arXiv subjects

Willem Hundsdorfer

Publications and source records attributed to Willem Hundsdorfer.

17 recordsLinked to original sources

arXiv2018

Extrapolation-Based Super-Convergent Implicit-Explicit Peer Methods with A-stable Implicit Part

In this paper, we extend the implicit-explicit (IMEX) methods of Peer type recently developed in [Lang, Hundsdorfer, J. Comp. Phys., 337:203--215, 2017] to a broader class of two-step methods that allow the construction of super-convergent IMEX-Peer methods with A-stable implicit part. IMEX schemes combine the necessary stability of implicit and low computational costs of explicit methods to efficiently solve systems of ordinary differential equations with both stiff and non-stiff parts included in the source term. To construct super-convergent IMEX-Peer methods with favourable stability properties, we derive necessary and sufficient conditions on the coefficient matrices and apply an extrapolation approach based on already computed stage values. Optimised super-convergent IMEX-Peer methods of order s+1 for s=2,3,4 stages are given as result of a search algorithm carefully designed to balance the size of the stability regions and the extrapolation errors. Numerical experiments and a comparison to other IMEX-Peer methods are included.

math.NA↗

arXiv2017

On Multistep Stabilizing Correction Splitting Methods with Applications to the Heston Model

In this note we consider splitting methods based on linear multistep methods and stabilizing corrections. To enhance the stability of the methods, we employ an idea of Bruno & Cubillos (2016) who combine a high-order extrapolation formula for the explicit term with a formula of one order lower for the implicit terms. Several examples of the obtained multistep stabilizing correction methods are presented, and results on linear stability and convergence are derived. The methods are tested in the application to the well-known Heston model arising in financial mathematics and are found to be competitive with well-established one-step splitting methods from the literature.

math.NA↗

arXiv2017

On the Construction of Splitting Methods by Stabilizing Corrections with Runge-Kutta Pairs

In this technical note a general procedure is described to construct internally consistent splitting methods for the numerical solution of differential equations, starting from matching pairs of explicit and diagonally implicit Runge-Kutta methods. The procedure will be applied to suitable second-order pairs, and we will consider methods with or without a mass conserving finishing stage. For these splitting methods, the linear stability properties are studied and numerical test results are presented.

math.NA↗

arXiv2017

Extrapolation-Based Implicit-Explicit Peer Methods with Optimised Stability Regions

In this paper we investigate a new class of implicit-explicit (IMEX) two-step methods of Peer type for systems of ordinary differential equations with both non-stiff and stiff parts included in the source term. An extrapolation approach based on already computed stage values is applied to construct IMEX methods with favourable stability properties. Optimised IMEX-Peer methods of order p = 2, 3, 4, are given as result of a search algorithm carefully designed to balance the size of the stability regions and the extrapolation errors. Numerical experiments and a comparison to other implicit-explicit methods are included.

math.NA↗

arXiv2016

Patankar-Type Runge-Kutta Schemes for Linear PDEs

We study the local discretization error of Patankar-type Runge-Kutta methods applied to semi-discrete PDEs. For a known two-stage Patankar-type scheme the local error in PDE sense for linear advection or diffusion is shown to be of the maximal order ${\cal O}(Δt^3)$ for sufficiently smooth and positive exact solutions. However, in a test case mimicking a wetting-drying situation as in the context of shallow-water flows, this scheme yields large errors in the drying region. A more realistic approximation is obtained by a modification of the Patankar approach incorporating an explicit testing stage into the implicit trapezoidal rule.

math.NA↗

arXiv2013

Error Analysis of Explicit Partitioned Runge-Kutta Schemes for Conservation Laws

An error analysis is presented for explicit partitioned Runge-Kutta methods and multirate methods applied to conservation laws. The interfaces, across which different methods or time steps are used, lead to order reduction of the schemes. Along with cell-based decompositions, also flux-based decompositions are studied. In the latter case mass conservation is guaranteed, but it will be seen that the accuracy may deteriorate.

math.NA↗

arXiv2012

A comparison of 3D particle, fluid and hybrid simulations for negative streamers

In the high field region at the head of a discharge streamer, the electron energy distribution develops a long tail. In negative streamers, these electrons can run away and contribute to energetic processes such as terrestrial gamma-ray and electron flashes. Moreover, electron density fluctuations can accelerate streamer branching. To track energies and locations of single electrons in relevant regions, we have developed a 3D hybrid model that couples a particle model in the region of high fields and low electron densities with a fluid model in the rest of the domain. Here we validate our 3D hybrid model on a 3D (super-)particle model for negative streamers in overvolted gaps, and we show that it almost reaches the computational efficiency of a 3D fluid model. We also show that the extended fluid model approximates the particle and the hybrid model well until stochastic fluctuations become important, while the classical fluid model underestimates velocities and ionization densities. We compare density fluctuations and the onset of branching between the models, and we compare the front velocities with an analytical approximation.

physics.plasm-ph↗

arXiv2011

Spatially hybrid computations for streamer discharges: II. Fully 3D simulations

We recently have presented first physical predictions of a spatially hybrid model that follows the evolution of a negative streamer discharge in full three spatial dimensions; our spatially hybrid model couples a particle model in the high field region ahead of the streamer with a fluid model in the streamer interior where electron densities are high and fields are low. Therefore the model is computationally efficient, while it also follows the dynamics of single electrons including their possible run-away. Here we describe the technical details of our computations, and present the next step in a systematic development of the simulation code. First, new sets of transport coefficients and reaction rates are obtained from particle swarm simulations in air, nitrogen, oxygen and argon. These coefficients are implemented in an extended fluid model to make the fluid approximation as consistent as possible with the particle model, and to avoid discontinuities at the interface between fluid and particle regions. Then two splitting methods are introduced and compared for the location and motion of the fluid-particle-interface in three spatial dimensions. Finally, we present first results of the 3D spatially hybrid model for a negative streamer in air.

physics.plasm-ph↗

arXiv2011

Simulated avalanche formation around streamers in an overvolted air gap

We simulate streamers in air at standard temperature and pressure in a short overvolted gap. The simulation is performed with a 3D hybrid model that traces the single electrons and photons in the low density region, while modeling the streamer interior as a fluid. The photons are followed by a Monte-Carlo procedure, just like the electrons. The first simulation result is present here.

physics.plasm-ph↗

arXiv2009

3D hybrid computations for streamer discharges and production of run-away electrons

We introduce a 3D hybrid model for streamer discharges that follows the dynamics of single electrons in the region with strong field enhancement at the streamer tip while approximating the many electrons in the streamer interior as densities. We explain the method and present first results for negative streamers in nitrogen. We focus on the high electron energies observed in the simulation.

physics.plasm-ph↗

arXiv2009

Spatially hybrid computations for streamer discharges with generic features of pulled fronts: I. Planar fronts

Streamers are the first stage of sparks and lightning; they grow due to a strongly enhanced electric field at their tips; this field is created by a thin curved space charge layer. These multiple scales are already challenging when the electrons are approximated by densities. However, electron density fluctuations in the leading edge of the front and non-thermal stretched tails of the electron energy distribution (as a cause of X-ray emissions) require a particle model to follow the electron motion. As super-particle methods create wrong statistics and numerical artifacts, modeling the individual electron dynamics in streamers is limited to early stages where the total electron number still is limited. The method of choice is a hybrid computation in space where individual electrons are followed in the region of high electric field and low density while the bulk of the electrons is approximated by densities (or fluids). We here develop the hybrid coupling for planar fronts. First, to obtain a consistent flux at the interface between particle and fluid model in the hybrid computation, the widely used classical fluid model is replaced by an extended fluid model. Then the coupling algorithm and the numerical implementation of the spatially hybrid model are presented in detail, in particular, the position of the model interface and the construction of the buffer region. The method carries generic features of pulled fronts that can be applied to similar problems like large deviations in the leading edge of population fronts etc.

physics.plasm-ph↗

arXiv2007

Spatial coupling of particle and fluid models for streamers: where nonlocality matters

Particle models for streamer ionization fronts contain correct electron energy distributions, runaway effects and single electron statistics. Conventional fluid models are computationally much more efficient for large particle numbers, but create too low ionization densities in high fields. To combine their respective advantages, we here show how to couple both models in space. We confirm that the discrepancies between particle and fluid fronts arise from the steep electron density gradients in the leading edge of the fronts. We find the optimal position for the interface between models that minimizes computational effort and reproduces the results of a pure particle model.

physics.plasm-ph↗

arXiv2006

Photoionization in negative streamers: fast computations and two propagation modes

Streamer discharges play a central role in electric breakdown of matter in pulsed electric fields, both in nature and in technology. Reliable and fast computations of the minimal model for negative streamers in simple gases like nitrogen have recently been developed. However, photoionization was not included; it is important in air and poses a major numerical challenge. We here introduce a fast and reliable method to include photoionization into our numerical scheme with adaptice grids, and we discuss its importance for negative streamers. In particular, we identify different propagation regimes where photoionization does or does not play a role.

physics.plasm-ph↗

arXiv2006

The multiscale nature of streamers

Streamers are a generic mode of electric breakdown of large gas volumes. They play a role in the initial stages of sparks and lightning, in technical corona reactors and in high altitude sprite discharges above thunderclouds. Streamers are characterized by a self-generated field enhancement at the head of the growing discharge channel. We briefly review recent streamer experiments and sprite observations. Then we sketch our recent work on computations of growing and branching streamers, we discuss concepts and solutions of analytical model reductions, we review different branching concepts and outline a hierarchy of model reductions.

physics.plasm-ph↗

arXiv2006

Numerical convergence of the branching time of negative streamers

In sufficiently large gaps and electric fields, discharge streamers do branch. In [Arrayas et al., PRL 88, 174502 (2002)], we observed streamer branching numerically within a deterministic particle density model and explained it as a Laplacian instability of a thin space charge layer. Our numerical results were criticized in [Kulikovsky, PRL 89, 229401 (2002)]. We here present an adaptive grid refinement method for streamer simulations, and we carry out the first conclusive investigation on the effect of the numerical grid on streamer branching in different fields. On stepwise finer grids the branching time converges, hence streamer branching is for the first time predicted quantitatively.

physics.plasm-ph↗

arXiv2002

Branching of negative streamers in free flight

We recently have shown that a negative streamer in a sufficiently high homogeneous field can branch spontaneously due to a Laplacian instability, rather than approach a stationary mode of propagation with fixed radius. In our previous simulations, the streamer started from a wide initial ionization seed on the cathode. We here demonstrate in improved simulations that a streamer emerging from a single electron branches in the same way. In fact, though the evolving streamer is much more narrow, it branches after an even shorter propagation distance.

cond-mat.mtrl-sci↗

arXiv2002

Spontaneous Branching of Anode-Directed Streamers between Planar Electrodes

Non-ionized media subject to strong fields can become locally ionized by penetration of finger-shaped streamers. We study negative streamers between planar electrodes in a simple deterministic continuum approximation. We observe that for sufficiently large fields, the streamer tip can split. This happens close to Firsov's limit of `ideal conductivity'. Qualitatively the tip splitting is due to a Laplacian instability quite like in viscous fingering. For future quantitative analytical progress, our stability analysis of planar fronts identifies the screening length as a regularization mechanism.

nlin.PS↗