## Nonlinear Conservation Laws and Finite Volume Methods for Astrophysical Fluid Flow (1998)

Venue: | Computational Methods for Astrophysical Fluid Flow, 27th Saas-Fee Advanced Course Lecture Notes |

Citations: | 6 - 0 self |

### BibTeX

@INPROCEEDINGS{Leveque98nonlinearconservation,

author = {Randall J. Leveque},

title = {Nonlinear Conservation Laws and Finite Volume Methods for Astrophysical Fluid Flow},

booktitle = {Computational Methods for Astrophysical Fluid Flow, 27th Saas-Fee Advanced Course Lecture Notes},

year = {1998},

pages = {1--160},

publisher = {Springer-- Verlag. Available}

}

### OpenURL

### Abstract

Contents 1. Introduction : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 4 1.1 Software : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 6 1.2 Notation : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 7 1.3 Classification of differential equations : : : : : : : : : : : : : : : : : : : : : : : 7 2. Derivation of conservation laws : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 10 2.1 The Euler equations of gas dynamics : : : : : : : : : : : : : : : : : : : : : : : 13 2.2 Dissipative fluxes : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 14 2.3 Source terms : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 14 2.4 Radiative trans

### Citations

1339 | Classical Electrodynamics - Jackson - 1999 |

719 | Fluid Mechanics - Landau, Lifshitz - 1959 |

514 |
A Multigrid Tutorial
- Briggs, Henson, et al.
- 2000
(Show Context)
Citation Context ...preconditioned conjugate gradient methods and multigrid methods have been developed for solving sparse linear systems arising from elliptic equations. There is a huge literature on these methods; see =-=[41]-=-, [101], [244], [255] for general discussions of such methods and [171], [172] for some astrophysical applications. More generally, a linear elliptic equation has the form Lq = f ; (1.2) where L is so... |

425 |
Numerical Methods for Conservation laws
- LeVeque
- 1990
(Show Context)
Citation Context ... 11 behavior of solutions of nonlinear conservation laws and the development of modern numerical methods for their solution. This section of these notes is largely a condensed version of my ETH Notes =-=[141]-=-, and is intended to be an overview of the important points for readers who are already familiar with this theory. More details can be found in [141] for those who are not. Numerous other books are al... |

418 |
Hyperbolic Systems of Conservation Laws
- Lax
- 1957
(Show Context)
Citation Context ... so on, with q m\Gamma1 connected to q r by an m-shock. If q l and q r are sufficiently close together then this can always be achieved. Lax proved a stronger version of this in his fundamental paper =-=[135]-=-. (By considering rarefaction waves also, 38 LeVeque --- Saas-Fee Notes q l qr q q Fig. 3.8. Construction of a 2-shock solution to the Riemann problem for the isothermal equations with left and right ... |

372 |
Adaptive mesh refinement for hyperbolic partial differential equations
- Berger, Oliger
- 1983
(Show Context)
Citation Context ...ng this idea to allow increased resolution in some regions. This approach, illustrated in Fig. 6.1, was pioneered by Berger and Oliger [27] and has been developed further by Berger and coworkers [19] =-=[25]-=- [28] [26]. Refinement on rectangular patches has the advantage that the data structures remain relatively simple, and consist of a nested set of grids. On each grid a standard finite-volume method is... |

351 |
Local Adaptive Mesh Refinement for Shock Hydrodynamics
- Berger, Colella
- 1989
(Show Context)
Citation Context ...is idea to allow increased resolution in some regions. This approach, illustrated in Fig. 6.1, was pioneered by Berger and Oliger [27] and has been developed further by Berger and coworkers [19] [25] =-=[28]-=- [26]. Refinement on rectangular patches has the advantage that the data structures remain relatively simple, and consist of a nested set of grids. On each grid a standard finite-volume method is used... |

324 |
Iterative Methods for Solving Linear Systems
- Greenbaum
- 1997
(Show Context)
Citation Context ...ditioned conjugate gradient methods and multigrid methods have been developed for solving sparse linear systems arising from elliptic equations. There is a huge literature on these methods; see [41], =-=[101]-=-, [244], [255] for general discussions of such methods and [171], [172] for some astrophysical applications. More generally, a linear elliptic equation has the form Lq = f ; (1.2) where L is some elli... |

298 |
Numerical Solution of the Navier-Stokes Equations
- Chorin
- 1968
(Show Context)
Citation Context ...in such a way that (8.19) remains satisfied. There is no local evolution equation for p; it must be determined using this global constraint. One common approach is the projection method introduced in =-=[51] (se-=-e also, e.g., [23], [20], [185]). In each time step one first advances (8.20) without the rp term to obtain u , i.e., we solve u t + u \Delta ru = ��r 2 u (8.21) over time \Deltat. The resulting u... |

258 |
High resolution schemes for hyperbolic conservation laws
- Harten
- 1983
(Show Context)
Citation Context ...thod satisfy TV (Q n+1 )sTV (Q n ) : (4.48) For a scalar conservation law, the exact solution has nonincreasing variation and so this is a reasonable condition to impose on a numerical method. Harten =-=[103]-=- introduced the use of this criterion in developing and analyzing numerical methods. For a scalar equation, steps 2 and 3 of Algorithm RSA are TVD, and so the overall method is TVD provided that the r... |

198 |
Hyperbolic Systems of Conservation Laws and the Mathematical Theory of Shock Waves
- Lax
- 1972
(Show Context)
Citation Context ...t points for readers who are already familiar with this theory. More details can be found in [141] for those who are not. Numerous other books are also available on this theory, including [60], [97], =-=[136]-=-, [221], [256]. Conservation laws are time-dependent systems of partial differential equations (usually nonlinear) with a particularly simple structure. In one space dimension the equations take the f... |

187 | Efficient implementation of weighted ENO schemes
- Jiang, Shu
- 1996
(Show Context)
Citation Context ...lation based on the behavior of the solution in order to avoid interpolating across a discontinuity. Some references on ENO schemes, and more recent work on WENO (weighted ENO), include [104], [109], =-=[119]-=-, [155], [215], [216]. Numerical Methods 83 4.8.2 Flux-vector splitting. We now return to the question of defining a first-order accurate interface flux and consider an alternative to the Godunov meth... |

183 |
Numerical Approximation of Hyperbolic Systems of Conservation Law
- Godlewski, Raviart
- 1996
(Show Context)
Citation Context ...portant points for readers who are already familiar with this theory. More details can be found in [141] for those who are not. Numerous other books are also available on this theory, including [60], =-=[97]-=-, [136], [221], [256]. Conservation laws are time-dependent systems of partial differential equations (usually nonlinear) with a particularly simple structure. In one space dimension the equations tak... |

156 |
Uniformly high order accurate essentially non-oscillatory schemes
- Harten, Engquist, et al.
- 1987
(Show Context)
Citation Context ...-order interpolation based on the behavior of the solution in order to avoid interpolating across a discontinuity. Some references on ENO schemes, and more recent work on WENO (weighted ENO), include =-=[104]-=-, [109], [119], [155], [215], [216]. Numerical Methods 83 4.8.2 Flux-vector splitting. We now return to the question of defining a first-order accurate interface flux and consider an alternative to th... |

155 |
A mathematical introduction to fluid mechanics
- Chorin, Marsden
- 1990
(Show Context)
Citation Context ...rning. The total energy E takes the form E = p fl \Gamma 1 + 1 2 %u 2 + q 0 %Z ; (5.25) where q 0 is the heat release. We consider the particular case of a Chapman-Jouguet detonation wave. See, e.g., =-=[53], [6-=-0], [97], [257] for more discussion of combustion in gas dyanamics, and also Muller's notes on nuclear burning and supernova explosions. For �� ? 0 we expect some region of finite width across whi... |

142 |
A model for collision processes in gases. I. Small amplitude processes in charged and neutral one-component system
- Bhatnagar, Gross, et al.
- 1954
(Show Context)
Citation Context ...ted by considering the collisionless Boltzman equation as a model of gas dynamics, as discussed in [108]. More recently the BGK approximation to the collision term, due to Bhatnagar, Gross, and Krook =-=[31]-=-, has been used to obtain more accurate Boltzman schemes. BGK methods have the advantage of not requiring Riemann solutions as such, and may be particularly useful for low density gases. Prendergast a... |

140 |
The dynamics of general relativity
- Arnowitt, Deser, et al.
- 1962
(Show Context)
Citation Context ...re then solved by marching forward in the time-like direction that links these manifolds together, advancing fl forward in "time". This approach was originally proposed by Arnowitt, Deser, a=-=nd Misner [6]-=- and is also known as the ADM formalism. The equations they proposed have a Hamiltonian structure. More recently, other 3 + 1 formulations of the Einstein equations have been proposed that give first-... |

136 |
A second-order projection method for the incompressible Navier–Stokes equations
- Bell, Colella, et al.
- 1989
(Show Context)
Citation Context ...mains satisfied. There is no local evolution equation for p; it must be determined using this global constraint. One common approach is the projection method introduced in [51] (see also, e.g., [23], =-=[20], [1-=-85]). In each time step one first advances (8.20) without the rp term to obtain u , i.e., we solve u t + u \Delta ru = ��r 2 u (8.21) over time \Deltat. The resulting u will not be divergence-free... |

133 |
On Upstream Differencing and Godunov-Type Schemes for Hyperbolic Conservation
- Harten, Lax, et al.
- 1983
(Show Context)
Citation Context ...t would arise from this particular Riemann data, and the intermediate state can then be calculated by the condition of conservation. This approach is developed by Harten, Lax, and van Leer originally =-=[108]-=- and improved by Einfeldt [78]. 4.6.4 The Roe solver. One of the most popular Riemann solvers currently in use is due to Roe [195]. The idea is to determine an approximate Riemann solutionsq(x; t) by ... |

109 |
die partiellen Differenzengleichungen der Mathematischen
- Courant, Friedrichs, et al.
- 1928
(Show Context)
Citation Context ...rant, Friedrichs, and Lewy, who wrote a fundamental paper in 1928 that was essentially the first paper on the stability and convergence of finite difference methods for partial differential equations =-=[61]-=-. To understand this general condition, we must discuss the domain of dependencesof a time-dependent PDE. For the advection equation (1.5), the solution q(X; T ) at some fixed point (X; T ) depends on... |

102 |
Supersonic Flow and Shock Waves
- Courant, Friedrichs
- 1948
(Show Context)
Citation Context ...the important points for readers who are already familiar with this theory. More details can be found in [141] for those who are not. Numerous other books are also available on this theory, including =-=[60]-=-, [97], [136], [221], [256]. Conservation laws are time-dependent systems of partial differential equations (usually nonlinear) with a particularly simple structure. In one space dimension the equatio... |

100 |
Numerical Computation of Internal and External Flows
- Hirsch
- 1990
(Show Context)
Citation Context ...op one particular set of methods of this type. The development of such methods has a long and rich history, and numerous related methods can be found in the literature. Books such as [2], [86], [97], =-=[111]-=-, [130], [141], [178], [185], [226], [242] contain descriptions of these methods and pointers to the literature. 56 LeVeque --- Saas-Fee Notes Rather than viewing Q n i as an approximation to the sing... |

95 | Multidimensional upwind methods for hyperbolic conservation laws
- Colella
(Show Context)
Citation Context ...ng these terms would arise naturally from the fractional step procedure, but with an unsplit method they must be explicitly modeled. A wide variety of unsplit methods have been developed, e.g., [22], =-=[56]-=-, [85], [200], [205], [219], [220], [243], [266], A multi-dimensional version of the wave-propagation algorithms described above has recently been developed, yielding a class of algorithms that follow... |

86 |
Computational methods in ordinary differential equations
- LAMBERT
- 1973
(Show Context)
Citation Context ...der to maintain stability. Methods for stiff problems are necessarily implicit in time. 4.1.7 Stiff ODE's. The classic example is the case of a stiff ordinary differential equation. See, for example, =-=[131]-=-, [142], or any text on numerical methods for ordinary differential equations. As an example, consider the ODE u 0 (t) = (cos t \Gamma u) \Gamma sin t : (4.13) One particular solution is the function ... |

84 |
Universality and scaling in gravitational collapse of a massless scalar field
- Choptuik
- 1993
(Show Context)
Citation Context ...powerful technique for solving astrophysical problems efficiently, especially for problems in three space dimensions. Related AMR methods have been used elsewhere in computational astrophysics,se.g., =-=[48]-=-, [75], [163], [228]. 7. Computational difficulties The high-resolution finite methods discussed in these lectures can be used very successfully for a wide range of problems. They are not foolproof, h... |

67 |
Three-dimensional adaptive mesh refinement for hyperbolic conservation laws
- Bell, Berger, et al.
- 1994
(Show Context)
Citation Context ...pplying this idea to allow increased resolution in some regions. This approach, illustrated in Fig. 6.1, was pioneered by Berger and Oliger [27] and has been developed further by Berger and coworkers =-=[19]-=- [25] [28] [26]. Refinement on rectangular patches has the advantage that the data structures remain relatively simple, and consist of a nested set of grids. On each grid a standard finite-volume meth... |

67 |
Flux-corrected transport. I. SHASTA, a fluid transport algorithm that works
- Boris, Book
(Show Context)
Citation Context ...). The same idea in the context of flux limiting, reducing the magnitude of the numerical flux to avoid oscillations, was introduced in the flux-corrected transport (FCT) algorithms of Boris and Book =-=[37]-=-. We can view this as creating a hybrid algorithm that is second order accurate in smooth regions but which reduces to a more robust first-order algorithm near discontinuities. This idea of hybridizat... |

61 |
Uniformly high-order accurate nonoscillatory schemes I
- Harten, Osher
- 1987
(Show Context)
Citation Context ...interpolation based on the behavior of the solution in order to avoid interpolating across a discontinuity. Some references on ENO schemes, and more recent work on WENO (weighted ENO), include [104], =-=[109]-=-, [119], [155], [215], [216]. Numerical Methods 83 4.8.2 Flux-vector splitting. We now return to the question of defining a first-order accurate interface flux and consider an alternative to the Godun... |

60 |
Computational Techniques for Fluid Dynamics
- Fletcher
- 1992
(Show Context)
Citation Context ...e will develop one particular set of methods of this type. The development of such methods has a long and rich history, and numerous related methods can be found in the literature. Books such as [2], =-=[86]-=-, [97], [111], [130], [141], [178], [185], [226], [242] contain descriptions of these methods and pointers to the literature. 56 LeVeque --- Saas-Fee Notes Rather than viewing Q n i as an approximatio... |

59 |
Self-adjusting grid methods for one-dimensional hyperbolic conservation laws
- Harten, Hyman
- 1983
(Show Context)
Citation Context ...efaction wave, it is necessary to modify the approximate Riemann solver in order to obtain entropy satisfying solutions. There are various ways to do this. One approach, discussed by Harten and Hyman =-=[105]-=-, is outlined in [141]. Various other approaches have also been used, e.g., [54], [112], [263]. Numerical Methods 67 Figure 4.6(c) shows the same test problem as in Fig. 4.6(b) but with the entropy fi... |

56 | Levermore. Numerical schemes for hyperbolic conservation laws with stiff relaxation terms
- Jin, David
- 1996
(Show Context)
Citation Context ...mann problem where the source terms are taken into account. A wide variety of different approaches to handling stiff source terms have been explored in the recent literature. A few examples are [24], =-=[120], [1-=-84], One can also avoid problems by making sure that the grid is sufficiently fine, and the time step small enough, that the source terms are fully resolved. For �� fixed, no matter how small, the... |

44 | Direct Eulerian MUSCL scheme for gas dynamics - Colella - 1985 |

44 |
Theoretical and Numerical Structure for Reacting Shock Wave
- Colella, Majda, et al.
- 1986
(Show Context)
Citation Context ...asonable at first glance and yet are propagating at nonphysical speeds due to purely numerical artifacts. This was observed in a simple model combustion problem studied by Colella, Majda, and Roytburd=-=[58]. Although in practi-=-cal problems there may be dozens of chemical species of importance, in this model there are only two chemical species: "unburnt gas" and "burnt gas". By using % as one of our varia... |

43 | Adaptive mesh refinement using wave-propagation algorithms for hyperbolic systems
- Berger, LeVeque
- 1998
(Show Context)
Citation Context ...sure that the formulas used to update the solution on each grid are consistent with one another. In particular, when a conservation law is being solved we must preserve global conservation. See [28], =-=[30]-=- for details on how this is done. Berger's work has recently been combined with the clawpack code to create amrclaw [29], [30], which can be used on quite general hyperbolic systems. Figure 6.1 shows ... |

41 |
An upwind differencing scheme for the equations of ideal magnetohydrodynamics
- Brio, Wu
- 1988
(Show Context)
Citation Context ...richer structure of compound waves that involve both shock and rarefaction waves in the same family traveling together. The existence of such waves was first discovered computationally by Brio and Wu =-=[44]-=-, [259] who used one of the earliest Riemann-based methods for MHD. The analysis of intermediate waves in MHD is currently an active area of research, and there is still some controversy about what th... |

37 |
A second-order projection method for variabledensity flows
- Bell, Marcus
- 1992
(Show Context)
Citation Context ...19) remains satisfied. There is no local evolution equation for p; it must be determined using this global constraint. One common approach is the projection method introduced in [51] (see also, e.g., =-=[23], [2-=-0], [185]). In each time step one first advances (8.20) without the rp term to obtain u , i.e., we solve u t + u \Delta ru = ��r 2 u (8.21) over time \Deltat. The resulting u will not be divergenc... |

37 | Automatic adaptive grid refinement for the Euler equations
- Berger, Jameson
- 1985
(Show Context)
Citation Context ...ea to allow increased resolution in some regions. This approach, illustrated in Fig. 6.1, was pioneered by Berger and Oliger [27] and has been developed further by Berger and coworkers [19] [25] [28] =-=[26]-=-. Refinement on rectangular patches has the advantage that the data structures remain relatively simple, and consist of a nested set of grids. On each grid a standard finite-volume method is used to s... |

34 | Relativistic Fluids and Magneto-Fluids - Anile - 1989 |

32 |
Efficient Solution Algorithms for the Riemann Problem for Real Gases,n
- Colella, Glaz
(Show Context)
Citation Context ... method. More details on the solution of the Riemann 46 LeVeque --- Saas-Fee Notes problem can also be found in x81 of [60]. Solving the Riemann problem for non-ideal gases can be more difficult. See =-=[57]-=-, for example. 4. Numerical Methods in One Dimension 4.1 Finite difference theory For students not familiar with numerical methods, it may be useful to review some of the fundamental ideas and issues ... |

31 |
On Godunov-type methods for gas dynamics
- Einfeldt
- 1988
(Show Context)
Citation Context ...lar Riemann data, and the intermediate state can then be calculated by the condition of conservation. This approach is developed by Harten, Lax, and van Leer originally [108] and improved by Einfeldt =-=[78]-=-. 4.6.4 The Roe solver. One of the most popular Riemann solvers currently in use is due to Roe [195]. The idea is to determine an approximate Riemann solutionsq(x; t) by solving a constant coefficient... |

31 | On finite-difference approximations and entropy conditions for shocks - Hyman, Harten, et al. - 1976 |

28 |
Stable and entropy satisfying approximations for transonicc flow calculations
- Engquist, Osher
- 1980
(Show Context)
Citation Context ...he basis for the Osher solver, which is a generalization of the Engquist-Osher method for scalar conservation laws, first developed in the context of the small disturbance equation for transonic flow =-=[81]-=-. Details for general conservation laws can be found in [181]. 4.6.3 The HLLE solver. Another approach is to approximate the full Riemann solution by a single intermediate state bounded by two waves m... |

28 |
Transitional waves for conservation laws
- Isaacson, Marchesin, et al.
- 1990
(Show Context)
Citation Context ... where similar problems arise, and in more general mathematical investigations. Along with the papers mentioned above, a small sample of interesting recent work includes [17], [42], [43], [91], [92], =-=[114]-=-, [127], [154], [173], [207], [208], [209], [229], [261], [262]. Zachary, Malagoli, and Colella [265] discuss a numerical approach to dealing with nonstrict hyperbolicity, based on ideas in [21]. 8.4.... |

27 |
Hyperbolic reductions for Einstein’s equations
- Friedrich
- 1996
(Show Context)
Citation Context ...on is best. The number of variables and equations varies between different formulations, ranging from 30 to 58, so these systems are quite large. Some possibilities are discussed in [35], [36], [50], =-=[93]-=-. Hyperbolic formulations have received recent attention for theoretical as well as numerical work. The well-developed mathematical theory of hyperbolic conservation laws can be used to gain new insig... |

23 | A wave propagation method for three-dimensional hyperbolic conservation laws
- Langseth, LeVeque
- 2000
(Show Context)
Citation Context ...s, but also helps to determine whether the numerical method is isotropic or suffers from grid-orientation effects that lead to the results being better resolved in some directions than in others. See =-=[134]-=-, [144], [147] for some examples of such numerical tests. 6.1 Dimensional splitting The simplest way to extend the finite volume methods developed in one dimension to more dimensions is to use dimensi... |

22 | Trangenstein, Higher order Godunov methods for general systems of conservation laws
- Bell, Colella, et al.
- 1989
(Show Context)
Citation Context ...92], [114], [127], [154], [173], [207], [208], [209], [229], [261], [262]. Zachary, Malagoli, and Colella [265] discuss a numerical approach to dealing with nonstrict hyperbolicity, based on ideas in =-=[21]-=-. 8.4.1 Artificial resistivity. We obtain a hyperbolic problem in MHD by assuming the fluid is perfectly conducting, i.e., we ignore the dissipative effect of resistivity along with viscosity and heat... |

21 |
An unsplit, higher order Godunov method for scalar conservation laws in multiple dimensions
- Bell, Dawson, et al.
- 1988
(Show Context)
Citation Context ...plitting these terms would arise naturally from the fractional step procedure, but with an unsplit method they must be explicitly modeled. A wide variety of unsplit methods have been developed, e.g., =-=[22]-=-, [56], [85], [200], [205], [219], [220], [243], [266], A multi-dimensional version of the wave-propagation algorithms described above has recently been developed, yielding a class of algorithms that ... |

19 |
The piecewise-parabolic method (PPM) for gas-dynamical simulations
- Colella, Woodward
- 1984
(Show Context)
Citation Context ...ciples have been developed in the past two decades. One of the algorithms of this type that is best known in the astrophysics community is the piecewise parabolic method (PPM) of Woodward and Colella =-=[59]-=-, which uses a piecewise quadratic reconstruction, with appropriate limiting. 4.7.5 Total variation. How much should we limit the slope in a piecewise linear reconstruction? Ideally we would like to h... |

18 | Capturing Shock Reflections: An Improved Flux Formula
- Donat, Marquina
- 1996
(Show Context)
Citation Context ... may be particularly useful for low density gases. Prendergast and Xu [189] give a nice introduction to this approach. 4.8.3 The Marquina flux. A flux-vector splitting recently introduced by Marquina =-=[73]-=-, [157] appears to be very robust on difficult astrophysical computations. Let r p i andsp (Q i ) be the right eigenvectors and eigenvalues of f 0 (Q i ), respectively, and decompose both Q i and f(Q ... |

15 |
The generalized Riemann problem for reactive flows
- Ben-Artzi
- 1989
(Show Context)
Citation Context ...ed Riemann problem where the source terms are taken into account. A wide variety of different approaches to handling stiff source terms have been explored in the recent literature. A few examples are =-=[24], [1-=-20], [184], One can also avoid problems by making sure that the grid is sufficiently fine, and the time step small enough, that the source terms are fully resolved. For �� fixed, no matter how sma... |