How good is random information? Approximation in the Hilbert space setting
Sprache des Vortragstitels:
Englisch
Original Tagungtitel:
Minisymposium Algorithms for High-Dimensional Approximation (and Integration) Problems
Sprache des Tagungstitel:
Englisch
Original Kurzfassung:
To find optimal algorithms for numerical problems like function approximation, one has to choose the information optimally, and this might be difficult in practice. Choosing the information functionals as iid random functionals and using the optimal algorithm for the chosen information may be almost as good as using optimal information or may be much worse, depending on the problem setting. We analyse this problem in the Hilbert space setting and give a partial answer in this case about the power of random information.
Sprache der Kurzfassung:
Englisch
Vortragstyp:
Hauptvortrag / Eingeladener Vortrag auf einer Tagung