Coupling kinetic and continuum using data-driven maximum entropy distribution

An important class of multi-scale flow scenarios deals with an interplay between kinetic and continuum phenomena. While hybrid solvers provide a natural way to cope these settings, two issues restrict their performance. Foremost, the inverse problem implied by estimating distributions has be addressed, boundary conditions for solver. The next issue comes from defining robust yet accurate switching criterion solvers. This study introduces data-driven kinetic-continuum coupling, where Maximum-Entropy-Distribution (MED) is employed parametrize arising field variables. Two regression methodologies Gaussian-Processes (GPs) Artificial-Neural-Networks (ANNs) are utilized predict MEDs efficiently. Hence MED estimates carry out besides providing criterion. To achieve latter, breakdown parameter defined means Fisher information distance computed estimates. We test performance our devised estimators recovering bi-modal densities. Next, integrated into solution algorithm. Here Direct Simulation Monte-Carlo (DSMC) Smoothed-Particle Hydrodynamics (SPH) chosen as solvers, respectively. monatomic gas inside Sod's shock tube investigated, DSMC-SPH coupling realized applying Very good agreements respect benchmark solutions along promising speed-up observed in reported cases.