Palestrante: Prof. Ryszard Bialecki, Silesian University of Technology, Poland
DATA: 04/12/2014 (quinta-feira)
LOCAL: Bloco G, Sala 219B, Centro de Tecnologia, Ilha do Fundão.
Proper Orthogonal Decomposition is technique of extracting the most meaningful portion of information from large sets of multidimensional vectors. By rotating the coordinate system, it is possible to reduce the number of degrees of freedom while filtering out the noise from the data set.
The presentation will first give a general theory behind the POD method and its numerous variants. Then the way how the procedure can be used to reduce the order of degrees of freedom in transient Finite Elements calculations will be shown. Another fields of application namely the multiscale problems and approximation of radiation properties of gases will be discussed next. The thrust of the presentation will be on using POD in inverse analysis, where not only the feature of reducing the DOFs but also the filtration property proved to produce very efficient algorithms.
The seminar will be illustrated by numerous numerical examples.