Refine
Has Fulltext
- yes (17)
Is part of the Bibliography
- yes (17)
Year of publication
Document Type
- Doctoral Thesis (17)
Language
- English (17) (remove)
Keywords
- Optimierung (17) (remove)
Institute
This thesis, first, is devoted to the theoretical and numerical investigation of an augmented Lagrangian method for the solution of optimization problems with geometric constraints, subsequently, as well as constrained structured optimization problems featuring a composite objective function and set-membership constraints. It is then concerned to convergence and rate-of-convergence analysis of proximal gradient methods for the composite optimization problems in the presence of the Kurdyka--{\L}ojasiewicz property without global Lipschitz assumption.
Optimization problems with composite functions deal with the minimization of the sum
of a smooth function and a convex nonsmooth function. In this thesis several numerical
methods for solving such problems in finite-dimensional spaces are discussed, which are
based on proximity operators.
After some basic results from convex and nonsmooth analysis are summarized, a first-order
method, the proximal gradient method, is presented and its convergence properties are
discussed in detail. Known results from the literature are summarized and supplemented by
additional ones. Subsequently, the main part of the thesis is the derivation of two methods
which, in addition, make use of second-order information and are based on proximal Newton
and proximal quasi-Newton methods, respectively. The difference between the two methods
is that the first one uses a classical line search, while the second one uses a regularization
parameter instead. Both techniques lead to the advantage that, in contrast to many similar
methods, in the respective detailed convergence analysis global convergence to stationary
points can be proved without any restricting precondition. Furthermore, comprehensive
results show the local convergence properties as well as convergence rates of these algorithms,
which are based on rather weak assumptions. Also a method for the solution of the arising
proximal subproblems is investigated.
In addition, the thesis contains an extensive collection of application examples and a detailed
discussion of the related numerical results.
This thesis is concerned with applying the total variation (TV) regularizer to surfaces and different types of shape optimization problems. The resulting problems are challenging since they suffer from the non-differentiability of the TV-seminorm, but unlike most other priors it favors piecewise constant solutions, which results in piecewise flat geometries for shape optimization problems.The first part of this thesis deals with an analogue of the TV image reconstruction approach [Rudin, Osher, Fatemi (Physica D, 1992)] for images on smooth surfaces. A rigorous analytical framework is developed for this model and its Fenchel predual, which is a quadratic optimization problem with pointwise inequality constraints on the surface. A function space interior point method is proposed to solve it. Afterwards, a discrete variant (DTV) based on a nodal quadrature formula is defined for piecewise polynomial, globally discontinuous and continuous finite element functions on triangulated surface meshes. DTV has favorable properties, which include a convenient dual representation. Next, an analogue of the total variation prior for the normal vector field along the boundary of smooth shapes in 3D is introduced. Its analysis is based on a differential geometric setting in which the unit normal vector is viewed as an element of the two-dimensional sphere manifold. Shape calculus is used to characterize the relevant derivatives and an variant of the split Bregman method for manifold valued functions is proposed. This is followed by an extension of the total variation prior for the normal vector field for piecewise flat surfaces and the previous variant of split Bregman method is adapted. Numerical experiments confirm that the new prior favours polyhedral shapes.
ADMM-Type Methods for Optimization and Generalized Nash Equilibrium Problems in Hilbert Spaces
(2020)
This thesis is concerned with a certain class of algorithms for the solution of constrained optimization problems and generalized Nash equilibrium problems in Hilbert spaces. This class of algorithms is inspired by the alternating direction method of multipliers (ADMM) and eliminates the constraints using an augmented Lagrangian approach. The alternating direction method consists of splitting the augmented Lagrangian subproblem into smaller and more easily manageable parts.
Before the algorithms are discussed, a substantial amount of background material, including the theory of Banach and Hilbert spaces, fixed-point iterations as well as convex and monotone set-valued analysis, is presented. Thereafter, certain optimization problems and generalized Nash equilibrium problems are reformulated and analyzed using variational inequalities and set-valued mappings. The analysis of the algorithms developed in the course of this thesis is rooted in these reformulations as variational inequalities and set-valued mappings.
The first algorithms discussed and analyzed are one weakly and one strongly convergent ADMM-type algorithm for convex, linearly constrained optimization. By equipping the associated Hilbert space with the correct weighted scalar product, the analysis of these two methods is accomplished using the proximal point method and the Halpern method.
The rest of the thesis is concerned with the development and analysis of ADMM-type algorithms for generalized Nash equilibrium problems that jointly share a linear equality constraint. The first class of these algorithms is completely parallelizable and uses a forward-backward idea for the analysis, whereas the second class of algorithms can be interpreted as a direct extension of the classical ADMM-method to generalized Nash equilibrium problems.
At the end of this thesis, the numerical behavior of the discussed algorithms is demonstrated on a collection of examples.
This thesis is concerned with the solution of control and state constrained optimal control problems, which are governed by elliptic partial differential equations. Problems of this type are challenging since they suffer from the low regularity of the multiplier corresponding to the state constraint. Applying an augmented Lagrangian method we overcome these difficulties by working with multiplier approximations in $L^2(\Omega)$. For each problem class, we introduce the solution algorithm, carry out a thoroughly convergence analysis and illustrate our theoretical findings with numerical examples.
The thesis is divided into two parts. The first part focuses on classical PDE constrained optimal control problems. We start by studying linear-quadratic objective functionals, which include the standard tracking type term and an additional regularization term as well as the case, where the regularization term is replaced by an $L^1(\Omega)$-norm term, which makes the problem ill-posed. We deepen our study of the augmented Lagrangian algorithm by examining the more complicated class of optimal control problems that are governed by a semilinear partial differential equation.
The second part investigates the broader class of multi-player control problems. While the examination of jointly convex generalized Nash equilibrium problems (GNEP) is a simple extension of the linear elliptic optimal control case, the complexity is increased significantly for pure GNEPs. The existence of solutions of jointly convex GNEPs is well-studied. However, solution algorithms may suffer from non-uniqueness of solutions. Therefore, the last part of this thesis is devoted to the analysis of the uniqueness of normalized equilibria.
Maps are the main tool to represent geographical information. Users often zoom in and out to access maps at different scales. Continuous map generalization tries to make the changes between different scales smooth, which is essential to provide users with comfortable zooming experience.
In order to achieve continuous map generalization with high quality, we optimize some important aspects of maps. In this book, we have used optimization in the generalization of land-cover areas, administrative boundaries, buildings, and coastlines. According to our experiments, continuous map generalization indeed benefits from optimization.
The Software Defined Networking (SDN) paradigm offers network operators numerous improvements in terms of flexibility, scalability, as well as cost efficiency and vendor independence. However, in order to maximize the benefit from these features, several new challenges in areas such as management and orchestration need to be addressed. This dissertation makes contributions towards three key topics from these areas.
Firstly, we design, implement, and evaluate two multi-objective heuristics for the SDN controller placement problem. Secondly, we develop and apply mechanisms for automated decision making based on the Pareto frontiers that are returned by the multi-objective optimizers. Finally, we investigate and quantify the performance benefits for the SDN control plane that can be achieved by integrating information from external entities such as Network Management Systems (NMSs) into the control loop. Our evaluation results demonstrate the impact of optimizing various parameters of softwarized networks at different levels and are used to derive guidelines for an efficient operation.
Lagrange Multiplier Methods for Constrained Optimization and Variational Problems in Banach Spaces
(2018)
This thesis is concerned with a class of general-purpose algorithms for constrained minimization problems, variational inequalities, and quasi-variational inequalities in Banach spaces.
A substantial amount of background material from Banach space theory, convex analysis, variational analysis, and optimization theory is presented, including some results which are refinements of those existing in the literature. This basis is used to formulate an augmented Lagrangian algorithm with multiplier safeguarding for the solution of constrained optimization problems in Banach spaces. The method is analyzed in terms of local and global convergence, and many popular problem classes such as nonlinear programming, semidefinite programming, and function space optimization are shown to be included as special cases of the general setting.
The algorithmic framework is then extended to variational and quasi-variational inequalities, which include, by extension, Nash and generalized Nash equilibrium problems. For these problem classes, the convergence is analyzed in detail. The thesis then presents a rich collection of application examples for all problem classes, including implementation details and numerical results.
In this work, multi-particle quantum optimal control problems are studied in the framework of time-dependent density functional theory (TDDFT).
Quantum control problems are of great importance in both fundamental research and application of atomic and molecular systems. Typical applications are laser induced chemical reactions, nuclear magnetic resonance experiments, and quantum computing.
Theoretically, the problem of how to describe a non-relativistic system of multiple particles is solved by the Schrödinger equation (SE). However, due to the exponential increase in numerical complexity with the number of particles, it is impossible to directly solve the Schrödinger equation for large systems of interest. An efficient and successful approach to overcome this difficulty is the framework of TDDFT and the use of the time-dependent Kohn-Sham (TDKS) equations therein.
This is done by replacing the multi-particle SE with a set of nonlinear single-particle Schrödinger equations that are coupled through an additional potential.
Despite the fact that TDDFT is widely used for physical and quantum chemical calculation and software packages for its use are readily available, its mathematical foundation is still under active development and even fundamental issues remain unproven today.
The main purpose of this thesis is to provide a consistent and rigorous setting for the TDKS equations and of the related optimal control problems.
In the first part of the thesis, the framework of density functional theory (DFT) and TDDFT are introduced. This includes a detailed presentation of the different functional sets forming DFT. Furthermore, the known equivalence of the TDKS system to the original SE problem is further discussed.
To implement the TDDFT framework for multi-particle computations, the TDKS equations provide one of the most successful approaches nowadays. However, only few mathematical results concerning these equations are available and these results do not cover all issues that arise in the formulation of optimal control problems governed by the TDKS model.
It is the purpose of the second part of this thesis to address these issues such as higher regularity of TDKS solutions and the case of weaker requirements on external (control) potentials that are instrumental for the formulation of well-posed TDKS control problems. For this purpose, in this work, existence and uniqueness of TDKS solutions are investigated in the Galerkin framework and using energy estimates for the nonlinear TDKS equations.
In the third part of this thesis, optimal control problems governed by the TDKS model are formulated and investigated. For this purpose, relevant cost functionals that model the purpose of the control are discussed.
Henceforth, TDKS control problems result from the requirement of optimising the given cost functionals subject to the differential constraint given by the TDKS equations. The analysis of these problems is novel and represents one of the main contributions of the present thesis.
In particular, existence of minimizers is proved and their characterization by TDKS optimality systems is discussed in detail.
To this end, Fréchet differentiability of the TDKS model and of the cost functionals is addressed considering \(H^1\) cost of the control.
This part is concluded by deriving the reduced gradient in the \(L^2\) and \(H^1\) inner product.
While the \(L^2\) optimization is widespread in the literature, the choice of the \(H^1\) gradient is motivated in this work by theoretical consideration and by resulting numerical advantages.
The last part of the thesis is devoted to the numerical approximation of the TDKS optimality systems and to their solution by gradient-based optimization techniques.
For the former purpose, Strang time-splitting pseudo-spectral schemes are discussed including a review of some recent theoretical estimates for these schemes and a numerical validation of these estimates.
For the latter purpose, nonlinear (projected) conjugate gradient methods are implemented and are used to validate the theoretical analysis of this thesis with results of numerical experiments with different cost functional settings.
Today's Internet is no longer only controlled by a single stakeholder, e.g. a standard body or a telecommunications company.
Rather, the interests of a multitude of stakeholders, e.g. application developers, hardware vendors, cloud operators, and network operators, collide during the development and operation of applications in the Internet.
Each of these stakeholders considers different KPIs to be important and attempts to optimise scenarios in its favour.
This results in different, often opposing views and can cause problems for the complete network ecosystem.
One example of such a scenario are Signalling Storms in the mobile Internet, with one of the largest occurring in Japan in 2012 due to the release and high popularity of a free instant messaging application.
The network traffic generated by the application caused a high number of connections to the Internet being established and terminated.
This resulted in a similarly high number of signalling messages in the mobile network, causing overload and a loss of service for 2.5 million users over 4 hours.
While the network operator suffers the largest impact of this signalling overload, it does not control the application.
Thus, the network operator can not change the application traffic characteristics to generate less network signalling traffic.
The stakeholders who could prevent, or at least reduce, such behaviour, i.e. application developers or hardware vendors, have no direct benefit from modifying their products in such a way.
This results in a clash of interests which negatively impacts the network performance for all participants.
The goal of this monograph is to provide an overview over the complex structures of stakeholder relationships in today's Internet applications in mobile networks.
To this end, we study different scenarios where such interests clash and suggest methods where tradeoffs can be optimised for all participants.
If such an optimisation is not possible or attempts at it might lead to adverse effects, we discuss the reasons.