arXiv subjects

Roger Käppeli

Publications and source records attributed to Roger Käppeli.

10 recordsLinked to original sources

arXiv2025

Physical Constraint Preserving Higher Order Finite Volume Schemes for Divergence-Free Astrophysical MHD and RMHD

Higher order finite volume schemes for magnetohydrodynamics (MHD) and relativistic magnetohydrodynamics (RMHD) are very valuable because they allow us to carry out astrophysical simulations with very high accuracy. However, astrophysical problems sometimes have unusually large Mach numbers, exceptionally high Lorentz factors and very strong magnetic fields. All these effects cause higher order codes to become brittle and prone to code crashes. In this paper we document physical constraint preserving (PCP) methods for treating numerical MHD and RMHD. While unnecessary for standard problems, for stringent astrophysical problems these methods show their value. We describe higher order methods that allow divergence-free evolution of the magnetic field. We present a novel two-dimensional Riemann solver. This two-dimensional Riemann solver plays a key role in the design of PCP schemes for MHD and RMHD. We present a very simple PCP formulation and show how it is amalgamated with the evolution of face-centered magnetic fields. The methods presented here are time-explicit and do not add much to the computational cost. We show that the methods meet their design accuracies and work well on problems that would otherwise be considered too extreme for typical higher order Godunov methods of the type used in computational astrophysics.

astro-ph.IM↗

arXiv2024

Poseidon: Efficient Foundation Models for PDEs

We introduce Poseidon, a foundation model for learning the solution operators of PDEs. It is based on a multiscale operator transformer, with time-conditioned layer norms that enable continuous-in-time evaluations. A novel training strategy leveraging the semi-group property of time-dependent PDEs to allow for significant scaling-up of the training data is also proposed. Poseidon is pretrained on a diverse, large scale dataset for the governing equations of fluid dynamics. It is then evaluated on a suite of 15 challenging downstream tasks that include a wide variety of PDE types and operators. We show that Poseidon exhibits excellent performance across the board by outperforming baselines significantly, both in terms of sample efficiency and accuracy. Poseidon also generalizes very well to new physics that is not seen during pretraining. Moreover, Poseidon scales with respect to model and data size, both for pretraining and for downstream tasks. Taken together, our results showcase the surprising ability of Poseidon to learn effective representations from a very small set of PDEs during pretraining in order to generalize well to unseen and unrelated PDEs downstream, demonstrating its potential as an effective, general purpose PDE foundation model. Finally, the Poseidon model as well as underlying pretraining and downstream datasets are open sourced, with code being available at https://github.com/camlab-ethz/poseidon and pretrained models and datasets at https://huggingface.co/camlab-ethz.

cs.LG↗

arXiv2021

Curl constraint-preserving reconstruction and the guidance it gives for mimetic scheme design

Several important PDE systems, like magnetohydrodynamics and computational electrodynamics, are known to support involutions where the divergence of a vector field evolves in divergence-free or divergence constraint-preserving fashion. Recently, new classes of PDE systems have emerged for hyperelasticity, compressible multiphase flows, so-called first order reductions of the Einstein field equations, or a novel first order hyperbolic reformulation of Schrödinger's equation, to name a few, where the involution in the PDE supports curl-free or curl constraint-preserving evolution of a vector field. Since mimetic numerical schemes for the solution of the former class of PDEs are well-developed, we draw guidance from them for the solution of the latter class of PDEs. We show that a study of the curl constraint-preserving reconstruction gives us a great deal of insight into the design of consistent, mimetic schemes for these involutionary PDEs. The importance of multidimensional Riemann solvers in facilitating the design of such schemes is also documented. We study the problem of curl constraint-preserving reconstruction as it pertains to the design of mimetic discontinuous Galerkin (DG) and finite volume (FV) schemes for PDEs that support such an involution. This is done for two and three dimensional structured mesh problems where we deliver closed form expressions for the reconstruction. The role that this reconstruction plays in the curl-free, or curl-preserving prolongation of vector fields in adaptive mesh refinement (AMR) is also discussed. In two dimensions, a von Neumann analysis of structure-preserving DG-like schemes that mimetically satisfy the curl constraints, is also presented. Numerical results are also presented to show that the schemes meet their design accuracy.

math.NA↗

arXiv2021

Von Neumann Stability Analysis of DG-like and PNPM-like Schemes for PDEs that have Globally Curl-Preserving Evolution of Vector Fields

This paper examines a class of PDEs where some part of the PDE system evolves a vector field whose curl remains zero or grows in proportion to specified source terms. Such PDEs are referred to as curl-free or curl-preserving respectively. In this paper we catalogue a class of DG-like schemes for such PDEs. To retain the globally curl-free or curl-preserving constraints, the components of the vector field, as well as their higher moments, have to be collocated at the edges of the mesh. They are updated by using potentials that are collocated at the vertices of the mesh. The resulting schemes : 1) do not blow up even after very long integration times, 2) do not need any special cleaning treatment, 3) can operate with large explicit timesteps, 4) do not require the solution of an elliptic system and 5) can be extended to higher orders using DG-like methods. The methods rely on a special curl-preserving reconstruction and they also rely on multidimensional upwinding. The Galerkin projection, so crucial to the design of a DG method, is now carried out in the edges of the mesh and yields a weak form update that uses potentials that are obtained at the vertices of the mesh with the help of a multidimensional Riemann solver. A von Neumann stability analysis of the curl-preserving methods is carried out and the limiting CFL numbers of this entire family of methods is catalogued in this work. The stability analysis confirms that with increasing order of accuracy, our novel curl-free methods have superlative phase accuracy while substantially reducing dissipation. We also show that PNPM-like methods retain much of the excellent wave propagation characteristics of the DG-like methods while offering a much larger CFL number and lower computational complexity.

math.NA↗

arXiv2020

High order discretely well-balanced methods for arbitrary hydrostatic atmospheres

We introduce novel high order well-balanced finite volume methods for the full compressible Euler system with gravity source term. They require no a priori knowledge of the hydrostatic solution which is to be well-balanced and are not restricted to certain classes of hydrostatic solutions. In one spatial dimension we construct a method that exactly balances a high order discretization of any hydrostatic state. The method is extended to two spatial dimensions using a local high order approximation of a hydrostatic state in each cell. The proposed simple, flexible, and robust methods are not restricted to a specific equation of state. Numerical tests verify that the proposed method improves the capability to accurately resolve small perturbations on hydrostatic states.

math.NA↗

arXiv2020

Well-balanced finite volume schemes for nearly steady adiabatic flows

We present well-balanced finite volume schemes designed to approximate the Euler equations with gravitation. They are based on a novel local steady state reconstruction. The schemes preserve a discrete equivalent of steady adiabatic flow, which includes non-hydrostatic equilibria. The proposed method works in Cartesian, cylindrical and spherical coordinates. The scheme is not tied to any specific numerical flux and can be combined with any consistent numerical flux for the Euler equations, which provides great flexibility and simplifies the integration into any standard finite volume algorithm. Furthermore, the schemes can cope with general convex equations of state, which is particularly important in astrophysical applications. Both first- and second-order accurate versions of the schemes and their extension to several space dimensions are presented. The superior performance of the well-balanced schemes compared to standard schemes is demonstrated in a variety of numerical experiments. The chosen numerical experiments include simple one-dimensional problems in both Cartesian and spherical geometry, as well as two-dimensional simulations of stellar accretion in cylindrical geometry with a complex multi-physics equation of state.

math.NA↗

arXiv2018

High-order well-balanced finite volume schemes for the Euler equations with gravitation

A high-order well-balanced scheme for the Euler equations with gravitation is presented. The scheme is able to preserve a spatially high-order accurate discrete representation of a large class of hydrostatic equilibria. It is based on a novel local hydrostatic reconstruction, which, in combination with any standard high-order accurate reconstruction procedure, achieves genuine high-order accuracy for smooth solutions close or away from equilibrium. The resulting scheme is very simple and can be implemented into any existing finite volume code with minimal effort. Moreover, the scheme is not tied to any particular form of the equation of state, which is crucial for example in astrophysical applications. Several numerical experiments demonstrate the robustness and high-order accuracy of the scheme nearby and out of hydrostatic equilibrium.

math.NA↗

arXiv2015

An advanced leakage scheme for neutrino treatment in astrophysical simulations

We present an Advanced Spectral Leakage (ASL) scheme to model neutrinos in the context of core-collapse supernovae and compact binary mergers. Based on previous gray leakage schemes, the ASL scheme computes the neutrino cooling rates by interpolating local production and diffusion rates (relevant in optically thin and thick regimes, respectively), separately for discretized values of the neutrino energy. Neutrino trapped components are also modeled, based on equilibrium and timescale arguments. The better accuracy achieved by the spectral treatment allows a more reliable computation of neutrino heating rates in optically thin conditions. The scheme has been calibrated and tested against Boltzmann transport in the context of Newtonian spherically symmetric models of core-collapse supernovae. ASL shows a very good qualitative and a partial quantitative agreement, for key quantities from collapse to a few hundreds of milliseconds after core bounce. We have proved the adaptability and flexibility of our ASL scheme coupling it to an axisymmetric Eulerian and to a three-dimensional SPH code to simulate core-collapse. Therefore, the neutrino treatment presented here is ideal for large parameter-space explorations, parametric studies, high-resolution tests, code developments, and long-term modeling of asymmetric configurations, where more detailed neutrino treatments are not available or currently computationally too expensive.

astro-ph.IM↗

arXiv2015

Construction of approximate entropy measure valued solutions for hyperbolic systems of conservation laws

Entropy solutions have been widely accepted as the suitable solution framework for systems of conservation laws in several space dimensions. However, recent results in \cite{CDL1,CDL2} have demonstrated that entropy solutions may not be unique. In this paper, we present numerical evidence that demonstrates that state of the art numerical schemes \emph{may not} necessarily converge to an entropy solution of systems of conservation laws as the mesh is refined. Combining these two facts, we argue that entropy solutions may not be suitable as a solution framework for systems of conservation laws, particularly in several space dimensions. Furthermore, we propose a more general notion, that of \emph{entropy measure valued solutions}, as an appropriate solution paradigm for systems of conservation laws. To this end, we present a detailed numerical procedure, which constructs stable approximations to entropy measure valued solutions and provide sufficient conditions that guarantee that these approximations converge to an entropy measure valued solution as the mesh is refined, thus providing a viable numerical framework for systems of conservation laws in several space dimensions. A large number of numerical experiments that illustrate the proposed schemes are presented and are utilized to examine several interesting properties of the computed entropy measure valued solutions.

math.NA↗

arXiv2012

Nucleosynthesis in core-collapse supernova explosions triggered by a quark-hadron phase transition

We explore heavy element nucleosynthesis in the explosion of massive stars which are triggered by a quark-hadron phase transition during the early post bounce phase of core-collapse supernovae. The present study is based on general relativistic radiation hydrodynamics simulations with three-flavor Boltzmann neutrino transport in spherical symmetry, which utilize a quark-hadron hybrid equation of state based on the MIT bag model for strange quark matter. The quark-hadron phase transition inside the stellar core forms a shock wave propagating towards the surface of the proto-neutron star. The shock wave results in an explosion and ejects neutron-rich matter which is piled up or accreting on the proto-neutron star. Later, during the cooling phase, the proto-neutron star develops a proton-rich neutrino-driven wind. We present a detailed analysis of the nucleosynthesis outcome in both neutron-rich and proton-rich ejecta and compare our integrated nucleosynthesis with observations of metal poor stars.

astro-ph.HE↗