Contributions in Merlin

We introduce non-homothetic preferences into an innovation-based growth model and study how income and wealth inequality affect economic growth. We identify a (positive) price effect -- where increasing inequality allows innovators to charge higher prices and (negative) market-size effects -- with higher inequality implying smaller markets for new goods and/or a slower transition of new goods into mass markets. It turns out that price effects dominate market-size effects. We also show that a redistribution from the poor to the rich may be Pareto improving for low levels of inequality.

Changes are a crucial part of the life-cycle of modern software systems. Common versioning systems such as CVS store version histories of source code. Usually, they are not capable of tracking changes on a more sophisticated level. They provide lexical but not syntactical change analysis. The existing Eclipse-plug-in ChangeDistiller bridges this gap by providing a sophisticated analysis of structural source code changes. It uses an abstract syntax tree (AST) representation of subsequent revisions of source code files and compares the trees by using a change detection algorithm for hierarchically structured information. The outcome is an edit script describing the operations that are necessary to transform the original version of the tree into the modified one. We aim at improving the sub-algorithm responsible for matching trees. It yields insufficiencies in terms of matching leaves in general, it often produces sub-optimal results for small subtrees, and it is not able to handle large number of changes adequately. To overcome this issues, we propose customized similarity measures and a similarity ranking algorithm for leaves, as well as dynamic modulation of the tree similarity thresholds whenever small tree structures are encountered. To prove our claims, we establish an extensive benchmark for investigating runtime performance and accuracy. The benchmark is based on the JUnit regression testing framework and relies on artificial source code examples, as well as on examples taken from a medium-sized real project.

null

null

Toward adaptive underwater locomotion, this paper presents the experimental results of fish-like swimming robots that we have newly developed. By using motor control with only one degree of freedom, these robots exhibit surprisingly rich behavioural diversity in three dimensional underwater environment. This paper focuses on some of the behavior variations, i.e. forward, turning, and vertical movement, which are required for the underwater three dimensional navigation. The visual behavior analysis shows that, even though there is only one motor, these behavior are possible because these robots exploit the unique interaction with the environment derived from the morphological properties. For better understanding how the material influences the swimming behaviour, a second robot with bending sensors implemented in its tail-fin measures the deflexion during swimming. Moreover, some of the behaviors demonstrated by these robots have a considerable similarity to those of biological systems, which would also contribute to understand the adaptive behavior of animals. Based on the experimental results, we speculate further issue on ``cheap"" underwater locomotion.

This demonstration illustrates the APDF tree: an adaptive tree that supports the effective and effcient computation of continuous density information. The APDF tree allocates more partition points in non-linear areas of the density function and fewer points in linear areas of the density function. This yields not only a bounded, but a tight control of the error. The demonstration explains the core steps of the computation of the APDF tree (split, kernel additions, tree optimization, kernel additions, unsplit) and demos the implementation for different datasets.

null

Integration of text and documents into database management systems has been the subject of much research. However, most of the approaches are limited to data retrieval. Collaborative text editing, i.e. the ability for multiple users to work on a document instance simultaneously, is rarely supported. Also, documents mostly consist of plain text only, and support very limited meta data storage or search. We address the problem by proposing an extended definition of document data type which comprises not only the text itself but also structural information such as layout, template and semantics, as well as document creation meta data. We implemented a new collaborative data type Document which supports document manipulation via a text editing API and extended SQL syntax (TX SQL), as detailed in this work. We report also on the search capabilities of our document management system and present some of the future challenges for collaborative document management.

Background: Current established models predicting mortality in acute coronary syndrome (ACS) patients are derived from randomised controlled trials performed in the 1990's, and are thus based on and predictive for selected populations. These scores perform inadequately in patients treated according to current guidelines. The aim of this study was to develop a model with improved predictive performance applicable to all kinds of ACS, based on outcomes in real world patients from the new millennium. Methods: The AMIS (Acute Myocardial Infarction in Switzerland)-Plus registry prospectively collects data from ACS patients admitted to 56 Swiss hospitals. Patients included in this registry between October 2001 and May 2005 (n = 7520) were the basis for model development. Modern data mining computational methods using new classification learning algorithms were tested to optimise mortality risk prediction using well-defined and non-ambiguous variables available at first patient contact. Predictive performance was quantified as ""area under the curve"" (AUC, range 0 - 1) in a receiver operator characteristic, and was compared to the benchmark risk score from the TIMI study group. Results were verified using 10-fold cross-validation. Results: Overall, hospital mortality was 7.5%. The final prediction model was based on the ""Averaged One-Dependence Estimators"" algorithm and included the following 7 input variables: 1) Age, 2) Killip class, 3) systolic blood pressure, 4) heart rate, 5) pre-hospital mechanical resuscitation, 6) history of heart failure, 7) history of cerebrovascular disease. The output of the model was an estimate of in-hospital mortality risk for each patient. The AUC for the entire cohort was 0.875, compared to 0.803 for the TIMI risk score. The AMIS model performed equally well for patients with or without ST elevation myocardial infarction (AUC 0.879 and 0.868, respectively). Subgroup analysis according to the initial revascularisation modality indicated that the AMIS model performed best in patients undergoing PCI (AUC 0.884 vs. 0.783 for TIMI) and worst in patients receiving no revascularisation therapy (AUC 0.788 vs. 0.673 for TIMI). The model delivered an acurate and reproducible prediction over the complete range of risks and for all kinds of ACS. Conclusions: The AMIS model performs about 10% better than established risk prediction models for hospital mortality in patients with all kinds of ACS in the modern era. Modern data mining algorithms proved useful to optimise the model development.

The prospect theory of Kahneman and Tversky (in Econometrica 47(2), 263–291, 1979) and the cumulative prospect theory of Tversky and Kahneman (in J. Risk uncertainty 5, 297–323, 1992) are descriptive models for decision making that summarize several violations of the expected utility theory. This paper gives a survey of applications of prospect theory to the portfolio choice problem and the implications for asset pricing. We demonstrate that prospect theory (and similarly cumulative prospect theory) has to be re-modelled if one wants to apply it to portfolio selection. We suggest replacing the piecewise power value function of Tversky and Kahneman (in J. Risk uncertainty 5, 297–323, 1992) with a piecewise negative exponential value function. This latter functional form is still compatible with laboratory experiments but it has the following advantages over and above Tversky and Kahneman’s piecewise power function: 1. The Bernoulli Paradox does not arise for lotteries with finite expected value. 2. No infinite leverage/robustness problem arises. 3. CAPM-equilibria with heterogeneous investors and prospect utility do exist. 4. It is able to simultaneously resolve the following asset pricing puzzles: the equity premium, the value and the size puzzle. In contrast to the piecewise power value function it is able to explain the disposition effect. Resolving these problems of prospect theory we show how it can be combined with mean–variance portfolio theory.