arXiv Science⌕ Search

arXiv subjects

Oliver Henze

Publications and source records attributed to Oliver Henze.

2 recordsLinked to original sources

Bayesian Inverse Problems for EUV reflectometry with regularized JKO scheme

We study a Bayesian inverse problem in EUV reflectometry, where optical constants and layer thicknesses are inferred from angle- and wavelength-resolved reflection measurements. The physical process is modeled and evaluated using an efficient forward model based on the transfer matrix method. As an alternative to computationally expensive sampling-based inference, we employ an Eulerian, entropically regularized Jordan-Kinderlehrer-Otto (JKO) scheme that minimizes the Kullback--Leibler divergence via proximal steps in the Wasserstein space. Using the Benamou--Brenier dynamic formulation and a Hopf-Cole transform, each JKO step is reformulated as a pair of heat equations coupled only through their initial and terminal conditions, enabling an efficient fixed-point solver. The resulting high-dimensional densities are discretized on tensor-product grids and represented in tensor-train format, exploiting Kronecker structure for scalable computation. Preconditioning based on estimated posterior moments further improves tractability. We test the method with numerical experiments on EUV reflectometry data for a ruthenium--silicon ($\ce{Ru}$ on $\ce{Si}$) specimen and compare the learned tensor-train posterior to ensemble Markov chain Monte Carlo (MCMC) reference samples. We demonstrate a use case in which we are able to reuse the tensor-train posterior to obtain samples for shifted target distributions with less computational effort than a comparable MCMC method.

math.NA↗

Fourth Order Accurate Finite Volume CWENO Scheme For Astrophysical MHD Problems

In this work, a simple fourth-order accurate finite volume semi-discrete scheme is introduced to solve astrophysical magnetohydrodynamics (MHD) problems on Cartesian meshes. Hydrodynamic quantities like density, momentum and energy are discretised as volume averages. The magnetic field and electric field components are discretised as area and line averages respectively, so as to employ the constrained transport technique, which preserves the solenoidality of the magnetic field to machine precision. The present method makes use of a dimension-by-dimension approach employing a 1-D fourth-order accurate centrally weighted essentially non-oscillatory (1D-CWENO4) reconstruction polynomial. A fourth-order accurate, strong stability preserving (SSP) Runge-Kutta method is used to evolve the semi-discrete MHD equations in time. Higher-order accuracy of the scheme is confirmed in various linear and nonlinear multi-dimensional tests and the robustness of the method in avoiding unphysical numerical artifacts in the solution is demonstrated through several complex MHD problems.

physics.comp-ph↗