Book of Abstracts
Book of Abstracts
Book of Abstracts
z
a on
Gui
mar
es2
01
4
Op mi
z
a on2014Conf
er
ence
S
CHOOLOFE
NGI
NE
E
RI
NG
UNI
VE
RS
I
TYOFMI
NHO
GUI
MARE
S
,PORTUGAL
J
UL
Y28-30,
2
014
BOOKOFABSTRACTS
Book of Abstracts
Sponsors
We would like to thank our sponsors for their contribution to the success of the conference.
O PTIMIZATION 2014
Contents
Opening Notes
iii
Organizing Committee
vii
Program Committee
ix
General Information
Social Program
Maps
Program Overview
11
Plenary Sessions
17
Sessions
23
Tourism
59
Index
63
O PTIMIZATION 2014
Opening Notes
Welcome to Optimization 2014!
On behalf of the Organizing and Program Committees, we are very pleased to welcome
you to Optimization 2014, the eighth edition of the Optimization series of conferences,
hosted by the Department of Production and Systems, School of Engineering, University
of Minho, in the city of Guimares.
We are very pleased to announce that we have more than 100 presentations and participants, from 12 different countries. We thus hope that the meeting will allow a fruitful
exchange of ideas and promote scientific collaborations.
The success of a conference depends on many factors, one of which is the prestige of the
plenary speakers. We would like to express our gratitude to the Optimization 2014 plenary
speakers for accepting our invitation. We are also grateful to those who kindly accepted
our invitation to organize a session. Finally, we are most thankful to all the participants
who will contribute to enhance the quality of the conference.
We are fully aware that the location of the meeting and the social program also play a
key role. We tried to select some of the gastronomic, cultural, and natural aspects of our
region. We hope you will enjoy the food, the wine, the music, and the landscape that this
part of Portugal has to offer.
On a personal note, we would like to express our pleasure in organizing this meeting.
Finally, we thank our colleagues from the organizing committee who generously contributed with their work and time to the success of the conference. We also thank our
sponsors for the support provided, which was crucial to this organization.
We hope that you enjoy Optimization 2014!
J. M. Valrio de Carvalho | Co-chair, Organizing Committee
A. Ismael F. Vaz | Co-chair, Organizing Committee
O PTIMIZATION 2014
iii
This meeting has international recognition as an important forum of discussion and exchange of ideas. It is the eighth edition of a series of international conferences in optimization organized in Portugal under the auspices of APDIO (the Portuguese Operational
Research Society).
O PTIMIZATION 2014
Organizing Committee
Jos Valrio de Carvalho
A. Ismael F. Vaz
Univ. Minho
Univ. Minho
M. Sameiro Carvalho
Lino Costa
Isabel Esprito Santo
Manuel Carlos Figueiredo
Teresa Monteiro
Lus Nunes Vicente
Univ. Minho
Univ. Minho
Univ. Minho
Univ. Minho
Univ. Minho
Univ. Coimbra
Co-chair
Co-chair
O PTIMIZATION 2014
vii
Program Committee
Lus Nunes Vicente
Univ. Coimbra
Agostinho Agra
Paula Amaral
Miguel F. Anjos
Antnio Pais Antunes
Carlos Henggeler Antunes
Domingos Cardoso
Jos Valrio de Carvalho
Jorge Orestes Cerdeira
Miguel Constantino
Ana Lusa Custdio
Francisco Saldanha da Gama
Jos Fernando Gonalves
Joo Gouveia
Lus Gouveia
Joaquim Joo Jdice
Miguel Sousa Lobo
Helena Ramalhinho Loureno
Carlos Luz
Joaquim R. R. A. Martins
Pedro Coimbra Martins
Jos Fernando Oliveira
Pedro Oliveira
Margarida Vaz Pato
Ana Paula Barbosa Pvoa
Rita Almeida Ribeiro
Antnio Jos Rodrigues
Joo Lus Soares
Tatiana Tchemisova
A. Ismael F. Vaz
Manuel V. C. Vieira
Univ. Aveiro
Univ. Nova Lisboa
cole Polytechnique Montral
Univ. Coimbra
Univ. Coimbra
Univ. Aveiro
Univ. Minho
Univ. Nova Lisboa
Univ. Lisboa
Univ. Nova Lisboa
Univ. Lisboa
Univ. Porto
Univ. Coimbra
Univ. Lisboa
Univ. Coimbra
INSEAD
Univ. Pompeu Fabra, Barcelona
Inst. Pol. Setbal
Univ. Michigan
Inst. Pol. Coimbra
Univ. Porto, INESC Porto
Univ. Porto
ISEG, Univ. Lisboa
IST, Univ. Lisboa
UNINOVA
Univ. Lisboa
Univ. Coimbra
Univ. Aveiro
Univ. Minho
Univ. Nova Lisboa
O PTIMIZATION 2014
Chair
ix
General Information
Language
The conference official language is English. No simultaneous translation will be provided.
Conference Venue
The conference is taking place in the Department of Production and Systems, School of
Engineering, University of Minho, Campus of Azurm, Guimares. A campus map may be
found in page 7, indicating the location of the buildings where sessions, coffee-breaks and
lunch take place. All buildings are within a walking distance.
Registration Desk
The registration desk will be located at the entrance of the Auditorium (Room 1). The registration desk will be open on Sunday 27th, from 16:30 to 19:00, and during the conference
days.
Internet Access
Free wireless access is available through the University Campus (network name: eduroam;
login: optimization2014@guest; password: optimization2014). Warning: This is an unsecured wireless network. Internet access is also available in two computers located in the
hall in front of the Auditorium (Room 1).
G ENERAL I NFORMATION
Ensure that talks respect the program order, to allow participants to jump between
sessions. If a speaker cancels or does not attend, the session schedule should be
respected, rather than shifting every talk backwards.
When appropriate inform the speaker of the remaining time till the end of the presentation.
VERY IMPORTANT: We ask session chairs to notify the session assistant about any
last minute changes or cancellations. These changes will be posted outside the meeting
rooms.
Lunches
Lunches will be served in the restaurant facilities. You are kindly required to wear your
badge. Vegetarian food is available.
2
O PTIMIZATION 2014
G ENERAL I NFORMATION
Coffee-breaks
Coffee, tea, juices, and bottled water will be available during the conference coffee-breaks.
Transportation
Transportation to the conference site is available in the morning. The BUS departures
from Hotel Guimares and goes by Hotel Ibis and Hotel Toural, arriving at the university
campus. More information is available at the hotel reception.
Guimares is a small city, where almost all visiting sites are in a walking distance. However, there is an urban transportation bus that circles the city. The BUS departures and
arrives at the Coach (BUS) Station. A map with the bus stops can be found at TUG website: http://tug.com.pt/horarios/h-l-01.pdf. You may buy a ticket for 1,70 from the BUS
driver (the ticket is valid for one trip).
O PTIMIZATION 2014
Social Program
Monday 28th July, from 18:30 to 19:30: Welcome Reception
The welcome reception will take place at university, next to Auditorium (outdoors).
O PTIMIZATION 2014
Maps
Campus map
Campus de Azurm - Alameda da Universidade
(GPS: [latitude: 41o 27 6.85 N - longitude: 8o 17 33.19 W])
O PTIMIZATION 2014
M APS
Conference Rooms
The sessions take place in five rooms: Room 1 (Auditorium), Room 2, Room 3, Room 4,
and Room 5.
O PTIMIZATION 2014
M APS
O PTIMIZATION 2014
Program Overview
Sunday
July27
08:45
09:00
09:00
10:00
10:00
10:30
Monday
July28
Opening
Tuesday
July29
Plenary
MaartenH.Van
derVlerk
Plenary
Franois
Vanderbeck
CoffeeBreak
CoffeeBreak
10:30
Wednesday
July30
Plenary
Nenad
Mladenovic
CoffeeBreak
Parallel
Sessions
Mon.A
Parallel
Sessions
Tue.A
09:30
10:30
10:30
11:00
11:00
Parallel
Sessions
Wed.A
12:30
12:30
12:30
12:30
Lunch
Lunch
Lunch
14:00
14:00
14:00
14:00
Plenary
SergeGratton
Plenary
MasaoFukushima
Plenary
TimKelley
15:00
15:00
15:00
Closing
Parallel
Sessions
Mon.B
Parallel
Sessions
Tue.B
CoffeeBreak
CoffeeBreak
15:00
16:30
16:30
17:00
17:00
Conference
registration
Parallel
Sessions
Mon.C
18:30
18:30
19:00
Parallel
Sessions
Tue.C
ConferenceTour
(15:1519:45)
Welcome
Reception
(18:3019:30)
Conference
Banquet
(19:4522:30)
O PTIMIZATION 2014
11
P ROGRAM O VERVIEW
12
O PTIMIZATION 2014
P ROGRAM O VERVIEW
O PTIMIZATION 2014
13
P ROGRAM O VERVIEW
14
O PTIMIZATION 2014
P ROGRAM O VERVIEW
Jos Brando Iterated local search algorithm for the open vehicle routing problem with time windows
O PTIMIZATION 2014
15
P ROGRAM O VERVIEW
16
O PTIMIZATION 2014
Plenary Sessions
M ASAO F UKUSHIMA
Faculty of Science and Engineering, Nanzan University, Seto, Aichi, Japan,
[email protected]
S OME R ESULTS IN M ULTI -L EADER -F OLLOWER G AMES
Tuesday, July 29, 14:00-15:00, Room 1/Auditorium
Chair: Joaquim Joo Jdice
Multi-leader-follower games play an important role in modeling complex conflict
situations that arise, for example, in economics and engineering. In this talk, we
discuss some issues in studying such games, and then particularly mention our
recent result on a class of multi-leader-follower games under uncertainty with
some special and simplified structure. By means of the robust optimization technique, the game is first formulated as a robust Nash equilibrium problem, and
then a generalized variational inequality problem. A numerical method is presented to compute a solution of the game.
Professor Masao Fukushima obtained all academic degrees in Engineering from Kyoto
University. Currently he is a full professor at the Faculty of Science and Engineering, Nanzan University, and professor emeritus at Kyoto University. His research interests include
nonlinear optimization, variational inequality and complementarily problems, parallel
optimization, nonsmooth optimization, global optimization, game theory, and applications in transportation, finance, data mining, etc. He has published over 200 papers in
peer reviewed journals and has been selected as an ISI Highly Cited Researcher in Mathematics. Professor Fukushima is one of the founders of the Pacific Optimization Research
Activity Group, and had served as the Chairman of the Working Committee. He is also
the founder and the Co-Editor of Pacific Journal of Optimization. Besides, he is currently
on the editorial boards of 16 international journals in optimization and operations research, including Computational Optimization and Applications, Optimization Methods
and Software, Journal of Optimization Theory and Applications, etc.
O PTIMIZATION 2014
17
P LENARY S ESSIONS
S ERGE G RATTON
CERFACS and INPT-IRIT, Toulouse, France, [email protected],[email protected]
O PTIMIZATION METHODS FOR LARGE SCALE INVERSE PROBLEMS
Monday, July 28, 14:00-15:00, Room 1/Auditorium
Chair: Lus Nunes Vicente
Solving non-convex optimisation problems on parallel computers in a challenge
that arise frequently in Geosciences, especially in the area of numerical prediction of physical phenomena. In this talk we discuss the state of the art algorithms
that are used in operational centers, for solving this problems. Most of them rely
on an adaptation of known optimization techniques, with an appropriate use of
the problem geometry. We will also discuss also the new directions that are considered for future systems, where, due to the increase of the problem complexity,
solution are expected to rely on stochastic approaches, and multi-domain, multifidelity or multi-grid ideas. Some results will be presented in the area of Data
Assimilation for Oceanography.
Serge Gratton got a PhD in applied mathematics in Toulouse. He is currently full professor
of applied mathematics in the University of Toulouse and director the CERFACS-IRIT joint
laboratory. His research nowadays mainly focuses on theoretical and practical aspects of
optimisation methods for large scale gradient based optimization or derivative free optimization. He is also the scientific coordinator of the CERFACS "Parallel Algorithms" team
working on numerical methods for solving systems of equations and optimization problems. He is on the editorial borad of Optimization Methods and Software and has coauthored 60 peer-reviewed papers in international journals covering theoretical and practical aspects of numerical algorithms.
T IM K ELLEY
North Carolina State University, USA, [email protected]
I MPLICIT F ILTERING AND H IDDEN C ONSTRAINTS
Wednesday, July 30, 14:00-15:00, Room 1/Auditorium
Chair: A. Ismael F. Vaz
Implicit Filtering is a derivative-free optimization method which evaluates the
objective function on a set of search directions. If the search fails to find a better point, the next step is to reduce the length of the direction vectors, updated
the directions, and try again. Otherwise one approximates a gradient based on
the values and tries a quasi-Newton step. For unconstrained problems where the
objective function is everywhere defined, the positive and negative coordinate directions work well.
18
O PTIMIZATION 2014
P LENARY S ESSIONS
For many problems, however, the objective function may fail to return a value.
Among the reasons for this are failure of internal simulations or failure of a postiori tests for validity within the function. In this case one must do more than
compute a finite-difference approximation to the gradient and must also augment the direction set to avoid stagnation at non-critical points.
In this talk we will discuss how one resolves these problems, discuss some recent results on functions which have embedded Monte Carlo simulations, and
show, in the context of an application, how our implementation of implicit filtering can be used to solve problems with hidden constraints.
C. T. Kelley is a Drexel Professor of Mathematics at North Carolina State University. His research interests are in numerical methods for nonlinear equations and optimization and
applications of those methods. He is currently working on applications to neutron transport, subsurface hydrology, computational chemistry, and electronic structure computations. He is the author of four books and over 160 papers. He is a fellow of the Society for
Industrial and Applied Mathematics (SIAM), editor-in-chief of SIAM Review, and chair of
the SIAM Board of Trustees.
N ENAD M LADENOVIC
Brunel University, UK, [email protected]
C LUSTERING COMMUNITY NETWORKS BY VARIABLE NEIGHBOURHOOD SEARCH
Wednesday, July 30, 9:30-10:30, Room 1/Auditorium
Chair: Lus Gouveia
In this talk I will first briefly present basic steps and facts regarding Variable
Neighborhood Search (VNS) metaheuristic. Then I will present its three recent
successful implementation in solving clustering problems on community networks. Network clustering is currently hot topic since it can be applied to social
networks, telecommunication networks, biological, internet and all other networks. Many clustering criteria are proposed in the literature taking into account
different needs for grouping. I will present three such clustering criteria solved by
VNS: (1) VNS for divisive hierachical ratio/cut clustering; (2) VNS for multi diversity grouping; (3) Clique partitioning. Computational results and a comparative
analysis will be given too.
Professor Nenad Mladenovic is currently professor (International chair) at LAMIH - CNRS,
University of Valenciennes, France. He received all his degrees from the University of Belgrade, Yugoslavia (BA in mathematics, MSc and PhD in Operational Research). He started
his university career in Belgrade, but has since spent about 10 years in Montreal, Canada at
the Operational Research (OR) center GERAD and at McGill University. He was teaching at
University of Birmingham and Brunel university, UK during 9 years. He has participated in
O PTIMIZATION 2014
19
P LENARY S ESSIONS
more than 20 industrial projects and was leader of several industrial and research projects
in Yugoslavia, Canada, Belgium, U.K. and France. He has had visiting professor positions
in Canada, Spain, Hong Kong, Germany, France and Belgium. In 2012 he became a member of the Academia Europe (The European Academy of Science). He is also a president
of YuSIAM (Yugoslav Society of Industrial and Applied Mathematics), member of Serbian
Scientific society etc. He has written 21 books and chapters, and has published more than
150 papers in edited proceedings and journals (such as SIAM J Computing, Management
Science, Mathematical Programming, Operations Research, INFORMS J on Computing,
OR Letters, EJOR, JOGO, JORS). He serves on the editorial board of a number of journals.
His H and G research indices are 37 and 90 respectively, according to Google scholar. He
has organized many international conferences and was guest editor of more than 10 special issues.
F RANOIS VANDERBECK
University of Bordeaux, France, [email protected]
L AGRANGIAN DUALITY AND COLUMN GENERATION : A REVIEW OF SOLUTION METHODS ,
THEIR COMMON BASE AND SPECIFIC FEATURES , AND EXTENSIONS OF THE PARADIGM
Tuesday, July 29, 9:00-10:00, Room 1/Auditorium
Chair: J. M. Valrio de Carvalho
Large scale Mixed Integer Programming (MIP) optimization models are intricate to solve. When the model decomposes into blocs, for instance when optimizing the use of multiple ressources, relaxing the linking constraints in a Lagrangian way can make the model tractable. Exploiting such decomposition to
develop tight approximations has lead to tremendous progress in solving important applications. In this presentation, we review the alternative methods to solve
the so-called Lagrangian dual model or its linear programming counterpart, the
Dantzig-Wolfe master (associated with the column generation algorithm). We
highlight their drawbacks, their common base, and their specific features, while
discussing extentions of such methods.
Franois Vanderbeck is a Professor in Operations Research at the University of Bordeaux.
He is a member of the Mathematics Institute of Bordeaux and animates an INRIA research
team named RealOpt. Before that, he was a Lecturer at the University of Cambridge and
was associated with both the Engineering Department and the Judge Institute of Management Studies.
Franois holds an Engineering Degree in Applied Mathematics (from UCL), a Master
Degree in Operations Research (from MIT, ORC). He did his Ph.D. in Operations Research
at the Centre for Operations Research and Econometric (CORE, UCL) under the direction
of Professor Laurence A. Wolsey.
His research is in Combinatorial Optimisation, Integer Programming, Reformulation
and Decomposition Methods (Branch-and-Price algorithms) and their Applications (Plan20
O PTIMIZATION 2014
P LENARY S ESSIONS
ning and Scheduling problems, Logistic and Routing, Cutting and Packing problems, Production and Inventory control).
O PTIMIZATION 2014
21
Sessions
Mon.A.1, Monday, July 28, 10:30-12:30,
Room 1
O PTIMIZATION 2014
23
S ESSIONS
24
O PTIMIZATION 2014
S ESSIONS
O PTIMIZATION 2014
25
S ESSIONS
Scheduling (I)
Session chair: Antnio Pais Antunes
1. Automatic timetabling in higher education institutions - Bullet TimeTabler
education
Pedro Fernandes
([email protected])
Bullet Solutions, Armando Barbosa
In a time where the need to reduce costs has
become part of day to day reality of all Educational Institutions, it is inefficient to continue
manually performing those tasks that can be automated and optimized - the creation of timetables. In our previous work, a new automatic and
optimized generator of timetables for Higher Education Institutions was presented - the product Bullet TimeTabler Education (BTTE), which
is successfully used in more than half of the Portuguese Higher Education Schools, including the
10 major ones. BTTE has been improved over the
years, and some important updates to the aforementioned previous work are relevant enough
to be presented, as well as the next research directions that were chosen by Bullet Solutions to
continuously refine the BTTE software.
Keywords:
combinatorial optimization
26
O PTIMIZATION 2014
S ESSIONS
O PTIMIZATION 2014
27
S ESSIONS
28
O PTIMIZATION 2014
S ESSIONS
Keywords:
constrained optimization
O PTIMIZATION 2014
29
S ESSIONS
form
ter estimation
B=
30
N
X
j =1
c j c Tj
with c j IR m , c j 0 N IN .
O PTIMIZATION 2014
S ESSIONS
Scheduling (II)
Session chair:
Gama
Francisco Saldanha-da-
O PTIMIZATION 2014
31
S ESSIONS
additional inequalities. The results of an extensive set of computational tests performed with
the final model are reported.
32
O PTIMIZATION 2014
S ESSIONS
towards the optimum of the current LP problem, we use this optimum as the new solution
of the nonlinear problem. We study to what extent this modification preserves the theoretical
properties of the classic SLP. Further, this modified SLP algorithm is easier to implement and it
readily leads to a heuristic algorithm for mixedinteger nonlinear optimization problems. Interestingly, this heuristic has already been applied
in a real-life problem on gas network optimization.
Keywords: Sequential linear programming, successive, nonlinear optimization, KKT conditions
O PTIMIZATION 2014
33
S ESSIONS
the hospital processes and evaluating their efficiency. In order to obtain the optimal setting,
we combine derivative-free optimization methods with a simulation tool. The resulting framework has been tested on a real healthcare problem.
Keywords:
Simulation-based
optimization,
34
O PTIMIZATION 2014
S ESSIONS
O PTIMIZATION 2014
35
S ESSIONS
Routing (I)
Session chair: Agostinho Agra
1. Integrated approaches for the vehicle
routing and packing problems
Telmo Pinto ([email protected])
University of Minho, Cludio Alves, Jos
Valrio de Carvalho
The capacitated vehicle routing problem (CVRP)
has motivated many variants and several approaches for decades. In the recent past, some
authors proposed the integration of the CVRP
with the bin-packing problem. The combination
of these two problems arises in many real-world
situations. The resulting problem is denominated by capacitated vehicle routing problem
with loading constraints (L-CVRP). The LVRP is
NP-hard and thus difficult to solve to optimality when it considers instances with large number of customers. Not surprisingly, the vast majority of the approaches to the L-CVRP are based
on heuristics. However, some recent works have
addressed both the exact solution of the LVRP
and the exact solution of two-dimensional packing problems explicitly considering loading constraints. In this talk, we intend to describe new
approaches towards the exact resolution of the
LCVRP.
Keywords: Vehicle routing, packing, integer programming
2. Network flow formulation and variable neighborhood search for the location and routing problem with multiple
routes
Rita Macedo
([email protected]) Universit
de Valenciennes, France, Bassem Jarboui,
Bruna Ramos, Cludio Alves, Jos Valrio
de Carvalho, Sad Hanafi, Nenad
Mladenovic
36
O PTIMIZATION 2014
S ESSIONS
O PTIMIZATION 2014
37
S ESSIONS
38
O PTIMIZATION 2014
S ESSIONS
mensional manifold. The latter manifold corresponds to the subspace spanned by both a modified FW direction and the previous search direction computed. This may be regarded as a possible extension of the Conjugate Gradient method
to the constrained case. Global convergence results, similar to the FW method, may be established. We test our proposal over standard constrained problems from the literature.
Keywords: Conditional gradient, quadratic programming, conjugate gradient method
O PTIMIZATION 2014
39
S ESSIONS
40
O PTIMIZATION 2014
S ESSIONS
cretized formulations are more efficient for solving large sized sparse graphs.
Keywords: Cliques in graphs, discretized formulations, biological networks
O PTIMIZATION 2014
41
S ESSIONS
Engineering Applications
Session chair: Aldina Correia
1. Optimization of cogeneration systems
applying direct search methods
Ana Ferreira ([email protected])
University of Minho, Ana C. Ferreira,
Manuel L. Nunes, Lus B. Martins,
Senhorinha F. Teixeira
High efficiency cogeneration systems based on
Stirling engine was considered for optimization
using performance and costs criteria. A thermaleconomic model was developed so that the out-
42
2. Derivative-free optimization of a 3D
composite structure
Ricardo S ([email protected])
Institute of Polymers and
Composites/I3N, Minho University, A.
Ismael F. Vaz, F. W. J. van Hattum, J. M. A.
Csar de S , J. P. Nunes
A case study corresponding to the design optimization of a gantry, a highly critical component of an Industrial Laser Cutting Machine
(ILCM), is presented. The material and shape
in which the new gantry will be produced must
allow for adequate dynamic characteristics. Because of the flexibility in terms of shape and
thickness allowed by composite materials, optimization algorithms have in this field a natural application, allowing for the production of
lighter and cheaper structures. The objective
will be to minimize the structure weight and
cost, by varying its composite material thickness and orientations, while ensuring that the
gantry allows for a pre-established degree of precision. Since the objective function evaluation
O PTIMIZATION 2014
S ESSIONS
Convex Optimization
Session chair: Philipp Hungerlnder
1. KKT point and non-convexity
Suvendu Ranjan Pattanaik
([email protected]) NIT
Rourkela
In this article we consider a convex feasible set
described by inequality constraints that are continuous and not necessarily Lipschitz or convex.
We show that if the Slater constraint qualifica-
tion and a non-degeneracy condition are satisfied, then the Karush-Kuhn-Tucker type optimality condition is both necessary and sufficient. In this way we extend previous results
which are proved for Lipschitz and differentiable
inequalities.
Keywords: Convex optimization, KKT conditions,
non-smooth functions
O PTIMIZATION 2014
43
S ESSIONS
Data Assimilation
Session organized by: Serge Gratton
Session chair: Serge Gratton
1. Observation thinning in data assimilation computations
Ehouarn Simon
([email protected]) ENSEEIHT
and IRIT, Serge Gratton, Monserrat
Rincon-Camacho, Ehouarn Simon,
Philippe Toint
We propose to use an observation-thinning
method for the efficient numerical solution of
44
O PTIMIZATION 2014
S ESSIONS
large-scale incremental four dimensional (4DVar) data assimilation problems. This decomposition is based on exploiting an adaptive hierarchy of the observations. Starting with a
low-cardinality set and the solution of its corresponding optimization problem, observations
are successively added based on a posteriori error estimates. The particular structure of the sequence of associated linear systems allows the
use of a variant of the conjugate gradient algorithm which effectively exploits the fact that the
number of observations is smaller than the size
of the vector state in the 4D-Var model. The new
algorithm is tested on a 1D-wave equation and
on the Lorenz-96 system.
Keywords: Data assimilation, numerical algorithms, multilevel optimization
3. Diffusion solvers for correlation modelling and localization in ensemblevariational data assimilation
Jean Tshimanga Ilunga
([email protected]) CERFACS,
Toulouse, France, Anthony Weaver, Serge
Gratton
Differential operators derived from the explicit
or implicit solution of a diffusion equation can
be used to represent the action of a correlation
matrix on a given vector. This can be exploited
in large-scale applications of variational data assimilation to develop computationally efficient
methods for modelling background-error correlations and/or for localizing correlation matrices estimated directly from a small sample of ensemble perturbations. In this study, we present
theoretical and experimental results on the use
of preconditioned linear solvers for an implicit
formulation of a diffusion-based correlation operator. In particular, we focus our investigation on Krylov methods with Chebyshev acceleration.
Keywords: Linear solvers, Chebyshev acceleration, background-error correlations, data assimilation
Financial Optimization
Session chair: Lus Alberto Coelho
1. Integrated planning of cash-flows and
projects in a discrete-time model
Francisco Lemos
([email protected]) Master in
Information Systems and Management,
Pedro Coimbra Martins
Current days financial crisis brought new challenges for governments, imposing narrower conditions for managing public expenses. We discuss and plan the annual budget to supply a local airfield, being entirely supported by public
funds. The decision maker is intended to reduce the annual budget, along a 25 years stream,
O PTIMIZATION 2014
45
S ESSIONS
46
Routing (II)
Session chair: Jos Brando
1. Combination of ant colony optimisation and exact methods applied to routing problems
Samuel Carvalho ([email protected])
INESC TEC Porto, Ana Maria Rodrigues,
Jos Soeiro Ferreira
An approach based on a combined strategy between Ant Colony Optimization (ACO) and Exact Optimisation Methods to deal with a special
mixed capacitated arc routing problem, will be
presented. The topic of mixing Metaheuristics
and Mathematical Programming is also referred
in the recent literature as Matheuristics. The situation under study, in the context of solid waste
collection, comprises a heterogeneous fleet of
vehicles and different points of collection with
a limited number of visits to empty. The mathematical model linked to routing optimisation
will be described. ACO is a population-based
metaheuristic which has already revealed many
interesting results, namely in routing problems.
Computational results, some of which in connection with a real application, will be shown
and discussed.
Keywords: Ant colony optimisation, metaheuristics, matheuristics, routing problems, solid waste collection
O PTIMIZATION 2014
S ESSIONS
2. Routes for solid waste collection - a
real application
Ana Maria Rodrigues
([email protected]) INESC TEC Porto /
ISCAP - Instituto Politcnico do Porto,
Jos Soeiro Ferreira
This work was developed in cooperation with an
environmental company and addressed a solid
waste collection case arising in the north of Portugal. Two integrated phases are involved in the
modelling and solution procedures: (a) sectorization, in which large municipalities are partitioned into small regions, and (b) routing, to
construct routes for each sector. The presentation focus primarily on routing optimization,
taking into account: a non-homogeneous fleet
of vehicles starting and ending at a garage; the
vehicles are emptied at different places, landfills
and transfer stations; different types of containers some of which must be collected by specific
vehicles; the transfer stations may have limitations related to the number of daily discharges.
Three optimisation models and solution methods are described. The presentation also includes the computational results, part of which
based on the real application.
Keywords: Waste collection, routing, sectoring,
mixed capacitated arc routing problems
O PTIMIZATION 2014
47
S ESSIONS
48
straints are linear, we obtain a linear fractional programming problem. Integer variables
are usually necessary for modelling several real
world issues. MOILFP problems are difficult to
deal with and have not received much attention
in the literature. We will present the last developments focusing on the attempts to solve MOILFP
problems.
Keywords: Multiobjetive integer linear fractional
programming, algorithms
Transportation Planning
Session chair: Joao Luis Miranda
1. Optimization-based park & ride facility planning
Joana Cavadas
([email protected]) CITTA,
Department of Civil Engineering,
University of Coimbra, Antnio Pais
Antunes
Park & ride facilities (P&R) are parking lots located in the periphery of urban areas where
commuters can transfer from cars to public
transport in order to reach their final destinations, thus playing a potentially important role
in decreasing traffic congestion in city centers.
In this presentation, we describe an optimization model for assisting transport planners at
determining the best locations and sizes for a
set of P&R facilities as a function of the budget
available for their construction and operation.
The modal choice of commuters is assumed to
depend on the generalized travel costs of each
mode according to a logit model. The results
that can be obtained through the model are illustrated through hypothetical examples inspired
in real situations.
Keywords: Urban transport, modal choice, park &
ride, optimization model
O PTIMIZATION 2014
S ESSIONS
2. Strategic planning model for freefloating car-sharing systems
Miguel Gueifo Santos
([email protected]) CITTA,
Department of Civil Engineering,
University of Coimbra, Antnio Pais
Antunes
problems; (ii) the advanced session, on the sequence, to consider uncertainty and also how to
use multi-criteria decision making methods; (iii)
the final session, to perform the evaluation of
learning outcomes.
Keywords: Higher education, international cooperation, decision making, transportation, capacity expansion
O PTIMIZATION 2014
49
S ESSIONS
3. A new approach for the gradient sampling using nonmonotone line search
Lucas Eduardo Azevedo Simes
([email protected]) IMECC Unicamp (Brasil), Elias Salomo Helou,
Sandra Augusta Santos
An important tool for solving unconstrained
optimization problems involving a nonsmooth
function which is locally Lipschitz continuous
is the Gradient Sampling (GS) algorithm. After its emergence, many variations of this algorithm have been developed to improve its performance or to overcome implementation issues. This study has the goal to use nonmonotone line search to improve upon the results of
the original GS and to seek for an algorithm that
does not need any kind of differentiability test
during its execution. With these modifications
we intend to reduce the gap between the theoretical and the implemented algorithm.
Keywords: Nonmonotone line search, gradient
sampling methods, nonconvex optimization, nonsmooth optimization
50
O PTIMIZATION 2014
S ESSIONS
O PTIMIZATION 2014
51
S ESSIONS
4. Autonomous predictive driving
utilizing car2car communication
Jrgen Pannek
([email protected]) Universitt
Bremen
We consider a distributed non cooperative control setting in which systems are interconnected
via state constraints. Each of these systems is
governed by an agent which is responsible for
exchanging information with its neighbours and
computing a feedback law using a nonlinear
model predictive controller to avoid violations
of constraints. For this setting we present an algorithm which generates a parallelizable hierarchy among the systems. Moreover, we show both
feasibility and stability of the closed loop using
only abstract properties of this algorithm. To this
end, we utilize a trajectory based stability result
which we extend to the distributed setting. The
scope of the results is illustrated using numerical simulations for crossroad, roundabout and
takeover scenarios.
Keywords: Model predictive control, distributed
control, suboptimality, stabilization, car-to-car communication
Integer Programming
Session chair: Filipe Alvelos
1. Heuristics and relaxations for the diversity management problem
Srgio Marques ([email protected])
Universidade de Aveiro
New heuristic techniques to find and improve
feasible solutions for the optimal diversity management problem are discussed. Feasible solutions are derived from the linear programming
relaxation solution using rounding schemes. To
solve the linear relaxation we use a decomposi-
52
O PTIMIZATION 2014
S ESSIONS
3. On a formulation for the (timedependent) travelling salesman problem
Luis Gouveia ([email protected]) Univ
de Lisboa, DEIO, Centro de Investigao
Operacional, Maria Teresa Godinho,
Pierre Pesneau
We present a new formulation for the TimeDependent Travelling Salesman Problem
(TDTSP). We start by reviewing well known natural formulations with some emphasis on the
formulation by Picard and Queyranne (1978).
The main feature of this formulation is that it
uses, as a subproblem, an exact description of
the n-circuit problem. Then, we present a new
formulation that uses more variables and is
based on using, for each node, a stronger subproblem, namely a n-circuit subproblem with
the additional constraint that the corresponding
node is not repeated in the circuit. Although the
new model has more variables and constraints
than the original PQ model, the results given
from our computational experiments show
that the linear programming relaxation of the
new model gives, for many of the instances
tested, gaps that are close to zero. Thus, the
new model is worth investigating for solving
TDTSP instances. We also present an updated
classification of formulations for the asymmetric
travelling salesman problem (ATSP) where we
"insert" the new time-dependet formulation
presented in the first part of the talk.
Keywords: TSP, ILP reformulations
O PTIMIZATION 2014
53
S ESSIONS
2. A hybrid metaheuristic for the onedimensional cutting stock problem
Angelo Aliano Filho
([email protected])
Universidade Estadual de Campinas,
Antnio Carlos Moretti
This work deals with the problem of the OneDimensional Cutting Stock Problem. This problem is one of the most studied in Combinatorial Optimization and is highly complex to be
solved, because of nature and number of variables involved, making the direct application of
Branch type exact algorithms impractical. For
this reason, we applied a hybrid metaheuristic
named Genetic Algorithm with Simulated Annealing to solve it. Therefore, two heuristics to
(i) build cutting patterns and (ii) to find initial
solutions for the problem were developed. To
validate them, we generated 1800 problems tests
and compared the quality of the solutions of
our method with the solution of the Integer Linear Problem, which are obtained with the patterns generated by the method Column Generation. The results showed a good performance
and robustness of our algorithm, obtaining suboptimal solutions in a reasonable computational
time.
Keywords: Combinatorial optimization, cutting
stock problem, hybrid metaheuristic
54
ket chain, this project concerns the implementation of a Biased Random Key Genetic Algorithm
(BRKGA) for the Shelf Space Allocation Problem.
The BRKGA is a specific class of Genetic Algorithms (GA) that differs from common GA by its
solution representation and the way it combines
the individuals to create new generations. Using a BRKGA API, original ideas for solution decoders were designed, implemented and tested
with real case study instances.
Keywords: Shelf space, genetic algorithm, retail
4. A hybrid BRKGA and local search approach for the minimization of open
stacks problem
Jos Fernando Gonalves
([email protected]) Faculdade de
Economia do Porto
This paper describes a hybrid biased randomkey genetic algorithm (BRKGA) for the Minimization of Open Stacks Problem (MOSP). The
MOSP arises in a production system scenario,
and consists of determining a sequence of
cutting patterns that minimizes the maximum
number of opened stacks during the cutting
process. The approach proposed combines a
BRKGA and a local search procedure for generating the sequence of cutting patterns. A novel
fitness function for evaluating the quality of
the solutions is also developed. Computational
tests are presented using available instances
taken from the literature. The high-quality of
the solutions obtained validate the proposed
approach. Supported by Projects PTDC/EGEGES/117692/2010 and NORTE-07-0124-FEDER000057 financed by the North Portugal Regional
Operational Programme (ON.2 - O Novo Norte),
under the National Strategic Reference Framework (NSRF), through the European Regional
Development Fund (ERDF), and by national
funds, through the Portuguese funding agency,
Fundao para a Cincia e a Tecnologia (FCT).
O PTIMIZATION 2014
S ESSIONS
Semi-continuous
programming,
O PTIMIZATION 2014
55
S ESSIONS
equations
The main goal of this talk is to present our extention of the DuBois-Reymond necessary optimality condition and Noethers first theorem to
56
O PTIMIZATION 2014
S ESSIONS
Filho, Alexandre Cludio Botazzo Delbem
Wed.A.4, Wednesday, July 30, 11:00-12:30,
Room 4
O PTIMIZATION 2014
57
S ESSIONS
58
O PTIMIZATION 2014
Tourism
History of Guimares
Guimares has its origins in the distant 10th century. It was at this time that the Countess Mumadona Dias, widow of Hermenegildo Mendes ordered the construction of a
monastery which became the focal point for a settlement. For its defense she ordered a
Castle to be built on a hill a short distance away, thus creating a second nucleus of development. A street grew-up linking one to the other - the Rua de Santa Maria.
Later the monastery became a chapter house and acquired great importance due to the
privileges and donations by kings and nobility. It became a famous centre for pilgrimage
attracting the prayers and promises of the faithful drawn from all quarters.
While the town continued to grow inside the walls which were erected to defend it, the
orders of poor friars arrived in Guimares and made their contribution to shaping the
town. The twin nuclei subsequently merged into one so that by the 15th century the layout of the city within the walls had been established. Even after the construction of some
churches, monasteries and palaces and the creation of Largo da Misericrdia, its alignment was not significantly altered.
Guimares is recognized as the cradle of the nation and of Portuguese identity. Its magnificently preserved Historic Center was classified as a World Cultural Heritage site by UNESCO in December of 2011.
Museums
The Castle
In the 10th century the Countess Mumadona Dias, in her widowhood, ordered the construction of a monastery in her estates at Vimaranes - today Guimares. The constant
attacks by Moors and Normans led to the construction of a fort to guard and defend the
monks and the Christian community which lived in its purlieus, giving shape to the original castle.
With the formation of the County of Portucalem in the 12th century Count D. Henrique
and Countess D. Teresa came to live in Guimares. They enlarged and strengthened the
castle and according to tradition it was here that they took up residence. King Afonso
Henriques was probably born here.
Between the 13 th and 15th centuries various kings contributed to further improve the
O PTIMIZATION 2014
59
T OURISM
Castle. Throughout its history it was the place of dynastic conflicts, none more heroic
than the leading to the foundation of the kingdom of Portugal at the Battle of S. Mamede in
1128; for this reason it is also known as the Founding Castle or as the Castle of St. Mamede.
After losing its defensive function the Castle was abandoned to progressive decline until
the 20th century when it was declared a National Monument and completely restored.
Opening hours
daily: 10 a.m. to 6 p.m.
Free entry
Chapel of S. Miguel
This small chapel of great simplicity was built at the beginning of the 12th century in Romanesque style probably at the order of Count Henry. It has a symbolic link to the founding of the nation as tradition states that King Afonso Henriques was baptised here. Inside,
the floor is paved with the sepulchres attributed to the noble warriors who fought with
him at the founding of the nation.
It is classified as a National Monument.
Opening hours
daily: 10 a.m. to 6 p.m.
Free entry
O PTIMIZATION 2014
T OURISM
Rato.
In one of the rooms there is a display of weapons including examples of daggers, fire-arms
and parts of armour dating from the 15th and 16th centuries which were collected by the
Viscount of Pindela and subsequently acquired by the state.
The Palace is classified as a National Monument.
Opening hours
daily: 10 a.m. to 6 p.m.
Free entry on Sunday morning
61
T OURISM
Guimares cable-car
The cable-car extends over 1.7 km and climbs 400 m providing the passenger with an unforgettable ride between the city and Penha mountain lasting approximately 10 minutes.
Penha mountain offers the visitor a wide range of services and spaces. Apart from the
sanctuary there is a number of facilities including a camp-site, a guest-house, a mini-golf
course, keep-fit circuits, a horse-riding centre, picnic areas, walks restaurants, bars and
cafs. It is possible to explore numerous grottoes and enjoy the magnificent landscapes
which the natural vantage-points allow.
More information
Further information may be found at the municipality official website:
http://www.guimaraesturismo.com
62
O PTIMIZATION 2014
Index
Plenary Speakers
Session Organizers
B
Batista, Maria Gardnia Sousa . . . . Mon.B.5,
Page 33
Bomze, Immanuel . . . . . . . . Mon.B.2, Page 30
Brando, Jos . . . . . . . . . . . . . . Tue.B.3, Page 46
Brs, Carmo . . . . . . . . . . . . . . Mon.C.2, Page 35
Burdakov, Oleg . . . . . . . . . . . Mon.C.5, Page 38
C
Session Speakers
A
Agra, Agostinho . . . . . . . . . . Mon.C.3, Page 36
Alagador, Diogo . . . . . . . . . . Mon.B.4, Page 32
Aliano Filho, Angelo. . . . . . .Tue.C.4, Page 53
Alvelos, Filipe. . . . . . . . . . . . . .Tue.C.3, Page 52
Alves, Maria Joo . . . . . . . . . . Tue.B.4, Page 47
Amaral, Paula . . . . . . . . . . . . Mon.B.2, Page 30
Amaro, Ana . . . . . . . . . . . . . . . . Tue.A.3, Page 41
Antunes, Antnio Pais . . . . Mon.A.3, Page 26
O PTIMIZATION 2014
63
I NDEX
Elaoud, Semya . . . . . . . . . . . Mon.C.4, Page 37
G
Garmanjani, Rohollah . . . Mon.C.5, Page 38
Giang, Dinh Thanh . . . . . . . . Tue.A.5, Page 43
Gonalves, Jos Fernando . Tue.C.4, Page 53
Gonzlez-Daz, Julio. . . . . .Mon.B.5, Page 33
Gouveia, Luis . . . . . . . . . . . . . . Tue.C.3, Page 52
Grilo, Teresa . . . . . . . . . . . . . . Mon.A.4, Page 27
Gurol, Selime . . . . . . . . . . . . . . Tue.B.1, Page 44
O
Ober-Blbaum, Sina . . . . . . Tue.C.2, Page 51
P
J
Janeiro, Ana Carolina. . . . . .Tue.C.4, Page 53
Jubran, Aparecido Jorge . . Mon.A.5, Page 28
Jubran, Laura Martinson . Mon.B.4, Page 32
K
Knauer, Matthias . . . . . . . . . . Tue.C.2, Page 51
Kwanashie, Augustine . . . . . Tue.A.2, Page 39
L
Lemos, Francisco . . . . . . . . . . Tue.B.2, Page 45
Lopes, Isabel Cristina . . . . . Mon.B.3, Page 31
Lopes, Sofia . . . . . . . . . . . . . . . Wed.A.2, Page 56
64
S
S, Ricardo . . . . . . . . . . . . . . . . . Tue.A.4, Page 42
Sadeghi, Parisa . . . . . . . . . . . Mon.B.3, Page 31
Saldanha-da-Gama, Francisco . . . Mon.B.3,
Page 31
Sampaio, Phillipe R. . . . . . . Mon.C.1, Page 34
Santos, Miguel Gueifo . . . . Tue.B.5, Page 48
O PTIMIZATION 2014
I NDEX
Santos, Sandra Augusta . . . Tue.C.1, Page 49
Santos, Simo P.S. . . . . . . . . . Wed.A.2, Page 56
Silva, Sandra . . . . . . . . . . . . . . Wed.A.4, Page 57
Simes, Lucas Eduardo Azevedo . . Tue.C.1,
Page 49
Simon, Ehouarn . . . . . . . . . . . Tue.B.1, Page 44
Soualmi, Nacer . . . . . . . . . . . Mon.A.1, Page 23
Still, Georg . . . . . . . . . . . . . . . . Mon.B.2, Page 30
T
Teodoro, M. Filomena . . . . Wed.A.2, Page 56
Torres, Delfim F. M. . . . . . . . Mon.A.4, Page 27
Trltzsch, Anke . . . . . . . . . . . Mon.B.1, Page 29
Tshimanga Ilunga, Jean . . . Tue.B.1, Page 44
Tubertini, Paolo . . . . . . . . . . . Tue.A.2, Page 39
V
Vaz, A. Ismael F. . . . . . . . . . . . . Tue.B.4, Page 47
Y
Yang, Kai . . . . . . . . . . . . . . . . . . Mon.A.2, Page 24
Z
Zhang, Zaikun . . . . . . . . . . . . Mon.A.1, Page 23
Plenary Chairs
Masao Fukushima . . . . Joaquim Joo Jdice
Serge Gratton . . . . . . . . . . Lus Nunes Vicente
Tim Kelley . . . . . . . . . . . . . . . . . . A. Ismael F. Vaz
Nenad Mladenovic. . . . . . . . . . . .Lus Gouveia
Franois Vanderbeck . . . . . . . J. M. Valrio de
Carvalho
Maarten H. Van der Vlerk . . . . . . . . Francisco
Saldanha-da-Gama
Session Chairs
Mon.A.1, Page 23 . . . . . . . . . . . . . Serge Gratton
Mon.A.2, Page 24 . . . . . . . . . . . . . . Laura Palagi
Mon.A.3, Page 26 . . . . Antnio Pais Antunes
O PTIMIZATION 2014
65
Sunday
July27
08:45
09:00
09:00
10:00
10:00
10:30
Optimization2014ProgramOverview
Monday
Tuesday
July28
July29
Opening
Plenary
MaartenH.Van
derVlerk
Plenary
Franois
Vanderbeck
CoffeeBreak
CoffeeBreak
10:30
Wednesday
July30
Plenary
Nenad
Mladenovic
CoffeeBreak
Parallel
Sessions
Mon.A
Parallel
Sessions
Tue.A
09:30
10:30
10:30
11:00
11:00
Parallel
Sessions
Wed.A
12:30
12:30
12:30
12:30
Lunch
Lunch
Lunch
14:00
14:00
14:00
14:00
Plenary
SergeGratton
Plenary
MasaoFukushima
Plenary
TimKelley
15:00
15:00
15:00
Closing
Parallel
Sessions
Mon.B
Parallel
Sessions
Tue.B
CoffeeBreak
CoffeeBreak
16:30
16:30
17:00
17:00
Conference
registration
Parallel
Sessions
Mon.C
18:30
18:30
19:00
Parallel
Sessions
Tue.C
Welcome
Reception
(18:3019:30)
Conference
Banquet
(19:4522:30)
ConferenceTour
(15:1519:45)
15:00
Op mi
z
a on
Gui
mar
es2
01
4