In der Motorenentwicklung treten Optimierungsprobleme auf, die sich nur schwer mit klassischen Methoden der Optimierung lösen lassen. Daher untersucht diese Arbeit nichtlineare Verfahren der ein- und multikriteriellen Optimierung, die unter Einhaltung nichtlinearer Nebenbedingungen mit relativ wenigen Funktionswertberechnungen in der Lage sind globale Extrema zu finden. Vorgestellt werden ein Genetischer Algorithmus und zwei Ersatzmodell-gestützte Optimierungsverfahren, die in das Optimierungsmodul der IAV EngineeringToolbox integriert wurden. Die Tauglichkeit der Algorithmen wurde an technischen Beispielen (1D-Strömungssimulation, Kettentriebsoptimierung), sowie an geeigneten Testfunktionen überprüft.

For ensuring network survivability in case of single component failures many routing protocols provide a primary and a back up routing path for each origin destination pair. We address the problem of selecting these paths such that in the event of multiple failures, occuring with given probabilities, the total loss in routable demand due to both paths being intersected is small with high probability. We present a chance constraint model and solution approaches based on an explicit integer programming formulation, a robust formulation and a cutting plane approach that yield reasonably good solutions assuming that the failures are caused by at most two elementary events, which may each affect several network components.

There is a growing societal concern about excessive water and fertilizer use in agricultural systems. High water productivity while maintaining high crop yields can be achieved with appropriate irrigation scheduling. Moreover, freshwater pollution through nitrogen (N) leaching due to the widespread use of N fertilizers demands for an efficient N fertilization management. However, sustainable crop management requires good knowledge of soil water and N dynamics as well as of crop water and N demand. Crop growth models, which describe physical and physiological processes of crop growth as well as water and matter transport, are considered as powerful tools to assist in the optimization of irrigation and fertilization management. It is of a general nature that the reliability of simulation based predictions depends on the quality and quantity of the data used for model calibration and validation which can be obtained e.g. in field experiments. A lack of data or low data quality for model calibration and validation may cause low performance and large uncertainties in simulation results. The large number of model parameters to be calibrated requires appropriate calibration methods and a sequential calibration strategy. Moreover, a simulation based planning of the field design saves costs and expenditure while supporting maximal outcomes of experiments. An adjustment of crop growth modeling and experiments is required for model improvement and development to reliably predict crop growth and to generalize predicted results. In this research study, a new approach for simulation based optimal experimental design was developed aiming to integrate simulation models, experiments, and optimization methods in one framework for optimal and sustainable irrigation and N fertilization management. The approach is composed of three steps: 1. The preprocessing consists of the calibration and validation of the crop growth model based on existing experimental data, the generation of long time-series of climate data, and the determination of the optimal irrigation control. 2. The implementation comprises the determination and experimental application of the simulation based optimized deficit irrigation and N fertilization schedules and an appropriate experimental data collection. 3. The postprocessing includes the evaluation of the experimental results namely observed yield, water productivity (WP), nitrogen use efficiency (NUE), and economic aspects, as well as a model evaluation. Five main tools were applied within the new approach: an algorithm for inverse model parametrization, a crop growth model for simulating crop growth, water balance and N balance, an optimization algorithm for deficit irrigation and N fertilization scheduling, and a stochastic weather generator. Furthermore, a water flow model was used to determine the optimal irrigation control functions and for simulation based estimation of the optimal field design. The approach was implemented within three case studies presented in this work. The new approach combines crop growth modeling and experiments with stochastic optimization. It contributes to a successful application of crop growth modeling based on an appropriate experimental data collection. The presented model calibration and validation procedure using the collected data facilitates reliable predictions. The stochastic optimization framework for deficit irrigation and N fertilization scheduling proved to be a powerful tool to enhance yield, WP, NUE and profit.:Contents Nomenclature ..............................................................................................................................xii 1 Introduction..................................................................................................................................1 I Fundamentals and literature review ........................................................................................5 2 Water productivity in crop production ....................................................................................7 2.1 Water productivity .................................................................................................................7 2.2 Increase of crop yield ..........................................................................................................9 2.3 Irrigation ...............................................................................................................................10 2.3.1 Irrigation methods ...........................................................................................................10 2.3.2 Irrigation scheduling and irrigation control ................................................................11 2.3.3 The influence of the field design on profitability .......................................................12 2.4 The concept of controlled deficit irrigation ...................................................................14 3 Nitrogen use efficiency in crop production .........................................................................17 3.1 Nitrogen use efficiency ....................................................................................................18 3.2 N fertilization management .............................................................................................18 3.3 Combination of controlled deficit irrigation and deficit N fertilization ......................19 4 Crop growth modeling ............................................................................................................21 4.1 Physiological crop growth models ..................................................................................21 4.1.1 Model description of SVAT model Daisy ....................................................................22 4.1.2 Model description of crop growth model Pilote .........................................................24 4.2 Optimal experimental design for model parametrization ...........................................25 4.2.1 Experimental design ......................................................................................................25 4.2.2 Model parameter estimation ........................................................................................26 4.2.3 Model parameter estimation based on greenhouse data .......................................27 5 Irrigation and N fertilization scheduling ..............................................................................29 5.1 Irrigation scheduling .........................................................................................................29 5.2 N fertilization scheduling .................................................................................................30 5.3 Combination of irrigation and N fertilization scheduling ............................................30 II New approach for simulation based optimal experimental design ................................33 6 Preprocessing steps ...............................................................................................................37 6.1 Model parametrization and assessment .......................................................................37 6.1.1 Calibration of the soil parameters ...............................................................................38 6.1.2 Calibration of the plant parameters ............................................................................39 6.1.3 Model assessment .........................................................................................................41 6.1.4 Preliminary simulations for an optimal experimental layout ..................................43 6.2 Generation of long time-series of climate data ............................................................44 6.3 Determination of the optimal irrigation control functions ...........................................44 7 Stochastic optimization framework ......................................................................................47 7.1 Stochastic optimization framework .................................................................................47 7.1.1 Optimization algorithm ...................................................................................................47 7.1.2 Generation of the crop water (nitrogen) production functions ................................48 7.1.3 Application of the stochastic optimization framework ..............................................48 7.1.4 Crop growth model requirements ................................................................................49 8 Data collection during the experimentation .......................................................................51 9 Postprocessing steps .............................................................................................................55 9.1 Evaluation of the experimental results ...........................................................................55 9.1.1 Yield and total dry matter ..............................................................................................55 9.1.2 Water productivity and nitrogen use efficiency .........................................................55 9.1.3 Economic aspects ..........................................................................................................55 9.1.4 Evaluation of the model results ....................................................................................56 III Application of the new approach to three case studies ...................................................57 10 Evaluation of model transferability ....................................................................................59 10.1 Objectives and summary ................................................................................................59 10.2 Experimental site and experimental setup .................................................................61 10.3 Data collection during the experimentation ................................................................63 10.4 Calibration and validation of the model parameters .................................................63 10.4.1 Model setup and soil parametrization ......................................................................64 10.4.2 Plant parameter calibration and validation .............................................................67 10.5 Application of the stochastic optimization framework ...............................................75 10.5.1 Generation of the climate data ...................................................................................75 10.5.2 Estimation of the yield potential of wheat ................................................................75 10.5.3 Estimation of the water productivity potential of barley .........................................77 10.6 Discussion and conclusions ..........................................................................................81 11 Real-time irrigation scheduling ..........................................................................................83 11.1 Objectives and summary ................................................................................................83 11.2 Experimental site and field design ...............................................................................85 11.3 Data collection during the experiment ........................................................................86 11.4 Calibration and setup of the crop growth model Pilote .............................................87 11.5 Derivation of optimal irrigation control functions for different drip line spacings 88 11.5.1 Initial Hydrus 2D/3D simulations ...............................................................................88 11.5.2 Determination of the irrigation control functions .....................................................89 11.5.3 Verifying measurements ..............................................................................................92 11.6 Real-time deficit irrigation scheduling .........................................................................93 11.7 Evaluation of the experimental results .........................................................................96 11.7.1 Crop yields .....................................................................................................................96 11.7.2 Water productivity .........................................................................................................97 11.7.3 Prognostic simulations ................................................................................................98 11.7.4 Economic aspects ........................................................................................................99 11.8 Discussion and conclusions ........................................................................................100 12 Multicriterial optimization...................................................................................................103 12.1 Objectives and summary .............................................................................................103 12.2 Experimental site and experimental setup ...............................................................105 12.3 Data collection during the experiment ......................................................................105 12.4 Experimental layout ......................................................................................................106 12.5 Calibration and validation of the model parameters ..............................................107 12.5.1 Calibration of the soil parameters ...........................................................................107 12.5.2 Calibration and validation of the plant parameters .............................................107 12.5.3 Setup of SVAT model Daisy .....................................................................................108 12.6 Generation of the climate data ....................................................................................109 12.7 Optimized irrigation and N fertilization scheduling .................................................109 12.8 Evaluation of the experimental results .......................................................................111 12.8.1 Observed plant variables and weather data .........................................................111 12.8.2 Water productivities and nitrogen use efficiencies ...............................................111 12.8.3 Chlorophyll Meter values ..........................................................................................112 12.8.4 Recalculation of soil parameters .............................................................................113 12.9 Postprocessing simulations of yield and water and N dynamics..........................114 12.9.1 Yield predictions using Daisy 1D ............................................................................114 12.9.2 Yield predictions using Daisy 2D ............................................................................119 12.10 Discussion and conclusions .....................................................................................121 IV General discussion, conclusions and outlook ...............................................................123 13 General discussion ............................................................................................................125 14 General conclusions and outlook ....................................................................................133 Appendix ....................................................................................................................................134 A Tables and Figures ...............................................................................................................137 B Model setup and weather files ...........................................................................................145 List of Tables .............................................................................................................................153 List of Figures ............................................................................................................................153 References ................................................................................................................................159 / In der heutigen Gesellschaft gibt es zunehmend Bedenken gegenüber übermäßigem Wasser- und Düngereinsatz in der Landwirtschaft. Eine hohe Wasserproduktivität kann jedoch durch geeignete Bewässerungspläne mit hohen landwirtschaftlichen Erträgen in Einklang gebracht werden. Die mit der weitverbreiteten Stickstoffdüngung einhergehende Gewässerbelastung aufgrund von Stickstoffauswaschung erfordert zudem ein effizientes Stickstoffmanagement. Eine entsprechende ressourceneffiziente Landbewirtschaftung bedarf präzise Kenntnisse der Bodenwasser- und Stickstoffdynamiken sowie des Pflanzenwasser- und Stickstoffbedarfs. Als leistungsfähige Werkzeuge zur Unterstützung bei der Optimierung von Bewässerungs-und Düngungsplänen werden Pflanzenwachstumsmodelle eingesetzt, welche die physischen und physiologischen Prozesse des Pflanzenwachstums sowie die physikalischen Prozesse des Wasser- und Stofftransports abbilden. Hierbei hängt die Zuverlässigkeit dieser simulationsbasierten Vorhersagen von der Qualität und Quantität der bei der Modellkalibrierung und -validierung verwendeten Daten ab, welche beispielsweise in Feldversuchen erfasst werden. Fehlende Daten oder Daten mangelhafter Qualität bei der Modellkalibrierung und -validierung führen zu unzuverlässigen Simulationsergebnissen und großen Unsicherheiten bei der Vorhersage. Die große Anzahl an zu kalibrierenden Parametern erfordert zudem geeignete Kalibrierungsmethoden sowie eine sequenzielle Kalibrierungsstrategie. Darüber hinaus kann eine simulationsbasierte Planung des Versuchsdesigns Kosten und Aufwand reduzieren und zu weiteren experimentellen Erkenntnissen führen. Die Abstimmung von Pflanzenwachstumsmodellen und Versuchen ist zudem für die Modellentwicklung und -verbesserung sowie für eine Verallgemeinerung von Simulationsergebnissen unabdingbar. Im Rahmen dieser Arbeit wurde ein neuer Ansatz für ein simulationsbasiertes optimales Versuchsdesign entwickelt. Ziel war es, Simulationsmodelle, Versuche und Optimierungsmethoden in einem Ansatz für optimales und nachhaltiges Bewässerungs- und Düngungsmanagement zu integrieren. Der Ansatz besteht aus drei Schritten: 1. Die Vorbereitungsphase beinhaltet die auf existierenden Versuchsdaten basierende Kalibrierung und Validierung des Pflanzenwachstumsmodells, die Generierung von Klimazeitreihen und die Bestimmung der optimalen Bewässerungssteuerung. 2. Die Durchführungsphase setzt sich aus der Erstellung und experimentellen Anwendung der simulationsbasierten optimierten Defizitbewässerungs- und Stickstoffdüngungspläne und der Erfassung der relevanten Versuchsdaten zusammen. 3. Die Auswertungsphase schließt eine Evaluierung der Versuchsergebnisse anhand ermittelter Erträge, Wasserproduktivitäten (WP), Stickstoffnutzungseffizienzen (NUE) und ökonomischer Aspekte, sowie eine Modellevaluierung ein. In dem neuen Ansatz kamen im Wesentlichen folgende fünf Werkzeuge zur Anwendung: Ein Algorithmus zur inversen Modellparametrisierung, ein Pflanzenwachstumsmodell, welches das Pflanzenwachstum sowie die Wasser- und Stickstoffbilanzen abbildet, ein evolutionärer Optimierungsalgorithmus für die Generierung von defizitären Bewässerungs- und Stickstoffplänen und ein stochastischer Wettergenerator. Zudem diente ein Bodenwasserströmungsmodell der Ermittlung der optimalen Bewässerungssteuerung und der simulationsbasierten Optimierung des Versuchsdesigns. Der in dieser Arbeit vorgestellte Ansatz wurde in drei Fallbeispielen angewandt. Der neue Ansatz kombiniert Pflanzenwachstumsmodellierung und Experimente mit stochastischer Optimierung. Er leistet einen Beitrag zu einer erfolgreichen Pflanzenwachstumsmodellierung basierend auf der Erfassung relevanter Versuchsdaten. Die vorgestellte Modellkalibrierung und -validierung unter Verwendung der erfassten Versuchsdaten trug wesentlich zu zuverlässigen Simulationsergebnissen bei. Darüber hinaus stellt die hier vorgestellte stochastische Optimierung von defizitären Bewässerungs- und Stickstoffplänen ein leistungsfähiges Werkzeug dar, um Erträge, WP, NUE und den Profit zu erhöhen.:Contents Nomenclature ..............................................................................................................................xii 1 Introduction..................................................................................................................................1 I Fundamentals and literature review ........................................................................................5 2 Water productivity in crop production ....................................................................................7 2.1 Water productivity .................................................................................................................7 2.2 Increase of crop yield ..........................................................................................................9 2.3 Irrigation ...............................................................................................................................10 2.3.1 Irrigation methods ...........................................................................................................10 2.3.2 Irrigation scheduling and irrigation control ................................................................11 2.3.3 The influence of the field design on profitability .......................................................12 2.4 The concept of controlled deficit irrigation ...................................................................14 3 Nitrogen use efficiency in crop production .........................................................................17 3.1 Nitrogen use efficiency ....................................................................................................18 3.2 N fertilization management .............................................................................................18 3.3 Combination of controlled deficit irrigation and deficit N fertilization ......................19 4 Crop growth modeling ............................................................................................................21 4.1 Physiological crop growth models ..................................................................................21 4.1.1 Model description of SVAT model Daisy ....................................................................22 4.1.2 Model description of crop growth model Pilote .........................................................24 4.2 Optimal experimental design for model parametrization ...........................................25 4.2.1 Experimental design ......................................................................................................25 4.2.2 Model parameter estimation ........................................................................................26 4.2.3 Model parameter estimation based on greenhouse data .......................................27 5 Irrigation and N fertilization scheduling ..............................................................................29 5.1 Irrigation scheduling .........................................................................................................29 5.2 N fertilization scheduling .................................................................................................30 5.3 Combination of irrigation and N fertilization scheduling ............................................30 II New approach for simulation based optimal experimental design ................................33 6 Preprocessing steps ...............................................................................................................37 6.1 Model parametrization and assessment .......................................................................37 6.1.1 Calibration of the soil parameters ...............................................................................38 6.1.2 Calibration of the plant parameters ............................................................................39 6.1.3 Model assessment .........................................................................................................41 6.1.4 Preliminary simulations for an optimal experimental layout ..................................43 6.2 Generation of long time-series of climate data ............................................................44 6.3 Determination of the optimal irrigation control functions ...........................................44 7 Stochastic optimization framework ......................................................................................47 7.1 Stochastic optimization framework .................................................................................47 7.1.1 Optimization algorithm ...................................................................................................47 7.1.2 Generation of the crop water (nitrogen) production functions ................................48 7.1.3 Application of the stochastic optimization framework ..............................................48 7.1.4 Crop growth model requirements ................................................................................49 8 Data collection during the experimentation .......................................................................51 9 Postprocessing steps .............................................................................................................55 9.1 Evaluation of the experimental results ...........................................................................55 9.1.1 Yield and total dry matter ..............................................................................................55 9.1.2 Water productivity and nitrogen use efficiency .........................................................55 9.1.3 Economic aspects ..........................................................................................................55 9.1.4 Evaluation of the model results ....................................................................................56 III Application of the new approach to three case studies ...................................................57 10 Evaluation of model transferability ....................................................................................59 10.1 Objectives and summary ................................................................................................59 10.2 Experimental site and experimental setup .................................................................61 10.3 Data collection during the experimentation ................................................................63 10.4 Calibration and validation of the model parameters .................................................63 10.4.1 Model setup and soil parametrization ......................................................................64 10.4.2 Plant parameter calibration and validation .............................................................67 10.5 Application of the stochastic optimization framework ...............................................75 10.5.1 Generation of the climate data ...................................................................................75 10.5.2 Estimation of the yield potential of wheat ................................................................75 10.5.3 Estimation of the water productivity potential of barley .........................................77 10.6 Discussion and conclusions ..........................................................................................81 11 Real-time irrigation scheduling ..........................................................................................83 11.1 Objectives and summary ................................................................................................83 11.2 Experimental site and field design ...............................................................................85 11.3 Data collection during the experiment ........................................................................86 11.4 Calibration and setup of the crop growth model Pilote .............................................87 11.5 Derivation of optimal irrigation control functions for different drip line spacings 88 11.5.1 Initial Hydrus 2D/3D simulations ...............................................................................88 11.5.2 Determination of the irrigation control functions .....................................................89 11.5.3 Verifying measurements ..............................................................................................92 11.6 Real-time deficit irrigation scheduling .........................................................................93 11.7 Evaluation of the experimental results .........................................................................96 11.7.1 Crop yields .....................................................................................................................96 11.7.2 Water productivity .........................................................................................................97 11.7.3 Prognostic simulations ................................................................................................98 11.7.4 Economic aspects ........................................................................................................99 11.8 Discussion and conclusions ........................................................................................100 12 Multicriterial optimization...................................................................................................103 12.1 Objectives and summary .............................................................................................103 12.2 Experimental site and experimental setup ...............................................................105 12.3 Data collection during the experiment ......................................................................105 12.4 Experimental layout ......................................................................................................106 12.5 Calibration and validation of the model parameters ..............................................107 12.5.1 Calibration of the soil parameters ...........................................................................107 12.5.2 Calibration and validation of the plant parameters .............................................107 12.5.3 Setup of SVAT model Daisy .....................................................................................108 12.6 Generation of the climate data ....................................................................................109 12.7 Optimized irrigation and N fertilization scheduling .................................................109 12.8 Evaluation of the experimental results .......................................................................111 12.8.1 Observed plant variables and weather data .........................................................111 12.8.2 Water productivities and nitrogen use efficiencies ...............................................111 12.8.3 Chlorophyll Meter values ..........................................................................................112 12.8.4 Recalculation of soil parameters .............................................................................113 12.9 Postprocessing simulations of yield and water and N dynamics..........................114 12.9.1 Yield predictions using Daisy 1D ............................................................................114 12.9.2 Yield predictions using Daisy 2D ............................................................................119 12.10 Discussion and conclusions .....................................................................................121 IV General discussion, conclusions and outlook ...............................................................123 13 General discussion ............................................................................................................125 14 General conclusions and outlook ....................................................................................133 Appendix ....................................................................................................................................134 A Tables and Figures ...............................................................................................................137 B Model setup and weather files ...........................................................................................145 List of Tables .............................................................................................................................153 List of Figures ............................................................................................................................153 References ................................................................................................................................159

For ensuring network survivability in case of single component failures many routing protocols provide a primary and a back up routing path for each origin destination pair. We address the problem of selecting these paths such that in the event of multiple failures, occuring with given probabilities, the total loss in routable demand due to both paths being intersected is small with high probability. We present a chance constraint model and solution approaches based on an explicit integer programming formulation, a robust formulation and a cutting plane approach that yield reasonably good solutions assuming that the failures are caused by at most two elementary events, which may each affect several network components.

In der Motorenentwicklung treten Optimierungsprobleme auf, die sich nur schwer mit klassischen Methoden der Optimierung lösen lassen. Daher untersucht diese Arbeit nichtlineare Verfahren der ein- und multikriteriellen Optimierung, die unter Einhaltung nichtlinearer Nebenbedingungen mit relativ wenigen Funktionswertberechnungen in der Lage sind globale Extrema zu finden. Vorgestellt werden ein Genetischer Algorithmus und zwei Ersatzmodell-gestützte Optimierungsverfahren, die in das Optimierungsmodul der IAV EngineeringToolbox integriert wurden. Die Tauglichkeit der Algorithmen wurde an technischen Beispielen (1D-Strömungssimulation, Kettentriebsoptimierung), sowie an geeigneten Testfunktionen überprüft.

We consider a bilevel continuous knapsack problem where the leader controls the capacity of the knapsack, while the follower chooses a feasible packing maximizing his own profit. The leader’s aim is to optimize a linear objective function in the capacity and in the follower’s solution, but with respect to different item values. We address a stochastic version of this problem where the follower’s profits are uncertain from the leader’s perspective, and only a probability distribution is known. Assuming that the leader aims at optimizing the expected value of her objective function, we first observe that the stochastic problem is tractable as long as the possible scenarios are given explicitly as part of the input,which also allows to deal with general distributions using a sample average approximation. For the case of independently and uniformly distributed item values, we show that the problem is #P-hard in general, and the same is true even for evaluating the leader’s objective function. Nevertheless, we present pseudo-polynomial time algorithms for this case, running in time linear in the total size of the items.Based on this,we derive an additive approximation scheme for the general case of independently distributed item values, which runs in pseudo-polynomial time.

Diese Arbeit leistet Beiträge zu zwei Gebieten der mathematischen Programmierung: stochastische Optimierung und gemischt-ganzzahlige nichtlineare Optimierung (MINLP). Im ersten Teil erweitern wir quantitative Stetigkeitsresultate für zweistufige stochastische gemischt-ganzzahlige lineare Programme auf Situationen in denen Unsicherheit gleichzeitig in den Kosten und der rechten Seite auftritt, geben eine ausführliche Übersicht zu Dekompositionsverfahren für zwei- und mehrstufige stochastische lineare und gemischt-ganzzahlig lineare Programme, und diskutieren Erweiterungen und Kombinationen des Nested Benders Dekompositionsverfahrens und des Nested Column Generationsverfahrens für mehrstufige stochastische lineare Programme die es erlauben die Vorteile sogenannter rekombinierender Szenariobäume auszunutzen. Als eine Anwendung dieses Verfahrens betrachten wir die optimale Zeit- und Investitionsplanung für ein regionales Energiesystem unter Einbeziehung von Windenergie und Energiespeichern. Im zweiten Teil geben wir eine ausführliche Übersicht zum Stand der Technik bzgl. Algorithmen und Lösern für MINLPs und zeigen dass einige dieser Algorithmen innerhalb des constraint integer programming Softwaresystems SCIP angewendet werden können. Letzteres erlaubt uns die Verwendung schon existierender Technologien für gemischt-ganzzahlige linear Programme und constraint Programme für den linearen und diskreten Teil des Problems. Folglich konzentrieren wir uns hauptsächlich auf die Behandlung der konvexen und nichtkonvexen nichtlinearen Nebenbedingungen mittels Variablenschrankenpropagierung, äußerer Approximation und Reformulierung. In einer ausführlichen numerischen Studie untersuchen wir die Leistung unseres Ansatzes anhand von Anwendungen aus der Tagebauplanung und des Aufbaus eines Wasserverteilungssystems und mittels verschiedener Vergleichstests. Die Ergebnisse zeigen, dass SCIP ein konkurrenzfähiger Löser für MINLPs geworden ist. / This thesis contributes to two topics in mathematical programming: stochastic optimization and mixed-integer nonlinear programming (MINLP). In the first part, we extend quantitative continuity results for two-stage stochastic mixed-integer linear programs to include situations with simultaneous uncertainty in costs and right-hand side, give an extended review on decomposition algorithm for two- and multistage stochastic linear and mixed-integer linear programs, and discuss extensions and combinations of the Nested Benders Decomposition and Nested Column Generation methods for multistage stochastic linear programs to exploit the advantages of so-called recombining scenario trees. As an application of the latter, we consider the optimal scheduling and investment planning for a regional energy system including wind power and energy storages. In the second part, we give a comprehensive overview about the state-of-the-art in algorithms and solver technology for MINLPs and show that some of these algorithm can be applied within the constraint integer programming framework SCIP. The availability of the latter allows us to utilize the power of already existing mixed integer linear and constraint programming technologies to handle the linear and discrete parts of the problem. Thus, we focus mainly on the domain propagation, outer-approximation, and reformulation techniques to handle convex and nonconvex nonlinear constraints. In an extensive computational study, we investigate the performance of our approach on applications from open pit mine production scheduling and water distribution network design and on various benchmarks sets. The results show that SCIP has become a competitive solver for MINLPs.

Motivation and background The enormous amount of capabilities that every human learns throughout his life, is probably among the most remarkable and fascinating aspects of life. Learning has therefore drawn lots of interest from scientists working in very different fields like philosophy, biology, sociology, educational sciences, computer sciences and mathematics. This thesis focuses on the information theoretical and mathematical aspects of learning. We are interested in the learning process of an agent (which can be for example a human, an animal, a robot, an economical institution or a state) that interacts with its environment. Common models for this interaction are Markov decision processes (MDPs) and partially observable Markov decision processes (POMDPs). Learning is then considered to be the maximization of the expectation of a predefined reward function. In order to formulate general principles (like a formal definition of curiosity-driven learning or avoidance of unpleasant situation) in a rigorous way, it might be desirable to have a theoretical framework for the optimization of more complex functionals of the underlying process law. This might include the entropy of certain sensor values or their mutual information. An optimization of the latter quantity (also known as predictive information) has been investigated intensively both theoretically and experimentally using computer simulations by N. Ay, R. Der, K Zahedi and G. Martius. In this thesis, we develop a mathematical theory for learning in the sensorimotor loop beyond expected reward maximization. Approaches and results This thesis covers four different topics related to the theory of learning in the sensorimotor loop. First of all, we need to specify the model of an agent interacting with the environment, either with learning or without learning. This interaction naturally results in complex causal dependencies. Since we are interested in asymptotic properties of learning algorithms, it is necessary to consider infinite time horizons. It turns out that the well-understood theory of causal networks known from the machine learning literature is not powerful enough for our purpose. Therefore we extend important theorems on causal networks to infinite graphs and general state spaces using analytical methods from measure theoretic probability theory and the theory of discrete time stochastic processes. Furthermore, we prove a generalization of the strong Markov property from Markov processes to infinite causal networks. Secondly, we develop a new idea for a projected stochastic constraint optimization algorithm. Generally a discrete gradient ascent algorithm can be used to generate an iterative sequence that converges to the stationary points of a given optimization problem. Whenever the optimization takes place over a compact subset of a vector space, it is possible that the iterative sequence leaves the constraint set. One possibility to cope with this problem is to project all points to the constraint set using Euclidean best-approximation. The latter is sometimes difficult to calculate. A concrete example is an optimization over the unit ball in a matrix space equipped with operator norm. Our idea consists of a back-projection using quasi-projectors different from the Euclidean best-approximation. In the matrix example, there is another canonical way to force the iterative sequence to stay in the constraint set: Whenever a point leaves the unit ball, it is divided by its norm. For a given target function, this procedure might introduce spurious stationary points on the boundary. We show that this problem can be circumvented by using a gradient that is tailored to the quasi-projector used for back-projection. We state a general technical compatibility condition between a quasi-projector and a metric used for gradient ascent, prove convergence of stochastic iterative sequences and provide an appropriate metric for the unit-ball example. Thirdly, a class of learning problems in the sensorimotor loop is defined and motivated. This class of problems is more general than the usual expected reward maximization and is illustrated by numerous examples (like expected reward maximization, maximization of the predictive information, maximization of the entropy and minimization of the variance of a given reward function). We also provide stationarity conditions together with appropriate gradient formulas. Last but not least, we prove convergence of a stochastic optimization algorithm (as considered in the second topic) applied to a general learning problem (as considered in the third topic). It is shown that the learning algorithm converges to the set of stationary points. Among others, the proof covers the convergence of an improved version of an algorithm for the maximization of the predictive information as proposed by N. Ay, R. Der and K. Zahedi. We also investigate an application to a linear Gaussian dynamic, where the policies are encoded by the unit-ball in a space of matrices equipped with operator norm.

Motivation and background The enormous amount of capabilities that every human learns throughout his life, is probably among the most remarkable and fascinating aspects of life. Learning has therefore drawn lots of interest from scientists working in very different fields like philosophy, biology, sociology, educational sciences, computer sciences and mathematics. This thesis focuses on the information theoretical and mathematical aspects of learning. We are interested in the learning process of an agent (which can be for example a human, an animal, a robot, an economical institution or a state) that interacts with its environment. Common models for this interaction are Markov decision processes (MDPs) and partially observable Markov decision processes (POMDPs). Learning is then considered to be the maximization of the expectation of a predefined reward function. In order to formulate general principles (like a formal definition of curiosity-driven learning or avoidance of unpleasant situation) in a rigorous way, it might be desirable to have a theoretical framework for the optimization of more complex functionals of the underlying process law. This might include the entropy of certain sensor values or their mutual information. An optimization of the latter quantity (also known as predictive information) has been investigated intensively both theoretically and experimentally using computer simulations by N. Ay, R. Der, K Zahedi and G. Martius. In this thesis, we develop a mathematical theory for learning in the sensorimotor loop beyond expected reward maximization. Approaches and results This thesis covers four different topics related to the theory of learning in the sensorimotor loop. First of all, we need to specify the model of an agent interacting with the environment, either with learning or without learning. This interaction naturally results in complex causal dependencies. Since we are interested in asymptotic properties of learning algorithms, it is necessary to consider infinite time horizons. It turns out that the well-understood theory of causal networks known from the machine learning literature is not powerful enough for our purpose. Therefore we extend important theorems on causal networks to infinite graphs and general state spaces using analytical methods from measure theoretic probability theory and the theory of discrete time stochastic processes. Furthermore, we prove a generalization of the strong Markov property from Markov processes to infinite causal networks. Secondly, we develop a new idea for a projected stochastic constraint optimization algorithm. Generally a discrete gradient ascent algorithm can be used to generate an iterative sequence that converges to the stationary points of a given optimization problem. Whenever the optimization takes place over a compact subset of a vector space, it is possible that the iterative sequence leaves the constraint set. One possibility to cope with this problem is to project all points to the constraint set using Euclidean best-approximation. The latter is sometimes difficult to calculate. A concrete example is an optimization over the unit ball in a matrix space equipped with operator norm. Our idea consists of a back-projection using quasi-projectors different from the Euclidean best-approximation. In the matrix example, there is another canonical way to force the iterative sequence to stay in the constraint set: Whenever a point leaves the unit ball, it is divided by its norm. For a given target function, this procedure might introduce spurious stationary points on the boundary. We show that this problem can be circumvented by using a gradient that is tailored to the quasi-projector used for back-projection. We state a general technical compatibility condition between a quasi-projector and a metric used for gradient ascent, prove convergence of stochastic iterative sequences and provide an appropriate metric for the unit-ball example. Thirdly, a class of learning problems in the sensorimotor loop is defined and motivated. This class of problems is more general than the usual expected reward maximization and is illustrated by numerous examples (like expected reward maximization, maximization of the predictive information, maximization of the entropy and minimization of the variance of a given reward function). We also provide stationarity conditions together with appropriate gradient formulas. Last but not least, we prove convergence of a stochastic optimization algorithm (as considered in the second topic) applied to a general learning problem (as considered in the third topic). It is shown that the learning algorithm converges to the set of stationary points. Among others, the proof covers the convergence of an improved version of an algorithm for the maximization of the predictive information as proposed by N. Ay, R. Der and K. Zahedi. We also investigate an application to a linear Gaussian dynamic, where the policies are encoded by the unit-ball in a space of matrices equipped with operator norm.

In dieser Arbeit beschäftigen wir uns mit dem Wasserstein Baryzentrumproblem diskreter Wahrscheinlichkeitsmaße sowie mit dem population Wasserstein Baryzentrumproblem gegeben von a Fréchet Mittelwerts von der rechnerischen und statistischen Seiten. Der statistische Fokus liegt auf der Schätzung der Stichprobengröße von Maßen zur Berechnung einer Annäherung des Fréchet Mittelwerts (Baryzentrum) der Wahrscheinlichkeitsmaße mit einer bestimmten Genauigkeit. Für empirische Risikominimierung (ERM) wird auch die Frage der Regularisierung untersucht zusammen mit dem Vorschlag einer neuen Regularisierung, die zu den besseren Komplexitätsgrenzen im Vergleich zur quadratischen Regularisierung beiträgt. Der Rechenfokus liegt auf der Entwicklung von dezentralen Algorithmen zurBerechnung von Wasserstein Baryzentrum: duale Algorithmen und Sattelpunktalgorithmen. Die Motivation für duale Optimierungsmethoden ist geschlossene Formen für die duale Formulierung von entropie-regulierten Wasserstein Distanz und ihren Derivaten, während, die primale Formulierung nur in einigen Fällen einen Ausdruck in geschlossener Form hat, z.B. für Gaußsches Maß. Außerdem kann das duale Orakel, das den Gradienten der dualen Darstellung für die entropie-regulierte Wasserstein Distanz zurückgibt, zu einem günstigeren Preis berechnet werden als das primale Orakel, das den Gradienten der (entropie-regulierten) Wasserstein Distanz zurückgibt. Die Anzahl der dualen Orakel rufe ist in diesem Fall ebenfalls weniger, nämlich die Quadratwurzel der Anzahl der primalen Orakelrufe. Im Gegensatz zum primalen Zielfunktion, hat das duale Zielfunktion Lipschitz-stetig Gradient aufgrund der starken Konvexität regulierter Wasserstein Distanz. Außerdem untersuchen wir die Sattelpunktformulierung des (nicht regulierten) Wasserstein Baryzentrum, die zum Bilinearsattelpunktproblem führt. Dieser Ansatz ermöglicht es uns auch, optimale Komplexitätsgrenzen zu erhalten, und kann einfach in einer dezentralen Weise präsentiert werden. / In this thesis, we consider the Wasserstein barycenter problem of discrete probability measures as well as the population Wasserstein barycenter problem given by a Fréchet mean from computational and statistical sides. The statistical focus is estimating the sample size of measures needed to calculate an approximation of a Fréchet mean (barycenter) of probability distributions with a given precision. For empirical risk minimization approaches, the question of the regularization is also studied along with proposing a new regularization which contributes to the better complexity bounds in comparison with the quadratic regularization. The computational focus is developing decentralized algorithms for calculating Wasserstein barycenters: dual algorithms and saddle point algorithms. The motivation for dual approaches is closed-forms for the dual formulation of entropy-regularized Wasserstein distances and their derivatives, whereas the primal formulation has a closed-form expression only in some cases, e.g., for Gaussian measures.Moreover, the dual oracle returning the gradient of the dual representation forentropy-regularized Wasserstein distance can be computed for a cheaper price in comparison with the primal oracle returning the gradient of the (entropy-regularized) Wasserstein distance. The number of dual oracle calls in this case will be also less, i.e., the square root of the number of primal oracle calls. Furthermore, in contrast to the primal objective, the dual objective has Lipschitz continuous gradient due to the strong convexity of regularized Wasserstein distances. Moreover, we study saddle-point formulation of the non-regularized Wasserstein barycenter problem which leads to the bilinear saddle-point problem. This approach also allows us to get optimal complexity bounds and it can be easily presented in a decentralized setup.