PhD Seminar: Summer Term 2008
Kristian Spoerer (PhD)
(University of Nottingham, University of Leicester)
Single-player games (often called puzzles) have received considerable attention from the scientific community. Consequently, interesting insights into some puzzles, and into the approaches for solving them, have emerged. Examples of techniques and algorithms that might be applied to these problems include Logic, Evolutionary Computation and Neural Networks. Researchers have also looked at these problems from a Computational Complexity perspective with the aim of identifying hard problem instances. However, many puzzles have been neglected, possibly because they are unknown to many people. This presentation will focus on NP-Complete puzzles and discuss various aspects of them in the hope of motivating further research in this fascinating area, particularly for those puzzles that have received little scientific attention to date.
Sandra Aguirre Herrera
(Universidad Politecnica de Madrid)
The development of Joint Degrees is an important mechanism for opening higher education systems nationwide, adapting them to the international standard, and promoting quality assessment to a broader environment. Since e-Learning systems covers a wide range of academic programs, and as joint degrees such as e-learning are rapidly growing trends, finding a suitable solution that enables universities to design joint degrees through their own E-Learning Systems becomes necessary. In this talk, I will present the design of a Federated Service-Oriented Architecture, whose implementation will be based on Web Services. Our main objective is to allow ELearning Systems to be able to work in a collaborative way; combining their data and functionality through the integration of their services. These services will be combined and federated, according to collaboration agreements between Universities.
(University of Nottingham, Host: Daniela Petrisan)
Modularity is the key to designing and reasoning about large systems. Because of the lack of a clean theory of operational semantics the question of its modularity has long remained unanswered. In this talk, I will describe Turi's elegant categorical framework for SOS, also known as mathematical operational semantics. Then, I will propose an answer to the question of its modularity. Perhaps unsurprisingly, the proposed solution has many similarities with the way modular monadic interpreters are constructed.
(University of Birmingham, Host: Christian Kissig)
This talk will discuss the universal properties of a particular 2,3 Turing machine, including an introduction to the notion of universality in models of computer systems in general, to Turing machines, and to a particular 2,3 Turing machine found by Stephen Wolfram and which I proved was capable of exhibiting universal behaviour with the right initial conditions, including a broad outline of the proof.
(University of Paderborn, Host: Denes Bisztray)
Software models are used in software development for different purposes, e.g. for specifying the requirements, for documenting design aspects or for code generation. Models can also be used for testing purposes, e.g. for generating test cases and test scripts, as test oracles, and for defining the ending criteria. Model-based testing (MBT) advocates the systematic use of models for testing activities. MBT mostly uses models from requirements analysis and design phase so that the correctness of the implementation with respect to these models is checked.
In our research we use UML-based modeling techniques for MBT. "Visual contracts" are developed at University of Paderborn as a visual specification language. They are based on the theory of graph transformations and define the behavior of software systems by specifying pre- and post-conditions, which are UML diagrams. We use visual contracts for generating test cases. The expected test results can be computed by means of graph transformations.
(University of Wolverhampton, Host: Emilio Tuosto)
Ever since Descartes introduced planar coordinate systems, visual representations of data have become a widely accepted way of describing scientific phenomena. Most recently, wireless sensor networks have emerged as a technology which can provide high fidelity, multi-modal sense data over large geographic areas and long periods of time. The construction of accurate and efficient visualisations of sense data gathered from wireless sensor networks is an open problem.
In the first part of this talk I will present a complete system solution that solves global network data mapping problems through localized and distributed computation. This system changes the mechanism of use of certain capabilities in wireless sensor network; for instance, the interpolation and routing capabilities are realised as services of the network. Experiments showed that this technique performs favourably against similar techniques which do not exploit such services of the network to generate maps.
In the second part of this talk I will describe my first steps towards a general distributed and adaptive mapping framework. This framework exploits the knowledge of the underlying wireless sensor network application domain to maximize the information return by dynamically adapting to changes in the concrete reality of the sensed environment.
(Oxford University Computing Laboratory, Host: Christian Kissig)
Constraint satisfaction is a framework for describing combinatorial problems. Instead of explicitly specifying a solution procedure, one lists properties that any solution must have. These properties are called constraints. I discuss two representations of constraint satisfaction problems, complete and microstructure, and some ideas that follow from the perspective these afford. The complete representation leads to new decision problems with structures of unbounded arity. No polynomial time algorithm is known for solving these problems, but they can be solved in subexponential time. The microstructure is defined using direct products of relational structures. For constraints involving two variables at a time, the microstructure is a graph. This provides a transformation to INDEPENDENT SET, and leads to new classes of constraint satisfaction problems that were previously not known to be tractable. i
Brian Campbell (PhD)
(University of Edinburgh, Host: Christian Kissig)
Bounding the resource usage of programs is useful for ensuring the reliability and security of systems, especially where mobile code from an untrusted source is used.
I will talk about the development of a series of memory usage analyses based on a system by Hofmann and Jost. These assign hypothetical amounts of free memory called potential to data structures through type annotations, and produce constraints which ensure the potential of the input is sufficient to satisfy all allocations in the program. (This is similar to amortized computational complexity.)
The original system only bounds the heap memory requirements of a simple first-order functional programming language. I will also describe how the analysis can be extended to provide reasonable bounds on stack memory (which is my own PhD work), and new language features (further work by Jost). Finally, I will briefly contrast this approach with a few other analyses.
(Swansea University, Host: Christian Kissig)
In this talk I will present a theory and practice for the evaluation of test cases with respect to formal specifications. In particular, I will give a brief introduction of the specification language that we use: namely Csp-CASL. We use Csp-CASL to define and evaluate black-box tests for reactive systems. I will show how this approach is well suited to deal with refinement of specifications, and how it is possible to develop test cases already from very abstract specification.