Filtern
Erscheinungsjahr
- 2008 (42) (entfernen)
Dokumenttyp
Sprache
- Deutsch (29)
- Englisch (12)
- Mehrsprachig (1)
Volltext vorhanden
- ja (42) (entfernen)
Schlagworte
- Stress (3)
- Approximation (2)
- Sensibilisierung <Immunologie> (2)
- dendritic cells (2)
- psychotherapy (2)
- stress (2)
- 13. Jahrhundert (1)
- 13th century (1)
- 2D DIGE (1)
- Abfluss (1)
Institut
- Psychologie (13)
- Raum- und Umweltwissenschaften (8)
- Geschichte, mittlere und neuere (4)
- Germanistik (3)
- Mathematik (3)
- Rechtswissenschaft (3)
- Wirtschaftswissenschaften (3)
- Kunstgeschichte (2)
- Philosophie (1)
- Politikwissenschaft (1)
In this thesis, we investigate the quantization problem of Gaussian measures on Banach spaces by means of constructive methods. That is, for a random variable X and a natural number N, we are searching for those N elements in the underlying Banach space which give the best approximation to X in the average sense. We particularly focus on centered Gaussians on the space of continuous functions on [0,1] equipped with the supremum-norm, since in that case all known methods failed to achieve the optimal quantization rate for important Gauss-processes. In fact, by means of Spline-approximations and a scheme based on the Best-Approximations in the sense of the Kolmogorov n-width we were able to attain the optimal rate of convergence to zero for these quantization problems. Moreover, we established a new upper bound for the quantization error, which is based on a very simple criterion, the modulus of smoothness of the covariance function. Finally, we explicitly constructed those quantizers numerically.