000 10937 5370 7536n50q25H 947681 9659 192 718317 963680 169 9612 143 36571 5340 9612n50q25I 8614 8614 000 98756 861400 000 8614 000 36778 5410 8614n50q25J 9218 9218 000 9299 921800 000 9218 000 38069 5760 9218n50q30A 6520 6520 000 2378 652600 009 6520 000 4669 4360 6520n50q30B 9025 9025 000 13250 902500 000 9025 000 25204 5010 9025n50q30C 8151 8151 000 176740 815100 000 8151 000 28687 4840 8151n50q30D 9804 9804 000 42998 981390 010 9804 000 44152 5100 9804n50q30E 9086 9086 000 120829 908600 000 9086 000 26272 5060 9086n50q30F 7661 7661 000 24761 766620 007 7661 000 20783 5140 7661n50q30G 6986 6986 000 4572 699120 007 6986 000 6368 4960 6986n50q30H 831901 8699 457 718329 856410 295 8563 293 18897 4930 8563n50q30I 7964 7964 000 210601 800270 049 7964 000 24274 5300 7964n50q30J 8054 8054 000 30755 805400 000 8054 000 16288 5050 8054n50q35A 6410 6410 000 5453 641600 009 6410 000 4122 4230 6410n50q35B 872050 8963 278 718324 878550 075 8763 049 36047 5120 8763n50q35C 773063 7781 065 718326 781760 113 7781 065 31068 5180 7781n50q35D 919919 9404 223 718331 942580 246 9343 156 34279 5250 9343n50q35E 8708 8708 000 382826 871410 007 8708 000 27412 5100 8708n50q35F 7390 7390 000 215151 739970 013 7397 009 14834 4990 7390n50q35G 6681 6681 000 8418 668100 000 6681 000 5655 4900 6681n50q35H 792723 8331 509 718330 803300 133 8033 133 15793 4840 8033n50q35I 7328 7328 000 9101 733210 006 7328 000 13307 5050 7328n50q35J 782819 7981 195 718187 797630 189 7970 181 18943 4940 7970n50q40A 6249 6249 000 3561 624900 000 6249 000 2873 4100 6249n50q40B 818774 8288 122 718327 828800 122 8288 122 25035 4950 8288n50q40C 7158 7158 000 21736 715800 000 7158 000 27746 5060 7158n50q40D 852019 8709 222 718269 867260 179 8657 161 23707 5070 8657n50q40E 8075 8075 000 594443 808710 015 8075 000 17773 4940 8075n50q40F 700967 7042 046 718319 704800 055 7042 046 12014 4910 7042n50q40G 6447 6447 000 2115 644700 000 6447 000 3925 4800 6447n50q40H 756106 7606 059 718322 760960 064 7606 059 9545 4690 7606n50q40I 6926 6926 000 23083 692920 005 6926 000 11565 5120 6926n50q40J 7249 7249 000 27231 725480 008 7249 000 9321 4850 7249n50q45A 5975 5975 000 2899 597500 000 5975 000 1324 4110 5975n50q45B 7697 7697 000 370119 772850 041 7697 000 13778 4760 7697n50q45C 6693 6693 000 14538 669300 000 6693 000 12546 4800 6693n50q45D 8054 8173 148 718294 814010 107 8136 102 12378 4940 8136n50q45E 767150 7737 085 718280 773200 079 7732 079 12000 4670 7732n50q45F 6464 6464 000 8259 646400 000 6464 000 5490 4620 6464n50q45G 6273 6273 000 2453 627300 000 6273 000 3105 4810 6273n50q45H 7005 7005 000 204746 700790 004 7005 000 5198 4690 7005n50q45I 6423 6423 000 4512 642300 000 6423 000 4601 4730 6423n50q45J 6601 6601 000 13661 660100 000 6601 000 4516 4620 6601n50q1000A 5634 5634 000 2223 563400 000 5634 000 399 4110 5634n50q1000B 590386 5961 097 718326 595500 087 5955 087 791 4600 5955n50q1000C 5139 5139 000 2689 513900 000 5139 000 742 4600 5139

35

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn50q1000D 5836 5836 000 097 583600 000 5836 000 701 4600 5836n50q1000E 6023 6023 000 229940 602300 000 6023 000 668 4500 6023n50q1000F 5262 5262 000 623166 526200 000 5262 000 733 4600 5262n50q1000G 5864 5864 000 254687 586400 000 5864 000 787 4800 5864n50q1000H 5559 5559 000 95735 555900 000 5559 000 589 4500 5559n50q1000I 5462 5462 000 1145 546200 000 5462 000 624 4700 5462n50q1000J 5675 5675 000 10974 567500 000 5675 000 627 4400 5675n60q10A 16622 16622 000 481118 1670680 051 16626 002 360277 10750 16624n60q10B 1787414 20003 1191 718308 1812420 140 18092 122 360172 11950 18092n60q10C 1876530 21050 1218 718220 1901070 131 18962 105 360227 11460 18962n60q10D 24095 24095 000 48156 2413840 018 24123 012 360427 11240 24123n60q10E 1877706 19356 308 718321 1893540 084 18805 015 360251 11080 18805n60q10F 17303 17303 000 16802 1733790 020 17303 000 360231 12520 17303n60q10G 18713 18713 000 74890 1875200 021 18715 001 360249 12000 18715n60q10H 16888 16888 000 21103 1699490 063 16920 019 360215 11100 16888n60q10I 20469 20698 112 718321 2066580 096 20526 028 360508 12020 20526n60q10J 16717 16717 000 202781 1672510 005 16717 000 360213 11380 16717n60q15A 1194432 12628 572 718313 1216800 187 12156 177 263077 8100 12156n60q15B 12591 12591 000 94624 1259100 000 12591 000 196851 8360 12591n60q15C 1363460 16390 2021 718297 1376570 096 13760 092 354202 8700 13760n60q15D 17109 17109 000 376523 1710980 000 17109 000 360066 8460 17109n60q15E 1372071 13910 138 718213 1385280 096 13851 095 291099 8250 13851n60q15F 12392 12392 000 35686 1239920 006 12392 000 202812 8830 12392n60q15G 13092 13092 000 23572 1310260 008 13092 000 244064 8450 13092n60q15H 12417 12417 000 11677 1243760 017 12417 000 217147 8680 12417n60q15I 14695 14695 000 103042 1469500 000 14695 000 360077 9050 14695n60q15J 12111 12111 000 132629 1211280 001 12111 000 123631 8210 12111n60q20A 1030988 10454 140 718339 1040700 094 10407 094 249251 7740 10407n60q20B 11057 11057 000 716306 1108800 028 11057 000 276791 8270 11057n60q20C 1125361 11336 073 718305 1129870 040 11298 039 269880 7910 11298n60q20D 13842 13842 000 177627 1385800 012 13842 000 360077 7540 13842n60q20E 11478 11478 000 226720 1147800 000 11478 000 214989 7690 11478n60q20F 10774 10774 000 294769 1079980 024 10774 000 270986 8330 10774n60q20G 11523 11523 000 191091 1152300 000 11523 000 335274 8040 11523n60q20H 10290 10290 000 6328 1029360 003 10290 000 178885 7500 10290n60q20I 1228572 12506 179 718226 1236310 063 12354 056 360055 8290 12354n60q20J 10342 10342 000 4891 1034200 000 10342 000 103641 7450 10342n60q25A 870114 9318 709 718244 890020 229 8870 194 76127 6290 8870n60q25B 9378 9378 000 125364 939950 023 9378 000 144462 7160 9378n60q25C 989691 11090 1206 718193 1008930 194 10045 150 98465 6650 10045n60q25D 1189713 12490 498 718288 1197260 063 11943 039 233587 6550 11937n60q25E 9701 9701 000 45467 970360 003 9701 000 96564 6480 9701n60q25F 9284 9284 000 60248 928400 000 9284 000 107066 7000 9284n60q25G 984344 10054 214 718301 991800 076 9918 076 139330 6940 9918n60q25H 9366 9366 000 121342 936610 000 9366 000 73242 6460 9366n60q25I 10530 10530 000 379476 1053340 003 10530 000 120847 7030 10530n60q25J 9224 9224 000 31201 922650 003 9224 000 63882 6700 9224n60q30A 7810 7810 000 671977 784780 048 7843 042 31494 5890 7843n60q30B 8156 8156 000 609633 817090 018 8156 000 26233 6170 8156n60q30C 889241 10797 2142 718327 909000 222 9090 222 58186 6140 9090n60q30D 10159 10159 000 56202 1015900 000 10159 000 65827 5540 10159n60q30E 886417 9060 221 718308 901830 174 8949 096 41774 5840 8949n60q30F 8163 8163 000 271969 818590 028 8163 000 44028 6490 8163n60q30G 857106 8926 414 718177 866780 113 8665 110 41834 6300 8665n60q30H 8114 8114 000 110851 811760 004 8114 000 22252 5790 8114n60q30I 909266 9189 106 718242 918850 105 9188 105 72071 6230 9188n60q30J 8050 8050 000 108667 807450 030 8050 000 27169 5950 8050n60q35A 741747 7744 440 718324 761540 267 7602 249 36533 5740 7602n60q35B 7779 7779 000 574201 778940 013 7779 000 35148 5920 7779n60q35C 824764 8916 810 718212 837230 151 8359 135 49685 6300 8359n60q35D 968405 10091 420 718253 978570 105 9784 103 73088 5660 9784n60q35E 8283 8283 000 414332 832100 046 8283 000 36861 5880 8283n60q35F 7848 7848 000 491048 785980 015 7848 000 38287 6390 7848n60q35G 8104 8104 000 112529 810400 000 8104 000 47439 6450 8104n60q35H 7576 7576 000 18934 757600 000 7576 000 22477 5640 7576n60q35I 853025 9166 745 718200 871860 221 8675 170 53073 6260 8675n60q35J 7876 7876 000 311661 791770 053 7876 000 25885 5810 7876n60q40A 7112 7112 000 643893 711360 002 7112 000 19053 5550 7112

36

Working Paper

Erdoganet al 2015

ILSSBRP

Avgof all

() (s)Sol

() () (s)NV

expn60q40B 7259 7259 000 40834 725900 000 7259 000 12099 5500 7259n60q40C 791158 8070 200 718239 802140 139 8010 124 32908 6270 8010n60q40D 886150 8959 110 718143 894340 092 8942 091 47594 5680 8942n60q40E 7810 7810 000 169827 781000 000 7810 000 15355 5420 7810n60q40F 7314 7314 000 40102 732330 013 7314 000 21514 6130 7314n60q40G 774425 7860 149 718092 783510 117 7819 097 31484 6090 7818n60q40H 7059 7059 000 67018 705900 000 7059 000 15014 5540 7059n60q40I 8126 8126 000 479857 813110 006 8126 000 39496 6090 8126n60q40J 7407 7407 000 233985 740700 000 7407 000 10224 5330 7407n60q45A 6730 6730 000 171417 673000 000 6730 000 12242 5400 6730n60q45B 6990 6990 000 28031 699000 000 6990 000 8942 5500 6990n60q45C 755557 7630 099 718329 763000 099 7630 099 22051 6040 7630n60q45D 828554 8421 163 718134 841870 161 8395 132 21370 5410 8380n60q45E 7446 7446 000 46737 744600 000 7446 000 11861 5600 7446n60q45F 7075 7075 000 121778 707530 000 7075 000 13343 5880 7075n60q45G 7344 7344 000 328052 734400 000 7344 000 15150 5850 7344n60q45H 6743 6743 000 7259 674300 000 6743 000 7877 5430 6743n60q45I 7577 7577 000 48883 757700 000 7577 000 16804 5910 7577n60q45J 7043 7043 000 34901 704300 000 7043 000 6899 5420 7043n60q1000A 5773 5773 000 13108 577300 000 5773 000 1455 5420 5773n60q1000B 612467 6132 012 718334 613340 014 6132 012 1364 5500 6132n60q1000C 6135 6135 000 56441 613500 000 6135 000 2094 5810 6135n60q1000D 6223 6223 000 958 622300 000 6223 000 1129 5100 6223n60q1000E 6131 6131 000 11910 613100 000 6131 000 1282 5400 6131n60q1000F 581062 5879 118 718340 587800 116 5878 116 2496 5800 5878n60q1000G 6169 6169 000 4336 616900 000 6169 000 1578 5700 6169n60q1000H 5878 5878 000 695860 587800 000 5878 000 1612 5420 5878n60q1000I 594283 5970 046 718321 597000 046 5970 046 1907 5800 5970n60q1000J 6374 6374 000 2781 637400 000 6374 000 1057 5310 6374

lowastInstance in which the cost of our best feasible solution is 1 unit smaller than the cost of the optimal solution reported in

Erdogan et al (2015) (possibly due to rounding issues)

37

1 Introduction
- 31 Solution representation
- 32 Repairing infeasible solutions
- 33 Constructive Procedure
- 34 Local search
- 35 Perturbation mechanisms
- - 4 Computational experiments
  - - 41 Instances
    - 42 Impact of the perturbation mechanisms
    - 43 Parameter tuning
    - 44 Comparison with the literature
    - - 441 Results for instances with up to 60 stations
      - 442 Results for instances with 100 stations
      - 5 Concluding Remarks

Page 2: rebalancing problem - arXiv · 2016-05-05 · rebalancing problem F abio Cruz, Anand Subramanian Centro de Inform atica Universidade Federal da Para ba, CEP 58059-900, Jo~ao Pessoa,