Professional Documents
Culture Documents
Textbook On Combinatorial Optimization and Mechanism Design Problems Arising at Container Ports Sebastian Meiswinkel Ebook All Chapter PDF
Textbook On Combinatorial Optimization and Mechanism Design Problems Arising at Container Ports Sebastian Meiswinkel Ebook All Chapter PDF
https://textbookfull.com/product/evolutionary-computation-in-
combinatorial-optimization-arnaud-liefooghe/
https://textbookfull.com/product/international-humanitarian-law-
rules-solutions-to-problems-arising-in-warfare-and-controversies-
marco-sassoli/
https://textbookfull.com/product/extensions-of-dynamic-
programming-for-combinatorial-optimization-and-data-mining-
hassan-aboueisha/
https://textbookfull.com/product/design-of-trajectory-
optimization-approach-for-space-maneuver-vehicle-skip-entry-
problems-runqi-chai/
A New Bio inspired Optimization Algorithm Based on the
Self defense Mechanism of Plants in Nature Camilo
Caraveo
https://textbookfull.com/product/a-new-bio-inspired-optimization-
algorithm-based-on-the-self-defense-mechanism-of-plants-in-
nature-camilo-caraveo/
https://textbookfull.com/product/mobile-crowd-sensing-incentive-
mechanism-design-fen-hou/
https://textbookfull.com/product/value-chains-and-wto-disputes-
compliance-at-the-dispute-settlement-mechanism-aydin-baris-
yildirim/
https://textbookfull.com/product/on-blueberry-hill-1st-edition-
sebastian-barry/
https://textbookfull.com/product/optimization-in-the-real-world-
toward-solving-real-world-optimization-problems-1st-edition-
katsuki-fujisawa/
Produktion und Logistik
Sebastian Meiswinkel
On Combinatorial
Optimization and
Mechanism Design
Problems Arising
at Container Ports
Produktion und Logistik
Kontakt
Professor Dr. Thomas S. Spengler
Technische Universität Braunschweig
Institut für Automobilwirtschaft
und Industrielle Produktion
Mühlenpfordtstraße 23
38106 Braunschweig
On Combinatorial
Optimization and
Mechanism Design
Problems Arising at
Container Ports
With a foreword by Prof. Dr. Erwin Pesch
Sebastian Meiswinkel
Siegen, Germany
Springer Gabler
© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with
regard to jurisdictional claims in published maps and institutional affiliations.
This Springer Gabler imprint is published by the registered company Springer Fachmedien Wiesbaden
GmbH part of Springer Nature
The registered company address is: Abraham-Lincoln-Str. 46, 65189 Wiesbaden, Germany
Foreword
Globalisation and the internet are main driving forces for the rapid increase of freight
transport over the past years, and its growth is predicted to continue in the same rate for
the next decade or more amongst others due to China’s One Belt One Road (or New Silk
Road) initiative, that is intended to intensify the trade between Europe and Asia along
the former silk road.
Intermodal transport is mostly the transport of containers in one supply chain in
more than one transport mode, e.g, by rail, road or sea. While the long distances are
covered by huge container vessels that periodically operate between sea ports, or by
intercontinental trains, trucks are operating on the so called last mile between customers
and a rail transshipment terminal or a sea port in order to deliver or pick-up the freight.
Transshipment yards and container ports are main points where the transportation
mode of goods can be changed. Since rail and roads are operating at their capacity
limits, an acceleration of freight transport requires faster processes in order to improve
the efficiency of cargo handling at container terminals and to cope with the increasing
shipping volume.
In this book “On Combinatorial Optimization and Mechanism Design Problems Aris-
ing at Container Ports” the author concentrates exactly on that, analysing different trans-
portation problems arising at container terminals from a quantitative point of view.
In the first part, the author discusses situations where an operator of a container
terminal serves clients who can be assumed to behave selfish. If the operator of a container
terminal requires private information from the clients for a socially optimal decision,
it is necessary to provide an incentive or mechanism that all clients commit their true
information and do not try to influence the outcome by false information. Algorithmic
mechanism design is a research area that deals with the construction of these methods to
make the clients to be truthful. The author made a major contribution to this field by
improving the understanding of truthfulness in case of specific scheduling problems.
The second part of this book deals with real world optimization problems without
any private information that appear at container ports. Efficient handling of the internal
vi Foreword
processes is important for the terminal being competitive, since many terminals are under
high competitive pressure and therefore need to optimize their processes.
This book should be most suitable to researchers and students of logistics and oper-
ations research. In addition, the contents of this book might be very interesting to those
in industry who need to solve problems on the design, operation, and management of
container ports.
This thesis represents the result of my PhD study at the Chair of Management Information
Science at the University of Siegen. This time has been a challenging and long journey,
but I was fortunate enough to be accompanied by many people. I thank all of them,
particularly the ones below.
First, I would like to express my great gratitude to my supervisor Prof. Dr. Erwin
Pesch. He offered me the opportunity to work at the Chair of Management Information
Science and to conduct my PhD research. He was very open-minded in his supervision
and I enjoyed very much the freedom he gave me to pursue the research direction I
like. Throughout my study, he continuously supported me with his broad knowledge in
the fields of logistics and operations research. His professional comments and invaluable
advices have helped me improve the scientific quality of this thesis.
I would also like to thank Prof. Dr. Rob van Stee for serving as the second referee of
my thesis and PD Dr. Sergei Chubanov for being member of the doctoral committee.
My great gratitude goes also to Dr. Dominik Kreß who shares an office with me. The
considerable discussions we had and the constructive suggestions he made on my work
were very helpful. I would also like to thank Dr. Alena Otto for proofreading important
parts of my thesis.
I am thankful to my colleagues David Müller, Xiyu Li and Roswitha Eifler for sup-
porting me and for participating in my defense. My appreciation also goes to my former
colleagues Dr. Jenny Nossack.
Last but not least, I would like to express my gratitude to my family and friends and
especially to my wife Anja for the strong support they have given me during the time I
have been working on this thesis.
List of Tables xv
Bibliography 113
List of Figures
1.1 Algorithmic mechanism design (in case of direct revelation) and scheduling
games . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 (Direct revelation) algorithmic mechanism design (Kress et al., 2017) . . . 7
Container ports are important parts of global transport chains. They are used to transship
containers between vessels and vehicles for land transport. Large container ports often
operate own rail-road terminals in order to connect the port with the hinterland efficiently.
New challenges and perspectives for container ports, and therefore for intermodal
transport logistics, mainly arise due to the ongoing increase of the container flow and the
resulting need to improve the throughput of containers. One way to do so is to enlarge
the port by building new berths and other necessary structures. If this is not possible or
too expensive, another way to improve the throughput of an existing port is to improve
the utilisation efficiency of the used equipment. Danish Ship Finance (2016) reports that
the world’s container market demand in 2016 has increased by 2.5 % in comparison to
the demand in 2012. It is furthermore projected to increase by another 4.7 % from 2016
to 2019. Moreover, the Institute of Shipping Economics and Logistics (2016) states a 8.4
% increase of the size of the international container fleet in 2015 and a doubling of the
average size of new container ships since 2009. This development puts the entire logistic
chain, and especially container ports and megahubs, under high pressure.
In order to cope with the increasing container throughput, container ports automa-
tize processes and decision making (e.g., by using automated guided vehicles and (semi-)
automated quay and stacking cranes as implemented at Container Terminal Altenwerder
(CTA) in Hamburg, Germany). Automatization, however, induces the need for thorough
analysis of all involved tasks. Simulation studies and optimization techniques help to orga-
nize the container throughput and to reduce the ships’ berthing time, which is considered
the major cost driver and thus the main objective of a port’s operator (Steenken et al.,
2004). However, apart from this objective, container ports also need to stay competitive
and attractive to their customers. Topics such as quality of service, waiting times, sus-
tainability, or waste minimization are increasingly important to the customers and have
to be addressed.
© Springer Fachmedien Wiesbaden GmbH, part of Springer Nature 2018
S. Meiswinkel, On Combinatorial Optimization and Mechanism Design
Problems Arising at Container Ports, Produktion und Logistik,
https://doi.org/10.1007/978-3-658-22362-5_1
2 1. Introduction and Preliminaries
on the jobs in J, then we call them unrelated (parallel) machines. We refer to the speed
of machine i ∈ M for processing job j ∈ J as sij .
In the following, we define the data that is used to characterize a job j ∈ J in this
thesis. We denote by tij the processing time of job j ∈ J on machine i ∈ M . In case of
identical machines we can drop the index for the machine, i.e. tj = tij for all i ∈ M . If
the machines are uniform, then tij = tj /si with i ∈ M , where tj is the standard processing
time. Analogously, we have tij = tj /sij with i ∈ M for unrelated machines. All processing
times are assumed to be positive numbers. The arrival time or release date rj is the time
at which job j is ready for processing. If the release dates are the same for all jobs j ∈ J,
then it is assumed rj = 0 for all j ∈ J. The due date of the job is denoted by dj . It
describes a time by which j should be completed. The weight wj specifies the relative
importance of job j.
Next, we present some definitions concerning the schedules and optimality criteria.
A feasible schedule o is an assignment of all jobs from J to machines from M together
with starting and/or completion times for each job such that the following conditions are
satisfied:
• each job is processed by exactly one machine and each machine processes at most
one job at a time,
We denote the set of all feasible schedules of a given machine scheduling problem by O.
For a given schedule o ∈ O we denote the completion time of a job j ∈ J by Cj . It
describes a time at which the processing of job j is completed. Furthermore, we define a
mapping Cj : O → R≥0 to describe the influence of the specific solution on the completion
times. We denote by Uj a step function Uj : O → {0, 1}. If job j completes strictly after
dj in a schedule o ∈ O, Uj (o) is equal to 1. Otherwise Uj (o) is equal to 0. The load Li of a
machine i ∈ M is defined as the sum of the processing times of all jobs that are assigned
to machine i. Since the load depends on the realized schedule o, we define the mapping
Li : O → R≥0 .
Table 1.2 provides an overview over the machine scheduling related notation used in
this thesis.
Graham et al. (1979) present a widely used and generally accepted classification
scheme for machine scheduling problems. It represents a specific problem by a three-field
1.1. Notation and Terminology 5
wj weight of job j ∈ J
rj release date of job j ∈ J
dj due date or deadline of job j ∈ J
Cj completion time of job j ∈ J Cj : O → R≥0
Uj unit penalty of job j ∈ J: 1 if j completes strictly after dj , 0 otherwise Uj : O → {0, 1}
Li load of machine i ∈ M Li : O → R≥0
notation, α|β|γ, where α describes the machine environment, β refers to job character-
istics, and γ relates to the (global) performance measure (optimality criterion). Each
field of the triple includes multiple elements, e.g. α = α1 , α2 , . . . , that represent specific
problem properties. The empty symbol, ◦, denotes the default value of an element and is
skipped when a triple is actually specified. A detailed look in the classification scheme is
given in Section 2.3.1. We refer to Błażewicz et al. (2007) and Leung (2004) for a more
detailed discussion.
In this section we give an overview over the basic notation and terminology concerning
game theory and algorithmic mechanism design as in Kress et al. (2018b). The games
considered throughout this thesis have three basic elements: players, strategy spaces, and
utility functions. Furthermore, we will restrict ourselves to considering non-cooperative
games. That is, players cannot form coalitions in order to generate group decisions. In
the context of machine scheduling problems, players may be machines or jobs. More
generally, one may also think of “owners” of multiple machines or jobs that act as single
players. Each player has an associated strategy space that represents the options that
the player can select from when the game is played. For example, when the players
correspond to jobs, each job may be allowed to select a machine to be processed on.
A player’s utility function assigns a utility level to every vector of strategies, i.e. each
combination of strategies that can potentially be selected by all players. With respect to
machine scheduling problems, the utility level could, for instance, be the completion time
of a given job.
We will consider fairly specific problem settings in the field of algorithmic game theory
6 1. Introduction and Preliminaries
for machine scheduling problems. These settings are characterized by the existence of
(rational and selfish) players, who are typically referred to as agents and can make a
single claim on some piece of information that may affect the final schedule. Furthermore,
there exists a central authority/planner, who is in charge of designing an interaction
protocol, a rewarding scheme (e.g. payments among players), and a scheduling algorithm
that determines the final schedule. Within this scope, there are two main streams of
literature that differ in the type of information that the agents possess and in the way
that the information affects an instance of the considered machine scheduling problem
(see Figure 1.1). In this thesis, we will focus one of these streams, which presumes that
Public Information
problem instance
designs and controls
Central Authority/Planner Algorithm Rew. Scheme
schedule “rewards”
Figure 1.1: Algorithmic mechanism design (in case of direct revelation) and scheduling games
the agents have private information on their own characteristics. Jobs, for instance, may
privately know their due dates or job weights. The remaining data, e.g. the number
of machines and jobs, is usually assumed to be publicly known. The central planner
designs some protocol of interaction that the agents have to follow. This protocol may
be fairly general. We will, however, restrict ourselves to considering “direct protocols”
that allow the agents to solely (but not necessarily truthfully) announce concrete values
that represent their private information when the game begins. In terms of optimization
problems, these agents therefore fix a subset of parameters. When acting selfishly, they
will try to influence the solution determined by the scheduling algorithm by submitting
false information. However, by designing appropriate algorithms and rewarding schemes
that set the right incentives, the central planner can extract the true information of
these players, for example, in order to generate fair solutions with respect to some social
criterion that considers the interests of all agents. In the second stream, that we do not
consider in this thesis (the interested reader may refer to Heydenreich et al., 2007), the
(usually completely informed) agents, again pursuing selfish goals, commit decisions on
machine-job assignments and thus implicitly fix variables of optimization problems. We
can, for example, think of jobs that choose to be processed on specific machines.
We would like to stress the fact that the aforementioned fields of research are not
1.1. Notation and Terminology 7
always clearly separated in the literature. Similarly, the terms used to identify specific
problems within these fields may differ among different articles. We will follow Nisan
and Ronen (2001), who define (algorithmic) mechanism design to aim at “study[ing] how
privately known preferences [...] can be aggregated towards a ‘social choice’ ” (see also
Nisan and Ronen, 1999), which corresponds to the first stream described above. Our focus
on direct protocols is usually termed direct revelation (see, for example, Nisan, 2007).
Others use the term “algorithmic mechanism design” in a more general context, even
when there is no privately owned information (see, for instance, Immorlica et al., 2009).
Problems in the second stream are sometimes referred to as (machine) scheduling games
(see, for instance, Harks et al., 2011; Roughgarden and Tardos, 2007) or load balancing
games (Vöcking, 2007). These games are closely related to the categories of congestion
games (Rosenthal, 1973) and coordination mechanisms (Christodoulou et al., 2009a). In
all of these areas, one is usually interested in deciding whether (Nash) equilibria exist,
how (in-)efficient these equilibria are when compared to socially optimal solutions, and
how fast algorithms can compute them (Harks et al., 2011; Roughgarden and Tardos,
2007).
Based on the illustration in Figure 1.2, we will now describe the (direct revelation) algo-
rithmic mechanism design setting in the context of machine scheduling problems in more
detail. The corresponding notation used throughout this thesis in context of mechanism
o
v1t
p1 (v)
1
V1
Mechanism Schedule
Set A of agents
v1
... f
p1 , . . . , p|A|
f (v)
o
v|A|
t
v|A|
|A| p|A| (v)
V|A|
o private information
public information
Figure 1.2: (Direct revelation) algorithmic mechanism design (Kress et al., 2017)
A set of agents
Vk set of potential valuation functions for agent k ∈ A
V Cartesian product of sets Vk , k ∈ A V = V1 × · · · × V|A|
V−k Cartesian product of sets Vl , l ∈ A \ {k} V = V1 × · · · × Vk−1 × Vk+1 × · · · × V|A|
f social choice function/allocation rule f :V →O
vkt true valuation function of agent k ∈ A Vk 3 vkt : O → R
vk claimed valuation function of agent k ∈ A Vk 3 v k : O → R
pk payment function for agent k ∈ A pk : V → R
uk utility function of agent k ∈ A uk (vk , v−k ) = vkt (f (v)) + pk (v)
v−k vector of claimed valuation functions except vk , k ∈ A v−k = (v1 , . . . , vk−1 , vk+1 , . . . , v|A| )
v = (v1 , . . . , v|A| )
v vector of claimed valuation functions
v = (vk , v−k ), k ∈ A
sometimes referred to as the agent’s type. Negative values can, for example, relate to
costs incurred to a (job) agent due to waiting for being completed.
Each agent k ∈ A reports a valuation function vk , that may deviate from the true
valuation function vkt , to the mechanism. Each valuation function vk , k ∈ A, is element
of a publicly known set Vk . We define V := V1 × · · · × V|A| . Furthermore, we denote
the vector of all valuation functions reported to the mechanism by v = (v1 , . . . , v|A| )
and the vector of all valuation functions reported to the mechanism except of vk by
v−k = (v1 , . . . , vk−1 , vk+1 , . . . , v|A| ). For the sake of notational convenience, we will use v
and (vk , v−k ) interchangeably.
The mechanism itself is designed and controlled by a central planner. It is a pair
(f, p), composed of a social choice function f : V → O and a vector of payment functions
p := (p1 , . . . , p|A| ), with pk : V → R for all k ∈ A. The mechanism (f, p) is said to
implement the social choice function f . It is efficient, if f optimizes the given global
objective function (Heydenreich et al., 2008; Mitra, 2001, 2002). As described in Section
1.1.2, in the context of scheduling problems, the social choice function is an algorithm
that determines a feasible schedule based on the valuation functions reported to the
mechanism. It is also referred to as the scheduling rule or allocation rule. By controlling
the allocation rule and the payment functions, the cental planner can design mechanisms
with different features.
Each agent k ∈ A selfishly aims to maximize the utility function uk : V → R, which
is assumed to be quasi-linear, i.e. corresponds to the sum of the agent’s valuation of the
schedule (determined by the allocation rule) and the (potentially negative) corresponding
payment from the mechanism, uk (vk , v−k ) := vkt (f (vk , v−k )) + pk (vk , v−k ). Sometimes it
is reasonable to focus on individually rational mechanisms (also referred to as voluntary
participation mechanisms, see Auletta et al., 2004a), that assume (or feature) the utilities
1.1. Notation and Terminology 9
of each agent to always be non-negative (see, for instance, Hoeksma and Uetz, 2013; Nisan,
2007).
All of the above definitions assume a deterministic problem setting. Unless stated other-
wise, this will also be our standard assumption throughout the remainder of this thesis.
The literature, however, also considers two main non-deterministic settings. First, one
can assume the allocation rule to be non-deterministic, i.e. let the scheduling algorithm’s
logic employ some degree of randomness, or consider randomized payments. A resulting
mechanism is then referred to as a randomized mechanism (see, for example, Angel et al.,
2012; Nisan, 2007). Second, one can deviate from the assumption of the agents having no
information at all about the private information of the other agents. Sometimes, it may
be appropriate to assume that there exists some commonly known probability distribu-
tion over the private information of each player (Nisan, 2007). In both non-deterministic
cases, agents are usually assumed to maximize expected utilities. The definitions of the
standard setting carry over to the non-deterministic settings in a straightforward manner.
As indicated before, agents selfishly aim to maximize their (expected) utility functions
and may therefore lie about their true valuation functions. To overcome this problem, the
central planner may want to design the mechanism such that agents behave truthfully.
The literature considers different concepts of truthfulness. We will briefly outline the
concepts that are relevant for this thesis in this section.
A mechanism is (dominant strategy) incentive compatible or truthful (Nisan, 2007) if
it guarantees that reporting the true valuation function maximizes the utility function
of a rationally acting agent for all possible vectors of claimed valuation functions of the
other agents, i.e. if uk (vkt , v−k ) ≥ uk (vk , v−k ) for all k ∈ A, all vk ∈ Vk , and all v−k ∈ V−k .
In case of randomized mechanisms, articles usually apply an adapted notion of truth-
fulness, referred to as truthfulness in expectation. Formally, let E(uk (v)) denote the
expected value of the utility function of agent k ∈ A over the randomization of the mech-
anism. A mechanism is truthful in expectation if E(uk (vkt , v−k )) ≥ E(uk (vk , v−k )), for all
k ∈ A, vk ∈ Vk , and v−k ∈ V−k . Alternatively, one may slightly deviate from our def-
inition in Section 1.1.2, and define a randomized mechanism to allow distributions over
deterministic mechanisms. Then, a randomized mechanism is defined to be truthful in
the universal sense if every deterministic mechanism in the support is dominant strategy
incentive compatible (Nisan, 2007).
10 1. Introduction and Preliminaries
Similarly, when considering the case of publicly known probability distributions over
the type spaces of agents (that we will denote by Φk for agent k ∈ A) as described in
before, one can apply a weaker notion of truthfulness, referred to as Bayes-Nash incentive
compatibility (see, for example, Duives et al., 2015; Heydenreich et al., 2008; Hoeksma
and Uetz, 2013). Here, for each agent, telling the truth must be (weakly) dominant in
expectation over the publicly known distributions over the type spaces of the other agents.
One of the most important general results in the field of mechanism design is the
Vickrey-Clarke-Groves mechanism (VCG mechanism), that was suggested by Vickrey
(1961) and generalized by Clarke (1971) and Groves (1973). A mechanism is called a
VCG mechanism, if the social choice function maximizes social welfare, i.e. the sum of
the valuation functions of all agents, and if the payment functions pk (v), k ∈ A, are given
by
Xn
pk (v) = hk (v−k ) − vl (f (v)),
l=1,l6=k
In this section, we introduce the notation used for graphs in this thesis. The used nota-
tion follows standard combinatorial optimization books as Korte and Vygen (2008) and
Schrijver (2003).
A graph G = (V, E) consists of two (finite) sets V and E. The set V is referred to as
vertex set or node set and we define n := |V |. The elements of V are called vertices or
nodes. The set E is called edge set and we define m := |E|. The elements of E are referred
to as edges. Edges can be directed or undirected. An undirected edge e is represented
1.1. Notation and Terminology 11
In the following, we introduce some optimization problems that are defined on graphs and
considered in this thesis.
A matchings M on an undirected graphs G = (V, E) is a subset of E such that all
elements in M are disjoint. In other words, each node v ∈ V is incident to at most one
12 1. Introduction and Preliminaries
edge in M . The matching with maximal cardinality is called maximum matching. The
problem of finding a matching M on a graph G = (V, E) with maximal cardinality is
called maximum matching problem. Given an edge-weighted graph G = (V, E, w), the
weight w(M ) of a given matching M on G is the sum of the weights of all edges in M ,
P
i.e. w(M ) = e∈M we . Given an edge-weighted bipartite graph G = (U, V, E, w), the
problem of finding a maximum matching of minimum weight is referred to as min-sum
weighted matching. These basic matching problems are well studied problems. Further
details and information on these problems can be found, for example, in Burkard et al.
(2009).
Another well studied optimization problem is the (symmetric) traveling salesman
problem (TSP). Given an edge weighted undirected graph G = (V, E, w), the problem is
to find a cycle that contains all nodes in V with minimal weight. The weight of a circle
W is defined as the sum of the weights of all edges in W . In context of the TSP, the circle
W is often referred to as a tour. The asymmetric traveling salesman problem is defined
analogously on edge weighted directed graphs. Further information on the TSP and its
variants can be found, for example, in Gutin and Punnen (2007).
1.2 Outline
The remainder of this thesis is organized as follows. In Chapter 2, we present a review of
recent contributions in the field of machine scheduling problems in the context of algo-
rithmic mechanism design. We review the categories and characterizing problem features
of machine scheduling settings in the algorithmic mechanism design literature and extend
the widely accepted classification scheme of Graham et al. (1979) for scheduling prob-
lems to include aspects relating to mechanism design. Based on this hierarchical scheme,
we give a systematic overview of recent contributions in this field of research. In Chap-
ter 3, we consider two machine scheduling problems that arise at container ports when
dealing with customers. Both problems are investigated with methods from the field of
algorithmic mechanism design. The first part of the chapter is concerned with machine
scheduling problems in context of one-parameter valuation function domains. We consider
the problem of minimizing the total weighted completion time of all jobs and investigate
the truthfulness of List-Scheduling algorithms for parallel machines. In the case of one
machine, we come up with results for a budget-balanced VCG mechanism. In the second
part, we consider machine scheduling problems in context of two-parameter valuation
function domains. We derive a set of properties that is equivalent to the well-known
condition of cycle monotonicity, which is a general condition for truthful mechanisms in
non-convex valuation function domains. Our results utilize knowledge about the underly-
1.2. Outline 13
ing scheduling problem, so that the resulting properties are easier to implement and verify
than the general condition of cycle monotonicity. We illustrate the use of our results by
analyzing an example algorithm that has recently been proposed in the literature for the
case of one machine.
Chapter 4 and 5 are concerned with “classic” optimization problems. Both chapters
have in common, that they consider optimization problems dealing with internal container
movements between the quay and the storage area of the ports. In Chapter 4, we intro-
duce and analyze the Partitioning Min-Max Weighted Matching Problem (PMMWM).
PMMWM combines the problem of partitioning a set of vertices of a bipartite graph into
disjoint subsets of restricted size and the strongly NP-hard Min-Max Weighted Match-
ing (MMWM) Problem, that has recently been introduced in the literature. In contrast
to PMMWM, the latter problem assumes the partitioning to be given. Applications of
the PMMWM arise at small and midsize container ports when reach stackers are used
to transport containers from a temporary storage area to the long-term storage area.
Other applications of the PMMWM arise at the rail-road terminal of a container port.
We propose a MILP formulation for PMMWM and prove that the problem is NP-hard
in the strong sense. Two heuristic frameworks are presented. Both of them outperform
standard optimization software. Our extensive computational study proves that the al-
gorithms provide high quality solutions within reasonable time. Chapter 5 deals with a
straddle carrier routing problem that arises at container ports where containers need to
be exchanged between a storage area of the sea port and a small buffer for locally storing
loaded or unloaded containers within reach of the quay cranes. The problem is how to
route the container carrying straddle carriers such that a loading or unloading sequence at
the quay crane is respected and the turnaround time of the vessel is minimized. The prob-
lem is proven to be strongly NP-hard and we present a mixed-integer programm based
on the asymmetric traveling salesman problem with precedence constraints. We propose
two decomposition heuristics and compare the most promising one with an approach used
in practice. Computational experiments are based on real-world data. The thesis closes
with a summary and an outlook on future research in Chapter 6.
Chapter 2
in Section 2.5.
• in offline-settings, where all information regarding the problem is known or has been
announced (in contrast to online-settings) at the unique time of planning,
• and have private information on their own characteristics which they directly (but
not necessarily truthfully) announce by making a single claim,
Updated editions will replace the previous one—the old editions will
be renamed.
1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside the
United States, check the laws of your country in addition to the terms
of this agreement before downloading, copying, displaying,
performing, distributing or creating derivative works based on this
work or any other Project Gutenberg™ work. The Foundation makes
no representations concerning the copyright status of any work in
any country other than the United States.
• You pay a royalty fee of 20% of the gross profits you derive from
the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information
about donations to the Project Gutenberg Literary Archive
Foundation.”
• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.
1.F.
1.F.4. Except for the limited right of replacement or refund set forth in
paragraph 1.F.3, this work is provided to you ‘AS-IS’, WITH NO
OTHER WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR ANY PURPOSE.
Please check the Project Gutenberg web pages for current donation
methods and addresses. Donations are accepted in a number of
other ways including checks, online payments and credit card
donations. To donate, please visit: www.gutenberg.org/donate.
Most people start at our website which has the main PG search
facility: www.gutenberg.org.