Search

In this thesis we concentrate on instances of the GLE which can be represented in a Markovian form in an extended phase space. We extend previous results on the geometric ergodicity of this class of GLEs using Lyapunov techniques, which allows us to conclude ergodicity for a large class of GLEs relevant to molecular dynamics applications. The main body of this thesis concerns the numerical discretisation of the GLE in the extended phase space representation. We generalise numerical discretisation schemes which have been previously proposed for the underdamped Langevin equation and which are based on a decomposition of the vector field into a Hamiltonian part and a linear SDE. Certain desirable properties regarding the accuracy of configurational averages of these schemes are inherited in the GLE context. We also rigorously prove geometric ergodicity on bounded domains by showing that a uniform minorisation condition and a uniform Lyapunov condition are satisfied for sufficiently small timestep size. We show that the discretisation schemes which we propose behave consistently in the white noise and overdamped limits, hence we provide a family of universal integrators for Langevin dynamics. Finally, we consider multiple-time stepping schemes making use of a decomposition of the fluctuation-dissipation term into a reversible and non-reversible part. These schemes are designed to efficiently integrate instances of the GLE whose Markovian representation involves a high number of auxiliary variables or a configuration dependent fluctuation-dissipation term. We also consider an application of dynamics based on the GLE in the context of large scale Bayesian inference as an extension of previously proposed adaptive thermostat methods. In these methods the gradient of the log posterior density is only evaluated on a subset (minibatch) of the whole dataset, which is randomly selected at each timestep. Incorporating a memory kernel in the adaptive thermostat formulation ensures that time-correlated gradient noise is dissipated in accordance with the fluctuation-dissipation theorem. This allows us to relax the requirement of using i.i.d. minibatches, and explore a variety of minibatch sampling approaches.

Compared to other disciplines, software engineering as of today is still dependent on craftsmanship of highly-skilled workers. However, with constantly increasing complexity and efforts, existing software engineering approaches appear more and more inefficient. A paradigm shift towards industrial production methods seems inevitable. Recent advances in academia and practice have lead to the availability of industrial key principles in software development as well. Specialization is represented in software product lines, standardization and systematic reuse are available with component-based development, and automation has become accessible through model-driven engineering. While each of the above is well researched in theory, only few cases of successful implementation in the industry are known. This becomes even more evident in specialized areas of software engineering such as systems integration. Today’s IT systems need to quickly adapt to new business requirements due to mergers and acquisitions and cooperations between enterprises. This certainly leads to integration efforts, i.e. joining different subsystems into a cohesive whole in order to provide new functionality. In such an environment. the application of industrial methods for software development seems even more important. Unfortunately, software development in this field is a highly complex and heterogeneous undertaking, as IT environments differ from customer to customer. In such settings, existing industrialization concepts would never break even due to one-time projects and thus insufficient economies of scale and scope. This present thesis, therefore, describes a novel approach for a more efficient implementation of prior key principles while considering the characteristics of software development for systems integration. After identifying the characteristics of the field and their affects on currently-known industrialization concepts, an organizational model for industrialized systems integration has thus been developed. It takes software product lines and adapts them in a way feasible for a systems integrator active in several business domains. The result is a three-tiered model consolidating recurring activities and reducing the efforts for individual product lines. For the implementation of component-based development, the present thesis assesses current component approaches and applies an integration metamodel to the most suitable one. This ensures a common understanding of systems integration across different product lines and thus alleviates component reuse, even across product line boundaries. The approach is furthermore aligned with the organizational model to depict in which way component-based development may be applied in industrialized systems integration. Automating software development in systems integration with model-driven engineering was found to be insufficient in its current state. The reason herefore lies in insufficient tool chains and a lack of modelling standards. As an alternative, an XML-based configuration of products within a software product line has been developed. It models a product line and its products with the help of a domain-specific language and utilizes stylesheet transformations to generate compliable artefacts. The approach has been tested for its feasibility within an exemplarily implementation following a real-world scenario. As not all aspects of industrialized systems integration could be simulated in a laboratory environment, the concept was furthermore validated during several expert interviews with industry representatives. Here, it was also possible to assess cultural and economic aspects. The thesis concludes with a detailed summary of the contributions to the field and suggests further areas of research in the context of industrialized systems integration.