Randomized resolvent analysis
Jean Hélder Marques Ribeiro, Chi-An Yeh, Kunihiko Taira
Abstract
Randomized numerical algebra is incorporated into resolvent analysis to reduce large-scale resolvent operators to their low-rank approximations. The key to finding the resolvent modes accurately is to weigh the random test matrix using insights from the base flow. Turbulent flow over a NACA0012 airfoil at Re = 23,000 is used to demonstrate significant speedup and memory savings to accurately find principal resolvent modes.
Topics & Concepts
ResolventMathematicsResolvent formalismMatrix (chemical analysis)Pure mathematicsAirfoilMathematical analysisFlow (mathematics)Algebra over a fieldBase (topology)Applied mathematicsNumerical analysisType (biology)AlgorithmNumerical linear algebraModel Reduction and Neural NetworksMatrix Theory and AlgorithmsStochastic Gradient Optimization Techniques