Abstract for the talk at 18.08.2015 (15:15 h)Oberseminar NUMERIK UND WISSENSCHAFTLICHES RECHNEN
Alexander Litvinenko (King Abdullah University of Science and Technology (KAUST), Saudi Arabia)
Tensor methods in stochastic PDEs and in inverse problems
After discretization of stochastic PDEs (PDEs with uncertain coefficients) one obtains a high dimensional problem. Usual methods have very high exponential cost and do not work. Therefore we apply low-rank tensor methods to solve the problem and to perform post processing (computing different statistics) in an efficient way with a low computational cost. Some open problems will be highlighted at the end of the talk.