Residual Distribution. basics, recents developments, relations with other techniques

Size: px
Start display at page:

Download "Residual Distribution. basics, recents developments, relations with other techniques"

Transcription

1 Residual Distribution basics, recents developments, relations with other techniques MARIO RICCHIUTO July 19, 2012

2 INTRODUCTION with historical perspective t u+ F(u) = 0 In the 80 s numerical techniques devised to provide high order monotone approximation to solutions of hyperbolic C.L.s 1. A. Harten (J.Comput.Phys., 1983) : TVD conditions 2. Goodman, LeVeque (Math.Comp.) : TVD in mulitd = first order MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

3 INTRODUCTION with historical perspective t u+ F(u) = 0 In the 80 s numerical techniques devised to provide high order monotone approximation to solutions of hyperbolic C.L.s 1. A. Harten (J.Comput.Phys., 1983) : TVD conditions 2. Goodman, LeVeque (Math.Comp.) : TVD in mulitd = first order This has spawned a number of new approaches Monotonicity conditions 1. ENO/WENO (Harten, Osher, Engquist, Chakravarthy, Shu) 2. TVB conditions (Shu Math.Comp. 1987) 3. Positive coefficient schemes (Spekreijse Math.Comp. 1987) Discretization frameworks 1. Stabilized FE (or central) (Hughes, Morton, Ni, Lerat, Jameson) 2. Discontinuous Galerkin (Cockburn and Shu, starting 1988) 3. Roe s Fluctuation Splitting (starting 1986) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

4 INTRODUCTION with historical perspective t u+ F(u) = 0 Discontinuous Galerkin : smart and elegant combination of existing tools (approximation, Galerkin projection, Riemann solvers, limiters) to generate automatically arbitrary higher order schemes An instant hit... MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

5 INTRODUCTION with historical perspective t u+ F(u) = 0 Fluctuation Splitting/Residual Distribution : A more fundamental and robust approach [...] due to Roe (1986), is that of the genuinely multidimensional upwind schemes. These may be regarded as the true multi-d generalization of 1-D fluctuation splitting [...] These methods are best formulated on simplex-type (finite-element) grids and include newly developed, compact limiters for avoiding oscillations... MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

6 INTRODUCTION with historical perspective t u+ F(u) = 0 Fluctuation Splitting/Residual Distribution : A more fundamental and robust approach [...] due to Roe (1986), is that of the genuinely multidimensional upwind schemes. These may be regarded as the true multi-d generalization of 1-D fluctuation splitting [...] These methods are best formulated on simplex-type (finite-element) grids and include newly developed, compact limiters for avoiding oscillations... B. van Leer, (excerpt from Upwind high resolution methods for compressible flow: from donor cell to residual distribution, Commun.Comput.Phys. 1(2), 2006) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

7 INTRODUCTION with historical perspective t u+ F(u) = 0 Initial idea (P.L.Roe, Num. Meth. Fluid Dyn. 1982) : given values of u on the mesh, integral of F(u) over elements measures the error (fluctuation); decompose the fluctuation in signals allowing to evolve solution values to those solving the problem MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

8 INTRODUCTION with historical perspective t u+ F(u) = 0 Initial idea (P.L.Roe, Num. Meth. Fluid Dyn. 1982) : integral of F(u) over elements measures the error (fluctuation); decompose the fluctuation in signals allowing to evolve solution values to those solving the problem 80 and 90, under the lead of P.L. Roe, H. Deconinck : 1. Genuinely multidimensional upwinding (scalar) 2. Steady state hyperbolic decomposition 3. Each scalar hyperbolic component discretized using MU technique Genuinely multid upwind second order compact (nearest neighbor) second order, nonlinear, positive discretization. Very well adapted to steady supersonic, multidimensional Roe linearization, inexact decompositions in sub-critical case, no unsteady. MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

9 INTRODUCTION with historical perspective t u+ F(u) = 0 Initial idea (P.L.Roe, Num. Meth. Fluid Dyn. 1982) : integral of F(u) over elements measures the error (fluctuation); decompose the fluctuation in signals allowing to evolve solution values to those solving the problem 80 and 90, under the lead of P.L. Roe, H. Deconinck : Genuinely multid upwind second order compact (nearest neighbor) second order, nonlinear, positive discretization. Very well adapted to steady supersonic, multidimensional Roe linearization, inexact decompositions in sub-critical case, no unsteady. 20 years later with contributions of R. Abgrall, T.J. Barth, D. Caraeni, M. Hubbard, C.W. Shu et al. 1. Time dependent problems 2. Conservation without Roe lienarization 3. Higher (than second) accuracy 4. More general data approximation (including discontinuous) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

10 INTRODUCTION with historical perspective t u+ F(u) = 0 Initial idea (P.L.Roe, Num. Meth. Fluid Dyn. 1982) : integral of F(u) over elements measures the error (fluctuation); decompose the fluctuation in signals allowing to evolve solution values to those solving the problem 80 and 90, under the lead of P.L. Roe, H. Deconinck : Genuinely multid upwind second order compact (nearest neighbor) second order, nonlinear, positive discretization. Very well adapted to steady supersonic, multidimensional Roe linearization, inexact decompositions in sub-critical case, no unsteady. 20 years later with contributions of R. Abgrall, T.J. Barth, D. Caraeni, M. Hubbard, C.W. Shu et al. The method has come to a form which is more correctly referred to as a weighted residual method, many of the initial multidimensional upwind fluctuation splitting ideas could not (so far) be retained.. MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

11 INTRODUCTION with historical perspective t u+ F(u) = 0 Initial idea (P.L.Roe, Num. Meth. Fluid Dyn. 1982) : integral of F(u) over elements measures the error (fluctuation); decompose the fluctuation in signals allowing to evolve solution values to those solving the problem 80 and 90, under the lead of P.L. Roe, H. Deconinck 20 years later with contributions of R. Abgrall, T.J. Barth, D. Caraeni, M. Hubbard, C.W. Shu et al. Despite so many contributions the method never really caught up with other approaches based on more sound mathematical foundations (DG), and it definitely has a much lower level of maturity. But some ideas have stuck. Hybrid nature : in between finite element and finite volume. This allows easily to import/export ideas born in this framework to improve others and vice-versa... keeping alive the interest in the method MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

12 INTRODUCTION with historical perspective t u+ F(u) = 0 Initial idea (P.L.Roe, Num. Meth. Fluid Dyn. 1982) : integral of F(u) over elements measures the error (fluctuation); decompose the fluctuation in signals allowing to evolve solution values to those solving the problem Despite so many contributions the method never really caught up with other approaches based on more sound mathematical foundations (DG), and it definitely has a much lower level of maturity What am I going to tell you? MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

13 COURSE OUTLINE : Part I 1. Conservative FV discretization and fluctuations 2. Fluctuation splitting/residual distribution framework 3. Design principles 4. Limiters in reverse 5. Relations with other techniques MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

14 COURSE OUTLINE : Part II 1. Higher (than second) orders 2. Time dependent problems 3. Viscous problems 4. Free surface flows 5. Summary, perspectives MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

15 PART I OUTLINE Finite Volume schemes and Fluctuations Design criteria Nonlinear schemes and limiters Relations with other techniques MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

16 1 Conservative Finite Volume schemes and Fluctuations MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

17 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law t u+ x F(u) = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

18 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law Conservative FV : t u+ x F(u) = 0 x i du i dt + F(u L i+1/2,ur i+1/2 ) F(u L i 1/2,uR i 1/2 ) = 0 u i u i 1 u i+1 i 1 i i+1 i 1/2 i+1/2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

19 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law Conservative FV : t u+ x F(u) = 0 x i du i dt + F(u L i+1/2,ur i+1/2 ) F(u L i 1/2,uR i 1/2 ) = 0 u i 1 u i u i+1 (consistent) F(u,u) = F(u) (L-continuous) F(u,v) F(u,z) K F v z i 1 i i+1 i 1/2 i+1/2 (E-stable. Monotone, etc.) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

20 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law Conservative FV : u i t u+ x F(u) = 0 x i du i dt + F i+1/2 F i 1/2 = 0 u i 1 u i+1 i 1 i i+1 i 1/2 i+1/2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

21 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law Conservative FV : t u+ x F(u) = 0 x i du i dt + F i+1/2 F i 1/2 = 0 u i 1 u i u i+1 du i+1 x i+1 dt du i 1 x i 1 dt + F i+3/2 F i+1/2 = 0 + F i 1/2 F i 3/2 = 0 i 1 i i+1 i 1/2 i+1/2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

22 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law Conservative FV : t u+ x F(u) = 0 x i du i dt + F i+1/2 F i 1/2 = 0 u i u i 1 u i+1 i 1 i i+1 i 1/2 i+1/2 du i+1 x i+1 dt du i 1 x i 1 dt + F i+3/2 F i+1/2 = 0 + F i 1/2 F i 3/2 = 0 Conservation from flux cancelation at interfaces MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

23 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law Conservative FV : t u+ x F(u) = 0 x i du i dt +( F i+1/2 F i )+(F i F i 1/2 ) = 0 u i u i 1 u i+1 i 1 i i+1 i 1/2 i+1/2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

24 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law t u+ x F(u) = 0 Conservative FV : φ i+1/2 i φ i 1/2 i {}}{{}}{ du i x i dt + ( F i+1/2 F i )+(F i F i 1/2 ) = 0 u i u i 1 u i+1 i 1 i i+1 i 1/2 i+1/2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

25 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law Conservative FV : t u+ x F(u) = 0 du i x i dt +φi+1/2 i +φ i 1/2 i = 0 u i 1 u i u i+1 at i+1/2 conservation becomes : φ i+1/2 i +φ i+1/2 i+1 = ( F i+1/2 F i )+(F i+1 F i+1/2 ) i 1 i i+1 i 1/2 i+1/2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

26 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law Conservative FV : t u+ x F(u) = 0 du i x i dt +φi+1/2 i +φ i 1/2 i = 0 u i 1 u i u i+1 at i+1/2 conservation becomes : φ i+1/2 i +φ i+1/2 i+1 = F i+1 F i := φ i+1/2 i 1 i i+1 i 1/2 i+1/2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

27 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law Conservative FV : t u+ x F(u) = 0 du i x i dt +φi+1/2 i +φ i 1/2 i = 0 u i 1 u i u i+1 at i 1/2 conservation becomes : φ i 1/2 i +φ i 1/2 i 1 = F i F i 1 := φ i 1/2 i 1 i i+1 i 1/2 i+1/2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

28 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law t u+ x F(u) = 0 Conservative FV : u i du i x i dt +φi+1/2 i +φ i 1/2 i = 0 u i 1 u i+1 i 1 i 1/2 i + 1/2 i i+1 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

29 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law t u+ x F(u) = 0 Conservative FV : du i x i dt +φi+1/2 i +φ i 1/2 i = 0 u i 1 φ i 1/2 u i u i+1 i φ i 1/2 := xf, i 1 (fluctuation) i 1 i 1/2 i + 1/2 i i+1 φ i 1/2 = F i F i 1 (as before...) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

30 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law t u+ x F(u) = 0 Conservative FV : u i 1 φ i 1/2 i u i φ i+1/2 i du i x i dt +φi+1/2 i +φ i 1/2 i = 0 u i+1 i φ i 1/2 := xf, (fluctuation) i 1 φ i 1/2 i = F i F i 1/2 (splitting) i 1 i 1/2 i + 1/2 i i+1 φ i 1/2 i 1 = F i 1/2 F i 1 (splitting) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

31 Finite Volume schemes and Fluctuations in 1D Starting point : conservation law t u+ x F(u) = 0 Conservative FV : u i 1 φ i 1/2 i u i φ i+1/2 i du i x i dt +φi+1/2 i +φ i 1/2 i = 0 u i+1 i i+1 φ i 1/2 := xf, φ i+1/2 := xf i 1 i φ i 1/2 i = F i F i 1/2, φ i+1/2 i = F i+1/2 F i i 1 i 1/2 i + 1/2 i i+1 FV scheme in Fluctuation splitting form... still the same guy though MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

32 Finite Volume schemes and Fluctuations in 2D The multi-d case. Starting point : conservation law t u+ F(u) = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

33 Finite Volume schemes and Fluctuations in 2D l i j K n K il n i = 2( n K ij + nk il ) fij K nk ij C i f ij C j The FV scheme reads C i du i dt + j f ij F ˆndl = 0 i j MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

34 Finite Volume schemes and Fluctuations in 2D l i j K n K il n i = 2( n K ij + nk il ) K fij ij C i f ij C j The FV scheme reads C i du i dt + K i K j K fij K i F ˆndl = 0 j MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

35 Finite Volume schemes and Fluctuations in 2D l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij The FV scheme reads C i du i dt + K i K j K i F ij n K ij = 0 K ij n K ij j MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

36 Finite Volume schemes and Fluctuations in 2D l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij The FV scheme reads Discrete conservation C i du i dt + K i K j K i F ij n K ij = 0 F ij n K ij + F ji n K ji = 0 K ij n K ij j MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

37 Finite Volume schemes and Fluctuations in 2D l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij Using the identity K C i du i dt + j nk ij = 0 K i K j K i K ij ( F ij F i ) n K ij = 0 n K ij j MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

38 Finite Volume schemes and Fluctuations in 2D l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij The FV scheme reads C i du i dt + K i K j K i K ij ( F ij F i ) n K ij = 0 } {{ } φ K i n K ij j MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

39 Finite Volume schemes and Fluctuations in 2D l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij The FV scheme reads C i du i dt + K i Kφ K i = 0, φ K i = j K i K ij n K ij j ( F ij F i ) n K ij MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

40 Finite Volume schemes and Fluctuations l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij The FV scheme reads C i du i dt + Discrete conservation K i Kφ K i = 0, φ K i = j K F ij n K ij + F ji n K ji = 0 = j K i K ij n K ij j ( F ij F i ) n K ij φ K j = 1 F i n j := φ K 2 j K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

41 Finite Volume schemes and Fluctuations l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij The FV scheme reads C i du i dt + K i Kφ K i = 0, φ K i = j K i K ij n K ij j ( F ij F i ) n K ij Discrete conservation (F h continuous P 1 finite element approx.) φ K j = φ K = F h j K K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

42 Finite Volume schemes and Fluctuations l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij The FV scheme reads (F h continuous P 1 finite element approx.) φ K = K F h, i K ij Discrete conservation { }}{ φ K j = φ K j K C i du i dt + K i Kφ K i = 0, φ K i = ( F ij F i ) n K ij j K... but it s still the same guy..!! MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19, n K ij j

43 Finite Volume schemes and Fluctuations l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij The FV scheme reads (F h continuous P 1 finite element approx.) φ K = K F h, i K ij Discrete conservation { }}{ φ K i = φ K j K C i du i dt + K i Kφ K i = 0, φ K i = ( F ij F i ) n K ij j K... but it s still the same guy..!! MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19, n K ij j

44 2 Residual Distribution the framework MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

45 Residual Distribution framework Ω h F(u) = 0 in Ω u = g on Γ λ(u) = u F(u) (1) Γ λ Some notations... Consider Ω h tesselation of Ω Unknowns (Degrees of Freedom, DoF) : u i u(m i ) M i Ω h a given set of nodes (vertices +other dofs) u h : continuous polynomial interpolation u h = i ψ i u i MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

46 Residual Distribution framework 1. K Ω h compute : φ K = K F h (u h ) K φ K i i K 2. Distribution : φ K = Distribution coeff.s : φ K i =β K i φ K φ K φ K 1 φ K 2 φ K 3 3. Compute nodal values : solve algebraic system T i T φ K i = 0, i Ω h (2) i MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

47 Residual Distribution framework Seek the steady limit of C i du i dt + K i K φ K i t K i K φ K i = 0 (3) The idea of Residual Distribution or Fluctuation Splitting Fluctuations & Signals (Roe, Num.Meth.Fluid Dyn., 1982) From an initial guess, nodal values evolve to steady state due to signals proportional to cell residuals (Roe s fluctuation) 1 - Compute fluctuation 2 - Split 3 - Gather signals K φ K φ K 1 φ K 2 φ K 3 i 4 - Evolve eq. ((3)) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

48 Structural conditions Conservation. LW theorem : convergence (if...) to weak solution? Stability. which form of stability (energy/entropy, equivalent algebraic condition, convergence?), choice of φ K i Accuracy. characterization of the error, choice of φ K i Oscillations. monotonicity preserving schemes, choice of φ K i MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

49 Conservation Conservation, LW theorem for RD and F(u) = 0 Under some (standard) continuity assumptions on φ K and φ K i the discrete solution u h converges (if...!) to a weak solution of the continuous problem, provided that (Abgral, Barth SISC, 2002 ; Abgrall, Roe J.Sci.Comp., 2003) : φ K j (u h) = φ K (u h ) = j K K F h (u h ) ˆndl for some continuous approximation of the flux F h. MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

50 Conservation Conservation, LW theorem for RD and F(u) = 0 φ K j (u h) = φ K (u h ) = F h (u h ) ˆndl j K for some continuous approximation of the flux F h. K In practice, approach 1 Set F h = j K ψ jf j and integrate exactly. This gives the P 1 element residual seen for the FV scheme MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

51 Conservation Conservation, LW theorem for RD and F(u) = 0 φ K j (u h ) = φ K (u h ) = F h (u h ) ˆndl j K for some continuous approximation of the flux F h. In practice, approach 2 Set F h = F(v h ) with v some set of variables, v h = j ψ jv j, apply Gauss Formulae on K (Csik, Ricchiuto, Deconinck, J.Comput.Phys, 2002) K φ K (u h ) = f K f F h (x) ˆndl = G p f ω q F h (x q ) ˆn f f K q=1 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

52 Conservation Conservation, LW theorem for RD and F(u) = 0 φ K j (u h ) = φ K (u h ) = F h (u h ) ˆndl j K for some continuous approximation of the flux F h. In practice, approach 3 Set F h = F(v h ) with v some set of variables, v h = j ψ jv j, and integrate exactly of below truncation error (Deconinck, Struijs, Roe K Computers & Fluids, 1993 ; Abgrall, Barth SISC, 2002) φ K = K F v (v h) v h dk MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

53 Conservation Conservation, LW theorem for RD and F(u) = 0 φ K j (u h) = φ K (u h ) = F h (u h ) ˆndl j K for some continuous approximation of the flux F h. In practice Schemes are more often written so that we recover at the end φ K (u h ) = f K f K F h (x) ˆndl = f K G p f ω q F h (x q ) ˆn f q=1 for some (edge) continuous polynomial reconstruction F h (x) which remains one of the degrees of freedom of the method MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

54 Conservation Conservation, LW theorem for RD and F(u) = 0 φ K j (u h) = φ K (u h ) = F h (u h ) ˆndl j K for some continuous approximation of the flux F h. This only guarantees that if discontinuous solutions are approximated, the correct jump (Rankine-Hugoniot) conditions are recovered K What about stability, accuracy, etc.? MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

55 3 DESIGN CRITERIA Accuracy, stability, and all that jazz MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

56 Design criteria 1. K Ω h compute : φ K = F h (u h ) K φ K i i K 2. Distribution : φ K = Distribution coeff.s : φ K i =β K i φ K K φ K φ K 1 3. Compute nodal values : solve algebraic system C i du i dt + T i T φ K i = 0, t (4) φ K 2 φ K 3 i MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

57 What can we say about the stability of this method? First : what is stability? MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

58 What can we say about the stability of this method? First : what is stability? Assume, for h fixed you do u n+1 i = u n i ω i K i K φ K i (u n h ), ω i = t C i MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

59 What can we say about the stability of this method? First : what is stability? More abstractly (ω a scalar,e.g. ω = min i ω i ) for h fixed you do u n+1 = u n ω(a h u n f) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

60 What can we say about the stability of this method? First : what is stability? More abstractly (ω a scalar,e.g. ω = min i ω i ) for h fixed you do u n+1 = u n ω(a h u n f) A condition for convergence with n but h fixed (I ωa h )u 2 r u 2, u and with r < 1 which is equivalent to u t A h u 1 r 2ω u 2 + ω 2 A hu 2 C h u 2 0 u MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

61 What can we say about the stability of this method? First : what is stability? More abstractly (ω a scalar,e.g. ω = min i ω i ) for h fixed you do u n+1 = u n ω(a h u n f) A condition for convergence with n but h fixed (I ωa h )u 2 r u 2, u and with r < 1 which is equivalent to u t A h u 1 r 2ω u 2 + ω 2 A hu 2 C h u 2 0 u Coercivity... MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

62 Stability and energy Consider the steady limit of t u+ a u = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

63 Stability and energy Consider the steady limit of Semi-discrete counterpart t u+ a u = 0 C i du i dt + K i K φ K i = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

64 Stability and energy Consider the steady limit of Semi-discrete counterpart t u+ a u = 0 C i du i dt + K i K φ K i = 0 Energy budget The equivalent of the quantity u t A h u seen in the previous slides is u t A h u u i φ K i i Ω h K i K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

65 Stability and energy Consider the steady limit of Semi-discrete counterpart t u+ a u = 0 C i du i dt + K i K φ K i = 0 Energy budget The equivalent of the quantity u t A h u seen in the previous slides is u t A h u u i φ K i K Ω h i K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

66 Stability and energy Consider the steady limit of Semi-discrete counterpart t u+ a u = 0 C i du i dt + K i K φ K i = 0 Energy budget The equivalent of the quantity u t A h u seen in the previous slides is u t A h u u i φ K i K Ω h i K }{{} φ E K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

67 Stability and energy Consider the steady limit of Semi-discrete counterpart t u+ a u = 0 C i du i dt + K i K φ K i = 0 Energy budget The equivalent of the quantity u t A h u seen in the previous slides is u t A h u K Ω h φ E K What is φ E K? MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

68 Stability and energy Starting from Energy budget C i du i dt + K i K du i C i u i dt + i Ω h φ K i = 0 K Ω h φ E K = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

69 Stability and energy Starting from C i du i dt + K i K φ K i = 0 Energy budget i Ω h C i d dt ( ) u 2 i + φ E K 2 = 0 K Ω h MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

70 Stability and energy Starting from Energy budget C i du i dt + Ω h with the energy density and with E h = K i K de h dt + φ K i = 0 K Ω h φ E K = 0 E = u2 2 i Ω h E i ψ i (piecewise linear) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

71 Stability and energy Saying that 0 u t A h u is equivalent to K Ω h φ E K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

72 Stability and energy Saying that 0 < u t A h u K Ω h φ E K is equivalent to Energy stability Ω h de h dt = K Ω h φ E K 0 with the energy density and with E h = E = u2 2 i Ω h E i ψ i (piecewise linear) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

73 Stability and energy Saying that is equivalent to 0 < u t A h u K Ω h φ E K Energy stability (modulo boundary conditions) Ω h de h dt = Ω h E h a ˆndl δ E, δ E 0 what one would like is to find that φ E K = E h a ˆndl+δ K E, δe K 0 K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

74 Stability and upwinding Consider again the steady limit of t u+ a u = 0 A geometrical view of advection... MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

75 Stability and upwinding Consider again the steady limit of t u+ a u = 0 A geometrical view of advection... 1-target triangle The inlet region is an edge 1 node downstream : 1 target k j = a n j 2 > 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

76 Stability and upwinding Consider again the steady limit of t u+ a u = 0 A geometrical view of advection... 1-target triangle The inlet region is an edge 1 node downstream : 1 target k j = a n j 2 > 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

77 Stability and upwinding Consider again the steady limit of t u+ a u = 0 A geometrical view of advection... 2-target triangle The outlet region is an edge 2 nodes downstream : 2 targets k j = a n j 2 > 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

78 Stability and upwinding Consider again the steady limit of t u+ a u = 0 A geometrical view of advection... 2-target triangle The outlet region is an edge 2 nodes downstream : 2 targets k j = a n j 2 > 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

79 Stability and upwinding Consider now the semi-discrete RD advection equation : C i du i dt + K i K A geometrical view of advection... φ K i = 0 Multidimensional Upwinding (MU) Multidimensional Upwind (MU) schemes only split φ K to downstream nodes, i.e. those for which k j > 0. All MU schemes reduce to the same in the 1-target case. MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

80 Stability and upwinding Consider now the semi-discrete RD advection equation : C i du i dt + K i K A geometrical view of advection... φ K i = 0 Multidimensional Upwinding (MU) Multidimensional Upwind (MU) schemes In 1-target elements if k 1 > 0 (node 1 only node downstream) φ K 1 = φk, φ K 2 = φk 3 = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

81 Stability and upwinding Consider now the semi-discrete RD advection equation : C i du i dt + K i K A geometrical view of advection... φ K i = 0 Multidimensional Upwinding (MU) Multidimensional Upwind (MU) schemes In 2-targets elements if k 1 < 0 (node 1 only node upstream) φ K 1 = 0, φk 2 +φk 3 = φk MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

82 Stability and MU schemes Example 1 : Roe s optimal N scheme 1 2 φ K ( a 1) = K a1 u h 2 φ K ( a 2) = K a2 u h a 1 a φ N 1 = φ K φ N 2 = φ N 3 = 0 a a 1 φ N 3 = 0 φ N 1 = φ K ( a 1) φ N 2 = φ K ( a 2) The formula (Roe Cranfield U.Tech.Rep., 1987 ; Roe, Sidilkover SINUM, 1992) kj u j φ N i = k + i (u i u in ), u in = j K kj j K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

83 Stability and MU schemes Example 2 : the LDA scheme The LDA scheme reads 2 φ LDA i (u h ) = βi LDA φ K (u h ) K 3 where β LDA i = k+ i k j + j K a K 3 / K K 2 / K K 2 1 recalling that for the advection equation (u h piecewise linear) φ K (u h ) = a u h φ K 1 K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

84 Stability and MU The following properties can be easily shown : 1. MU schemes, 1-target (Deconinck, Ricchiuto Enc.Comput.Mech., 2007) φ E K = K E h a ˆndl +δ E K, δe K 0 2. N scheme energy stable (Barth, NASA 1996 ; Abgrall, Barth SISC, 2002) 3. LDA scheme, 2-targets (Deconinck, Ricchiuto Enc.Comput.Mech., 2007) φ E LDA = ( k + ) ( u 2 ) out j 2 u2 in + δ 2 LDA, E δlda E 0 j K }{{} NRG balance along streamline Multidimensional upwinding does the job... MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

85 Stability, upwinding, and dissipation 1. FV scheme (1st order upwind) NRG stable (Barth, NASA 1996 ; Abgrall, Barth SISC, 2002), also E-flux schemes by (Osher SINUM, 1984) 2. Streamline upwind finite element scheme SUPG, (Hughes, Brooks CMAME, 1982) : MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

86 Stability, upwinding, and dissipation 1. FV scheme (1st order upwind) NRG stable (Barth, NASA 1996 ; Abgrall, Barth SISC, 2002), also E-flux schemes by (Osher SINUM, 1984) 2. Streamline upwind finite element scheme (SUPG) (Hughes, Brooks CMAME, 1982) : ψ i F h (u h )+ a(u h ) ψ i τ a(u h ) u h = 0 Ω K Ω K h h can be written as the RD scheme φ K i = 0 with φ K i = K K i K ψ i F h (u h )+ a(u h ) ψ i τ a(u h ) u h K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

87 Stability, upwinding, and dissipation 1. FV scheme (1st order upwind) NRG stable (Barth, NASA 1996 ; Abgrall, Barth SISC, 2002), also E-flux schemes by (Osher SINUM, 1984) 2. Streamline upwind finite element scheme SUPG : φ SUPG i = ψ i a u h + a ψ i τ a u h K one easily checks that since j ψ j = 1 and j ψ j = 0 φ SUPG j = a u h = φ K (u h ) j K K K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

88 Stability, upwinding, and dissipation 1. FV scheme (1st order upwind) NRG stable (Barth, NASA 1996 ; Abgrall, Barth SISC, 2002), also E-flux schemes by (Osher SINUM, 1984) 2. Streamline upwind finite element scheme (SUPG) : φ E SUPG = K u 2 h 2 a ˆndl+ K a u h τ a u h } {{ } Streamline dissipation 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

89 Stability, upwinding, and dissipation 1. FV scheme (1st order upwind) NRG stable (Barth, NASA 1996 ; Abgrall, Barth SISC, 2002), also E-flux schemes by (Osher SINUM, 1984) 2. Streamline upwind finite element scheme (SUPG) : φ E SUPG = K u 2 h 2 a ˆndl+ K a u h τ a u h } {{ } Streamline dissipation 0 3. Lax-Friedrich s/rusanov scheme φ LF i = ψ i a u h +α LF (u i u j ) K j K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

90 Stability, upwinding, and dissipation 1. FV scheme (1st order upwind) NRG stable (Barth, NASA 1996 ; Abgrall, Barth SISC, 2002), also E-flux schemes by (Osher SINUM, 1984) 2. Streamline upwind finite element scheme (SUPG) : φ E SUPG = K u 2 h 2 a ˆndl+ 3. Lax-Friedrich s/rusanov scheme φ E LF = K K u 2 h 2 a ˆndl + α LF 3 a u h τ a u h } {{ } Streamline dissipation 0 (u i u j ) 2 i,j K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

91 Stability, upwinding, and dissipation Upwinding has beneficial effect in terms of energy stability MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

92 Design criteria : what is the truncation error? By Taylor expansion : no way (unless meshes with particular structure are considered) Error analysis based on variational form : 1. which variational form? 2. NRG stability not enough, no coercivity no tools for analysis Idea : use weak form to define error (consistency estimate) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

93 Design criteria : what is the truncation error? Idea : use weak form to define an integral truncation error ϕ F(u)dx+BCs = 0 Ω Ω ϕ F h (u h )dx+bcs = ε h with u a smooth exact (classical) solution This gives a consistency estimate.. What is ε h? MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

94 Design criteria, consistency analysis What do we have...? MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

95 Design criteria, consistency analysis What do we have...? Consider 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) with w h a continuous polynomial approximation of degree k (e.g standard Lagrange elements) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

96 Design criteria, consistency analysis Continuous Lagrange elements P 1 P 2 P 3 Q 1 Q 2 Q 3 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

97 Design criteria, consistency analysis What do we do...? Consider 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) Take the steady RD scheme K i K approximating F in node i φ K i (u h) = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

98 Design criteria, consistency analysis What do we do...? Consider 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) Formally replace the nodal values of u h, computed by the scheme, with those of the exact solution w, exaclty as done in finite difference TE analysis MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

99 Design criteria, consistency analysis What do we do...? Consider 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) We obtain K i K φ K i (w h ) 0 since of course the nodal values of the exact solution w do not verify the discrete equations MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

100 Design criteria, consistency analysis What do we do...? Consider 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) Given ϕ a C0 r (Ω) class function, r large enough, define ǫ h := φ K i (w h ) ϕ i i Ω h K i K A global measure of how much the discrete equations differ from the continuous one MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

101 Design criteria, consistency analysis What do we do...? Estimate ǫ h (Abgrall, Roe J.Sci.Comp., 2003 ; Ricchiuto, Abgrall, Deconinck J.Comput.Phys, 2007) ǫ h = K Ω h i K ϕ i φ K i (w h) = ǫ a +ǫ d ǫ a = ϕ h (F h (w h ) F(w)) Ω } h {{} approximation error ϕ h = ψ j ϕ j K Ω h j K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

102 Design criteria, consistency analysis What do we do...? Estimate ǫ h (Abgrall, Roe J.Sci.Comp., 2003 ; Ricchiuto, Abgrall, Deconinck J.Comput.Phys, 2007) ǫ h = K Ω h i K ϕ i φ K i (w h) = ǫ a +ǫ d ǫ d = K Ω h i,j K ϕ i ϕ j (φ K i (w h ) φ G i (w h )) n K DoF } {{ } distribution error φ G i (w h) = ψ i (F h (w h ) F(w)) (Galerkin proj.) K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

103 Design criteria, consistency analysis What do we do...? Estimate ǫ h (Abgrall, Roe J.Sci.Comp., 2003 ; Ricchiuto, Abgrall, Deconinck J.Comput.Phys, 2007) 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

104 Design criteria, consistency analysis What do we do...? Estimate ǫ h (Abgrall, Roe J.Sci.Comp., 2003 ; Ricchiuto, Abgrall, Deconinck J.Comput.Phys, 2007) 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) We get easily the estimates ǫ a = ϕ h (F h (w h ) F(w)) C ah k+1 Ω h MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

105 Design criteria, consistency analysis What do we do...? Estimate ǫ h (Abgrall, Roe J.Sci.Comp., 2003 ; Ricchiuto, Abgrall, Deconinck J.Comput.Phys, 2007) 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) We get easily the estimates φ G (w h ) = K ψ i (F h (w h ) F(w)) C a hk+2 ϕ i ϕ j h ϕ Ch MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

106 Design criteria, consistency analysis What do we do...? Estimate ǫ h (Abgrall, Roe J.Sci.Comp., 2003 ; Ricchiuto, Abgrall, Deconinck J.Comput.Phys, 2007) 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) We get easily the estimates ǫ d = ϕ i ϕ j K Ω h i,j K n K (φ K i (w h) φ G i (w h)) DoF C Ωh h 2 Ch (sup K sup i K φ K i (w h) +C a hk+2 ) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

107 Design criteria, consistency analysis What do we do...? Estimate ǫ h (Abgrall, Roe J.Sci.Comp., 2003 ; Ricchiuto, Abgrall, Deconinck J.Comput.Phys, 2007) 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) in L 2 from approximation theory, see e.g. (Ern, Guermond Springer, 2004) We get easily the estimates ǫ d C h 1 sup K sup φ K i (w h ) +C a h k+1 i K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

108 Design criteria, consistency analysis What do we do...? Estimate ǫ h (Abgrall, Roe J.Sci.Comp., 2003 ; Ricchiuto, Abgrall, Deconinck J.Comput.Phys, 2007) ǫ h C a h k+1 +C h 1 sup K sup φ K i (w h ) i K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

109 Design criteria, consistency analysis What do we do...? Estimate ǫ h (Abgrall, Roe J.Sci.Comp., 2003 ; Ricchiuto, Abgrall, Deconinck J.Comput.Phys, 2007) ǫ h C a h k+1 +C h 1 sup K sup φ K i (w h ) i K For a polynomial approximation of degree k, a sufficient condition to have a ǫ h Ch k+1 is (in 2d) φ K i (w h ) = O(h k+2 ), K h, i K A local TE condition.. MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

110 Design criteria, high order schemes For a polynomial approximation of degree k, a sufficient condition to have a ǫ h Ch k+1 is (in 2d) φ K i (w h ) = O(h k+2 ), K h, i K 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

111 Design criteria, high order schemes For a polynomial approximation of degree k, a sufficient condition to have a ǫ h Ch k+1 is (in 2d) φ K i (w h ) = O(h k+2 ), K h, i K 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) High order prototype 1, Petrov-Galerkin φ K i (u h) = ωi K F h(u h ), ωi K < C < K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

112 Design criteria, high order schemes For a polynomial approximation of degree k, a sufficient condition to have a ǫ h Ch k+1 is (in 2d) φ K i (w h ) = O(h k+2 ), K h, i K 1. w H k+1 smooth solution : F(w) = u F(w) w = 0 2. w w h = O(h k+1 ), F(w) F h (w h ) = O(h k+1 ) 3. (w w h ) = O(h k ), (F(w) F h (w h )) = O(h k ) High order prototype 2, accuracy preserving RD φ K i (u h ) = βi K F h (u h ) = βi K φ K (u h ), βi K < C < K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

113 High order schemes, examples LDA scheme (P 1 ) elements β LDA i = k + i ( j K k + j ) 1 SUPG ω K i = ψ i + a(u h ) ψ i τ, a(u h ) = u F(u h ) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

114 4 LIMITERS using them in reverse MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

115 Nonlinear high order schemes So far we have 1. A stability criterion requiring an upwind bias (other stabilization strategies mentioned later if time..) 2. An accuracy (consistency) criterion requiring bounded weights in the residual splitting How about discontinuity capturing? MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

116 Discontinuity capturing : positivity C i du i dt = K i K Positive coefficient scheme (Spekreijse, Math.Comp. 49, 1987) A scheme for which φ K i = c K ij(u i u j ) with c K ik 0 j K j i s said to be LED (Local Extremum Diminishing) φ K i MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

117 Discontinuity capturing : positivity C i un+1 i t u n i = K i K c K ij (un i un j ) j K j i Positive coefficient scheme (Spekreijse, Math.Comp. 49, 1987) When combined with Explicit Euler time integration (or with another boundedness preserving time integration scheme) LED leads to c ij u n j u n+1 i = j where c ij 0, j c K ij = 1 provided t C i c ij 1 j In this case the scheme is said (by abuse of language) to be positive MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

118 Discontinuity capturing : positivity C i un+1 i t u n i = K i K c K ij (un i un j ) j K j i Positive coefficient scheme (Spekreijse, Math.Comp. 49, 1987) When combined with Explicit Euler time integration (or with another boundedness preserving time integration scheme) LED leads to c ij u n j u n+1 i = j where c ij 0, j c K ij = 1 provided t C i c ij 1 j In this case the scheme is said (by abuse of language) to be positive MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

119 Discontinuity capturing : positivity u n+1 i = j c ij u n j with c ij 0, c K ij = 1 j Positive coefficient scheme (Spekreijse, Math.Comp. 49, 1987) A positive scheme verifies the discrete max principle minu n j un+1 i j maxu n j j MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

120 Positive schemes : examples Example 1 : Roe s optimal N scheme 1 2 φ K ( a 1) = K a1 u h 2 φ K ( a 2) = K a2 u h a 1 a φ N 1 = φ K φ N 2 = φ N 3 = 0 a a 1 φ N 3 = 0 φ N 1 = φ K ( a 1) φ N 2 = φ K ( a 2) The formula (Roe Cranfield U.Tech.Rep., 1987 ; Roe, Sidilkover SINUM, 1992) kj u j φ N i = k + i (u i u in ), u in = j K kj j K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

121 Nonlinear high order schemes Example 2 : Lax-Friedrich s distribution 2 φ T 1 LF φi = K ψ i F h +α LF (u i u j ) j K for positivity (scalar case) 3 α LF h K sup u F(u h (x)) x K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

122 Nonlinear high order schemes Bad news... (Godunov) All linear positive (LED) schemes are first order accurate... MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

123 Nonlinear high order schemes Where does the limiter come in Recall that one prototype of a high order scheme is φ K i (u h) = β K i φ K (u h ), β K i C < MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

124 Nonlinear high order schemes Where does the limiter come in Recall that one prototype of a high order scheme is φ K i (u h) = β K i φ K (u h ), β K i C < For linear positive coefficient schemes φ P i (u h) = j Kc K ij (u i u j ), c K ij 0 Formally we have c K ij (u i u j ) β P i (u h) = j K φ K (u h ) in general unbounded! MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

125 Nonlinear high order schemes Where does the limiter come in The idea : apply a limiter function to bound the distribution coefficient MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

126 Nonlinear high order schemes Where does the limiter come in The idea : apply a limiter function to bound the distribution coefficient βi LP (u h ) = ψ( βi P(u h) ) ψ ( βj P(u h) ) j K The scaling on the denominator guarantees that j β LP i = 1 What are the conditions on the limiter function ψ( )? MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

127 Nonlinear high order schemes Where does the limiter come in Linear positive coefficient schemes φ P i (u h) = j K β LP c K ij (u i u j ),c K ij 0 ; βp i (u h) = φp i (u h) φ K (u h ) unbounded i (u h ) = ψ( βi P(u h) ) ψ ( βj P(u ) limited distribution coefficient h) j K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

128 Nonlinear high order schemes Where does the limiter come in Linear positive coefficient schemes φ P i (u h ) = j K β LP c K ij(u i u j ),c K ij 0 ; β P i (u h ) = φp i (u h) φ K (u h ) unbounded i (u h ) = ψ( βi P(u h) ) ψ ( βj P(u ) limited distribution coefficient h) j K φ LP i (u h ) = βi LP φ K = βlp i β P i Provided ψ(r) 0 and ψ(r) r }{{} γ P i 0 φ P i = j K 0 we have c LP ij (u i u j ), c LP ij = γ P i c K ij 0! MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

129 High order schemes High order RD scheme 1. Compute cell residual φ K = K F h (u h ) dl 2. Compute linear positive distribution φ P i = j ck ij (u i u j ) 3. Limit β P i = φ P i /φk β LP i = ψ(β P i )/( j ψ(βp j )) 4. Distribute cell residual φ K i = β LP i φ K 5. Evolve C i du i dt = K i K φ K i until steady state MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

130 High order schemes High order RD scheme 1. Compute cell residual φ K = K F h (u h ) dl 2. Compute linear positive distribution φ P i = j ck ij (u i u j ) 3. Limit β P i = φ P i /φk β LP i = ψ(β P i )/( j ψ(βp j )) 4. Distribute cell residual φ K i = β LP i φ K 5. Evolve C i du i dt = K i K φ K i until steady state The simplest possible choice for ψ( ) is ψ(r) = max(0, r) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

131 Examples Rotational advection Scalar example : a u = 0 with a = (y,1 x) and bcs { cos(2π(x+0.5)) 2 if x [ 0.75, 0.25] u in = 0 otherwise y x inlet outlet MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

132 y y Examples (cont d) Rotational advection N and Limited N (LN) schemes N scheme 1 N LN Exact 0.8 outlet data x u LN scheme x x MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

133 y y Examples (cont d) Rotational advection LF and Limited LF (LLF) schemes LF scheme 1 LF LLF Exact 0.8 outlet data x u LLF scheme x x MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

134 Examples (cont d) Burger s equation Scalar example : F(u) = 0 with F(u) = (u, u2 2 ) and bcs u(x,y = 0) = 1.5 2x u in MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

135 y y Examples (cont d) Burger s equation N scheme N and Limited N (LN) schemes y = x 1 LN scheme u y = N LN x x MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

136 y y Examples (cont d) Burger s equation LF scheme LF and Limited LF (LLF) schemes y = x LLF scheme u LF LLF y = x x MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

137 Remarks on extension to systems Historical perspective Two approaches (Roe J.Comput.Phys, 1986 ; Nishikawa, Rad, Roe AIAA Conf. 2001) and (van der Weide, Deconinck Comput.Fluid Dyn., Wiley 1996) 1. Local projection (wave decomposition) of the continuous PDE to obtain (possibly decoupled) scalar equations discretized independently 2. Formal matrix generalization in which the scalar flux vector is replaced by a tensor and the k i = a n i /2 coefficients become matrix flux Jacobians MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

138 Remarks on extension to systems Practical implementation Hybrid of the two (Abgrall, Mezine J.Comput.Phys, 2004 ; Ricchiuto, Csik, Deconinck J.Comput.Phys, 2005) : Matrix formulation for linear first order schemes Projection onto characteristic directions to obtain scalar residuals to work with for the limiting procedure (similar to FV limiting on characteristic var.s) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

139 Example 1 : Mach 3.6 scramjet inlet (Euler, perfect gas) Mesh N scheme Nonlinear scheme MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

140 Example 2 : Mach 10 bow shock (Euler, perfect gas) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

141 6 ON THE RELATIONS WITH FEM, FV, DG, WENO FD, etc. etc. MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

142 Relations with other techniques Continuous, Stabilized Finite Elements By nature of the underlying approximation, these methods bear close resemblance to stabilized continuous Galerkin methods as e.g. the SUPG of (Hughes, Brooks CMAME, 1982) (bcs omitted) ψ i F h (u h )+ a(u h ) ψ i τ a(u h ) u h = 0 Ω K Ω K h h which, as seen, can be written as the RD scheme φ K i = 0 with φ K i = ψ i F h (u h )+ a(u h ) ψ i τ a(u h ) u h K i K K K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

143 Relations with other techniques Continuous, Stabilized Finite Elements More generally, a Petrov-Galerkin method with test space spanned by functions {ω i } i Ωh such that K Ω h i K = j Kω ωi K = 1 j K can be recast as a RD scheme φ K i = 0 with φ K i = K i K K ω K i F h(u h ) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

144 Relations with other techniques : RD and FV Which numerical flux defines your conservative statement? The first relation we have already seen : FV schemes can be written such that they sit in an element l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij i K ij n K ij j C i du i dt + K i Kφ K i = 0, φ K i = ( F ij F i ) n K ij j K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

145 Relations with other techniques : RD and FV Which numerical flux defines your conservative statement? The first relation we have already seen : FV schemes can be written such that they sit in an element Here conservation is expressed by the FV flux function F C i du i dt + K i Kφ K i = 0, φ K i = ( F ij F i ) n K ij j K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

146 Relations with other techniques : RD and FV Which numerical flux defines your conservative statement? A different view is to recast RD as FV on the median dual l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij C i du i dt + K i K j K i F ij n K ij = 0 K ij n K ij j MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

147 Relations with other techniques : RD and FV Which numerical flux defines your conservative statement? A different view is to recast RD as FV on the median dual l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij C i du i dt = K i K j K F ij n K ij i K ij n K ij What definition of F ij such F ij n K ij = βk i φk j K for a given φ K i = β K i φ K j MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

148 Relations with other techniques : RD and FV Which numerical flux defines your conservative statement? A different view is to recast RD as FV on the median dual l i j K n K il n i = 2( n K ij + nk il ) f C i C j f ij i K ij n K ij j C i du i dt = K i K j K F ij n K ij F RD ij What numerical flux defines local conservation on the median dual cell for RD? MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

149 Relations with other techniques : RD and FV Which numerical flux defines your conservative statement? A different view is to recast RD as FV on the median dual. C i du i dt = j K K i K j K F RD ij n K ij F RD ij n K ij = βk i ΦK, i F RD ij What numerical flux defines local conservation on the median dual cell for RD? Answer in (Abgrall, 2012) F RD ij n ij = Ψ i Ψ j with Ψ i = β K i φ K F i n i 2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

150 Relations with other techniques : RD and FV Which numerical flux defines your conservative statement? A different view is to recast RD as FV on the median dual. C i du i dt = K i K j K F RD ij n K ij F RD ij What numerical flux defines local conservation on the median dual cell for RD? Answer in (Abgrall, 2012) F RD ij n ij = Ψ i Ψ j with Ψ i = β K i φ K F i n i 2 Consistent 3-states numerical flux function This is still the RD scheme we start with MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

151 Relations with other techniques : RD and FV u i u i 1 u i+1 i 1 i i+1 i 1/2 i+1/2 FV fluxes from RD schemes Starting point : conservation law Conservative FV : t u+ x F(u) = 0 x i du i dt + F i+1/2 F i 1/2 = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

152 Relations with other techniques : RD and FV u i (x) u i 1 (x) u i+1 (x) F i+1/2 = F(u i(x i+1/2 ),u i+1(x i+1/2 )) F i 1/2 = F(u i 1(x i 1/2 ),u i(x i 1/2 )) i 1 i 1/2 i i+1 i+1/2 FV fluxes from RD schemes Conservative FV : x i du i dt + F i+1/2 F i 1/2 = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

153 Relations with other techniques : RD and FV u i 1 (x) u i (x) u i+1 (x) F i = F(u i(x i+1/2 ) F(u i(x i 1/2 )) φ i+1/2 i = F i+1/2 F(u i(x i+1/2 )) i 1 i 1/2 i i+1 i+1/2 φ i+1/2 i FV fluxes from RD schemes Conservative FV, reformulation : = F(u i+1(x i+1/2 )) F i+1/2 du i x i dt + F i +φ i+1/2 i +φ i 1/2 i = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

154 Relations with other techniques : RD and FV u i 1 (x) u i (x) u i+1 (x) F i = F(u i(x i+1/2 ) F(u i(x i 1/2 )) φ i+1/2 i = F i+1/2 F(u i(x i+1/2 )) i 1 i 1/2 i i+1 i+1/2 φ i+1/2 i FV fluxes from RD schemes Conservative FV, reformulation : = F(u i+1(x i+1/2 )) F i+1/2 du i x i dt + F i +φ i+1/2 i +φ i 1/2 i = 0 x i+1 du i+1 dt + F i+1 +φ i+3/2 i+1 +φ i+1/2 i+1 = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

155 Relations with other techniques : RD and FV u i 1 (x) u i (x) u i+1 (x) F i = F(u i(x i+1/2 ) F(u i(x i 1/2 )) φ i+1/2 i = F i+1/2 F(u i(x i+1/2 )) i 1 i 1/2 i i+1 i+1/2 φ i+1/2 i FV fluxes from RD schemes Conservative FV, reformulation : φ i+1/2 i = F(u i+1(x i+1/2 )) F i+1/2 +φ i+1/2 i+1 := φ i+1/2 = F(u i+1 (x i+1/2 )) F(u i (x i+1/2 )) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

156 Relations with other techniques : RD and FV u i (x) u i 1 (x) h l u i+1 (x) i x x + i 1/2 i 1/2 FV fluxes from RD schemes x i+1/2 x + i+1/2 Integrate over each control volume : no need for numerical flux (continuity through ghost elements) In each ghost element apply any RD scheme which will provide a definition for the numerical flux as x + i±1/2 x i±1/2 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

157 Relations with other techniques : RD and FV u i (x) u i 1 (x) h l u i+1 (x) i x x + i 1/2 i 1/2 FV fluxes from RD schemes x i+1/2 x + i+1/2 u n+1 i x i t u n i + F i +β i+1/2 i φ i+1/2 +β i 1/2 i φ i+1/2 = 0 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

158 Relations with other techniques : RD and FV u i (x) u i 1 (x) u i+1 (x) F i+1/2 = F(u i(x i+1/2 ),u i+1(x i+1/2 )) F i 1/2 = F(u i 1(x i 1/2 ),u i(x i 1/2 )) i 1 i 1/2 i i+1 i+1/2 FV fluxes from RD schemes u n+1 i x i t F RD u n i + F RD i+1/2 FRD i 1/2 = 0 i+1/2 =F(u i(x i+1/2 ))+β i+1/2 F RD i 1/2 =F(u i(x i 1/2 )) β i 1/2 i i φ i+1/2 φ i 1/2 MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

159 Relations with other techniques Other examples In (Chou, Shu J.Comput.Phys, 2006 ; Chou, Shu J.Comput.Phys, 2006) the authors propose a WENO Finite Difference scheme consisting of a RD technique using nodal WENO reconstructions instead of Lagrange approximation. The RD formulation permits to keep the simplicity of the WENO FD approach, while allowing high accurate solutions on non-smooth cartesian meshes MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

160 Relations with other techniques Other examples In (Abgrall, Shu Comm.Compu.Phys, 2009) DG schemes are recast as RD. The key is defining the fluctuation including a numerical flux F as φ K = F ˆndl A preliminary construction of a hybrid DG-RD nonlinear scheme is proposed K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

161 Relations with other techniques Other examples Residual distribution schemes based on discontinuous approximation explored in (Hubbard J.Comput.Phys, 2008 ; Abgrall Adv.Appl.Math.Mech, 2010 ; Hubbard, Ricchiuto Computers & Fluids, 2011). As before, the key is defining the fluctuation including a numerical flux F as φ K = F ˆndl RD techniques used to generate nonlinear schemes. The advantages of the discontinuous approximation are retained K MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

162 Relations with other techniques Important points 1. RD as a general framework to study weighted residual discretizations on general meshes 2. RD as a means of constructing non-oscillatory schemes 3. RD to define FV numerical fluxes, interesting applications in presence of source terms (see Part II) MARIO RICCHIUTO - Residual Distribution, Part I (CEMRACS 2012) July 19,

163 .. to be continued... M.R. web : mricchiu BACCHUS Inria team :

RECENT DEVELOPMENTS IN VERY HIGH ORDER RESIDUAL DISTRIBUTION SCHEMES

RECENT DEVELOPMENTS IN VERY HIGH ORDER RESIDUAL DISTRIBUTION SCHEMES RECENT DEVELOPMENTS IN VERY HIGH ORDER RESIDUAL DISTRIBUTION SCHEMES FOR INVISCID AND VISCOUS PROBLEMS. R. Abgrall Team Bacchus INRIA Bordeaux Sud Ouest and UniversitédeBordeauxTalence,France Roscoff,

More information

Divergence Formulation of Source Term

Divergence Formulation of Source Term Preprint accepted for publication in Journal of Computational Physics, 2012 http://dx.doi.org/10.1016/j.jcp.2012.05.032 Divergence Formulation of Source Term Hiroaki Nishikawa National Institute of Aerospace,

More information

Construction of very high order Residual Distribution Schemes for steady problems

Construction of very high order Residual Distribution Schemes for steady problems Construction of very high order Residual Distribution Schemes for steady problems Rémi Abgrall, Mario Ricchiuto, Cédric Tavé, Nadège Villedieu and Herman Deconinck Mathématiques Appliquées de Bordeaux,

More information

Discontinuous Galerkin methods Lecture 2

Discontinuous Galerkin methods Lecture 2 y y RMMC 2008 Discontinuous Galerkin methods Lecture 2 1 Jan S Hesthaven Brown University Jan.Hesthaven@Brown.edu y 1 0.75 0.5 0.25 0-0.25-0.5-0.75 y 0.75-0.0028-0.0072-0.0117 0.5-0.0162-0.0207-0.0252

More information

Hierarchical Reconstruction with up to Second Degree Remainder for Solving Nonlinear Conservation Laws

Hierarchical Reconstruction with up to Second Degree Remainder for Solving Nonlinear Conservation Laws Hierarchical Reconstruction with up to Second Degree Remainder for Solving Nonlinear Conservation Laws Dedicated to Todd F. Dupont on the occasion of his 65th birthday Yingjie Liu, Chi-Wang Shu and Zhiliang

More information

The RAMSES code and related techniques I. Hydro solvers

The RAMSES code and related techniques I. Hydro solvers The RAMSES code and related techniques I. Hydro solvers Outline - The Euler equations - Systems of conservation laws - The Riemann problem - The Godunov Method - Riemann solvers - 2D Godunov schemes -

More information

Weighted Essentially Non-Oscillatory limiters for Runge-Kutta Discontinuous Galerkin Methods

Weighted Essentially Non-Oscillatory limiters for Runge-Kutta Discontinuous Galerkin Methods Weighted Essentially Non-Oscillatory limiters for Runge-Kutta Discontinuous Galerkin Methods Jianxian Qiu School of Mathematical Science Xiamen University jxqiu@xmu.edu.cn http://ccam.xmu.edu.cn/teacher/jxqiu

More information

Improvement of convergence to steady state solutions of Euler equations with. the WENO schemes. Abstract

Improvement of convergence to steady state solutions of Euler equations with. the WENO schemes. Abstract Improvement of convergence to steady state solutions of Euler equations with the WENO schemes Shuhai Zhang, Shufen Jiang and Chi-Wang Shu 3 Abstract The convergence to steady state solutions of the Euler

More information

Chapter 1. Introduction

Chapter 1. Introduction Chapter 1 Introduction Many astrophysical scenarios are modeled using the field equations of fluid dynamics. Fluids are generally challenging systems to describe analytically, as they form a nonlinear

More information

Hierarchical Reconstruction with up to Second Degree Remainder for Solving Nonlinear Conservation Laws

Hierarchical Reconstruction with up to Second Degree Remainder for Solving Nonlinear Conservation Laws Hierarchical Reconstruction with up to Second Degree Remainder for Solving Nonlinear Conservation Laws Dedicated to Todd F. Dupont on the occasion of his 65th birthday Yingjie Liu, Chi-Wang Shu and Zhiliang

More information

Riemann Solvers and Numerical Methods for Fluid Dynamics

Riemann Solvers and Numerical Methods for Fluid Dynamics Eleuterio R Toro Riemann Solvers and Numerical Methods for Fluid Dynamics A Practical Introduction With 223 Figures Springer Table of Contents Preface V 1. The Equations of Fluid Dynamics 1 1.1 The Euler

More information

Math 660-Lecture 23: Gudonov s method and some theories for FVM schemes

Math 660-Lecture 23: Gudonov s method and some theories for FVM schemes Math 660-Lecture 3: Gudonov s method and some theories for FVM schemes 1 The idea of FVM (You can refer to Chapter 4 in the book Finite volume methods for hyperbolic problems ) Consider the box [x 1/,

More information

Finite Volume Schemes: an introduction

Finite Volume Schemes: an introduction Finite Volume Schemes: an introduction First lecture Annamaria Mazzia Dipartimento di Metodi e Modelli Matematici per le Scienze Applicate Università di Padova mazzia@dmsa.unipd.it Scuola di dottorato

More information

Numerical Solutions for Hyperbolic Systems of Conservation Laws: from Godunov Method to Adaptive Mesh Refinement

Numerical Solutions for Hyperbolic Systems of Conservation Laws: from Godunov Method to Adaptive Mesh Refinement Numerical Solutions for Hyperbolic Systems of Conservation Laws: from Godunov Method to Adaptive Mesh Refinement Romain Teyssier CEA Saclay Romain Teyssier 1 Outline - Euler equations, MHD, waves, hyperbolic

More information

A class of the fourth order finite volume Hermite weighted essentially non-oscillatory schemes

A class of the fourth order finite volume Hermite weighted essentially non-oscillatory schemes Science in China Series A: Mathematics Aug., 008, Vol. 51, No. 8, 1549 1560 www.scichina.com math.scichina.com www.springerlink.com A class of the fourth order finite volume Hermite weighted essentially

More information

AMS 529: Finite Element Methods: Fundamentals, Applications, and New Trends

AMS 529: Finite Element Methods: Fundamentals, Applications, and New Trends AMS 529: Finite Element Methods: Fundamentals, Applications, and New Trends Lecture 25: Introduction to Discontinuous Galerkin Methods Xiangmin Jiao SUNY Stony Brook Xiangmin Jiao Finite Element Methods

More information

Entropy stable high order discontinuous Galerkin methods. for hyperbolic conservation laws

Entropy stable high order discontinuous Galerkin methods. for hyperbolic conservation laws Entropy stable high order discontinuous Galerkin methods for hyperbolic conservation laws Chi-Wang Shu Division of Applied Mathematics Brown University Joint work with Tianheng Chen, and with Yong Liu

More information

NUMERICAL SOLUTION OF HYPERBOLIC PARTIAL DIFFERENTIAL EQUATIONS

NUMERICAL SOLUTION OF HYPERBOLIC PARTIAL DIFFERENTIAL EQUATIONS NUMERICAL SOLUTION OF HYPERBOLIC PARTIAL DIFFERENTIAL EQUATIONS JOHN A. TRANGENSTEIN Department of Mathematics, Duke University Durham, NC 27708-0320 Ш CAMBRIDGE ЩР UNIVERSITY PRESS Contents 1 Introduction

More information

High Order Accurate Runge Kutta Nodal Discontinuous Galerkin Method for Numerical Solution of Linear Convection Equation

High Order Accurate Runge Kutta Nodal Discontinuous Galerkin Method for Numerical Solution of Linear Convection Equation High Order Accurate Runge Kutta Nodal Discontinuous Galerkin Method for Numerical Solution of Linear Convection Equation Faheem Ahmed, Fareed Ahmed, Yongheng Guo, Yong Yang Abstract This paper deals with

More information

AA214B: NUMERICAL METHODS FOR COMPRESSIBLE FLOWS

AA214B: NUMERICAL METHODS FOR COMPRESSIBLE FLOWS AA214B: NUMERICAL METHODS FOR COMPRESSIBLE FLOWS 1 / 59 AA214B: NUMERICAL METHODS FOR COMPRESSIBLE FLOWS The Finite Volume Method These slides are partially based on the recommended textbook: Culbert B.

More information

HIGH ORDER NUMERICAL METHODS FOR TIME DEPENDENT HAMILTON-JACOBI EQUATIONS

HIGH ORDER NUMERICAL METHODS FOR TIME DEPENDENT HAMILTON-JACOBI EQUATIONS June 6, 7 :7 WSPC/Lecture Notes Series: 9in x 6in chapter HIGH ORDER NUMERICAL METHODS FOR TIME DEPENDENT HAMILTON-JACOBI EQUATIONS Chi-Wang Shu Division of Applied Mathematics, Brown University Providence,

More information

Advection / Hyperbolic PDEs. PHY 604: Computational Methods in Physics and Astrophysics II

Advection / Hyperbolic PDEs. PHY 604: Computational Methods in Physics and Astrophysics II Advection / Hyperbolic PDEs Notes In addition to the slides and code examples, my notes on PDEs with the finite-volume method are up online: https://github.com/open-astrophysics-bookshelf/numerical_exercises

More information

30 crete maximum principle, which all imply the bound-preserving property. But most

30 crete maximum principle, which all imply the bound-preserving property. But most 3 4 7 8 9 3 4 7 A HIGH ORDER ACCURATE BOUND-PRESERVING COMPACT FINITE DIFFERENCE SCHEME FOR SCALAR CONVECTION DIFFUSION EQUATIONS HAO LI, SHUSEN XIE, AND XIANGXIONG ZHANG Abstract We show that the classical

More information

A Multi-Dimensional Limiter for Hybrid Grid

A Multi-Dimensional Limiter for Hybrid Grid APCOM & ISCM 11-14 th December, 2013, Singapore A Multi-Dimensional Limiter for Hybrid Grid * H. W. Zheng ¹ 1 State Key Laboratory of High Temperature Gas Dynamics, Institute of Mechanics, Chinese Academy

More information

A recovery-assisted DG code for the compressible Navier-Stokes equations

A recovery-assisted DG code for the compressible Navier-Stokes equations A recovery-assisted DG code for the compressible Navier-Stokes equations January 6 th, 217 5 th International Workshop on High-Order CFD Methods Kissimmee, Florida Philip E. Johnson & Eric Johnsen Scientific

More information

A parametrized maximum principle preserving flux limiter for finite difference RK-WENO schemes with applications in incompressible flows.

A parametrized maximum principle preserving flux limiter for finite difference RK-WENO schemes with applications in incompressible flows. A parametrized maximum principle preserving flux limiter for finite difference RK-WENO schemes with applications in incompressible flows Tao Xiong Jing-ei Qiu Zhengfu Xu 3 Abstract In Xu [] a class of

More information

A NUMERICAL STUDY FOR THE PERFORMANCE OF THE RUNGE-KUTTA FINITE DIFFERENCE METHOD BASED ON DIFFERENT NUMERICAL HAMILTONIANS

A NUMERICAL STUDY FOR THE PERFORMANCE OF THE RUNGE-KUTTA FINITE DIFFERENCE METHOD BASED ON DIFFERENT NUMERICAL HAMILTONIANS A NUMERICAL STUDY FOR THE PERFORMANCE OF THE RUNGE-KUTTA FINITE DIFFERENCE METHOD BASED ON DIFFERENT NUMERICAL HAMILTONIANS HASEENA AHMED AND HAILIANG LIU Abstract. High resolution finite difference methods

More information

Info. No lecture on Thursday in a week (March 17) PSet back tonight

Info. No lecture on Thursday in a week (March 17) PSet back tonight Lecture 0 8.086 Info No lecture on Thursday in a week (March 7) PSet back tonight Nonlinear transport & conservation laws What if transport becomes nonlinear? Remember: Nonlinear transport A first attempt

More information

A New Fourth-Order Non-Oscillatory Central Scheme For Hyperbolic Conservation Laws

A New Fourth-Order Non-Oscillatory Central Scheme For Hyperbolic Conservation Laws A New Fourth-Order Non-Oscillatory Central Scheme For Hyperbolic Conservation Laws A. A. I. Peer a,, A. Gopaul a, M. Z. Dauhoo a, M. Bhuruth a, a Department of Mathematics, University of Mauritius, Reduit,

More information

A numerical study of SSP time integration methods for hyperbolic conservation laws

A numerical study of SSP time integration methods for hyperbolic conservation laws MATHEMATICAL COMMUNICATIONS 613 Math. Commun., Vol. 15, No., pp. 613-633 (010) A numerical study of SSP time integration methods for hyperbolic conservation laws Nelida Črnjarić Žic1,, Bojan Crnković 1

More information

Sung-Ik Sohn and Jun Yong Shin

Sung-Ik Sohn and Jun Yong Shin Commun. Korean Math. Soc. 17 (2002), No. 1, pp. 103 120 A SECOND ORDER UPWIND METHOD FOR LINEAR HYPERBOLIC SYSTEMS Sung-Ik Sohn and Jun Yong Shin Abstract. A second order upwind method for linear hyperbolic

More information

Fourier analysis for discontinuous Galerkin and related methods. Abstract

Fourier analysis for discontinuous Galerkin and related methods. Abstract Fourier analysis for discontinuous Galerkin and related methods Mengping Zhang and Chi-Wang Shu Abstract In this paper we review a series of recent work on using a Fourier analysis technique to study the

More information

AN OPTIMALLY ACCURATE SPECTRAL VOLUME FORMULATION WITH SYMMETRY PRESERVATION

AN OPTIMALLY ACCURATE SPECTRAL VOLUME FORMULATION WITH SYMMETRY PRESERVATION AN OPTIMALLY ACCURATE SPECTRAL VOLUME FORMULATION WITH SYMMETRY PRESERVATION Fareed Hussain Mangi*, Umair Ali Khan**, Intesab Hussain Sadhayo**, Rameez Akbar Talani***, Asif Ali Memon* ABSTRACT High order

More information

Zonal modelling approach in aerodynamic simulation

Zonal modelling approach in aerodynamic simulation Zonal modelling approach in aerodynamic simulation and Carlos Castro Barcelona Supercomputing Center Technical University of Madrid Outline 1 2 State of the art Proposed strategy 3 Consistency Stability

More information

Block-Structured Adaptive Mesh Refinement

Block-Structured Adaptive Mesh Refinement Block-Structured Adaptive Mesh Refinement Lecture 2 Incompressible Navier-Stokes Equations Fractional Step Scheme 1-D AMR for classical PDE s hyperbolic elliptic parabolic Accuracy considerations Bell

More information

2.2. Methods for Obtaining FD Expressions. There are several methods, and we will look at a few:

2.2. Methods for Obtaining FD Expressions. There are several methods, and we will look at a few: .. Methods for Obtaining FD Expressions There are several methods, and we will look at a few: ) Taylor series expansion the most common, but purely mathematical. ) Polynomial fitting or interpolation the

More information

arxiv: v1 [math.na] 11 Jun 2018

arxiv: v1 [math.na] 11 Jun 2018 High-order residual distribution scheme for the time-dependent Euler equations of fluid dynamics arxiv:1806.03986v1 [math.na] 11 Jun 2018 R. Abgrall, P. Bacigaluppi, S. Tokareva Institute of Mathematics,

More information

Order of Convergence of Second Order Schemes Based on the Minmod Limiter

Order of Convergence of Second Order Schemes Based on the Minmod Limiter Order of Convergence of Second Order Schemes Based on the Minmod Limiter Boan Popov and Ognian Trifonov July 5, 005 Abstract Many second order accurate non-oscillatory schemes are based on the Minmod limiter,

More information

ALGEBRAIC FLUX CORRECTION FOR FINITE ELEMENT DISCRETIZATIONS OF COUPLED SYSTEMS

ALGEBRAIC FLUX CORRECTION FOR FINITE ELEMENT DISCRETIZATIONS OF COUPLED SYSTEMS Int. Conf. on Computational Methods for Coupled Problems in Science and Engineering COUPLED PROBLEMS 2007 M. Papadrakakis, E. Oñate and B. Schrefler (Eds) c CIMNE, Barcelona, 2007 ALGEBRAIC FLUX CORRECTION

More information

TOTAL VARIATION DIMINISHING RUNGE-KUTTA SCHEMES

TOTAL VARIATION DIMINISHING RUNGE-KUTTA SCHEMES MATHEMATICS OF COMPUTATION Volume 67 Number 221 January 1998 Pages 73 85 S 0025-5718(98)00913-2 TOTAL VARIATION DIMINISHING RUNGE-KUTTA SCHEMES SIGAL GOTTLIEB AND CHI-WANG SHU Abstract. In this paper we

More information

ENO and WENO schemes. Further topics and time Integration

ENO and WENO schemes. Further topics and time Integration ENO and WENO schemes. Further topics and time Integration Tefa Kaisara CASA Seminar 29 November, 2006 Outline 1 Short review ENO/WENO 2 Further topics Subcell resolution Other building blocks 3 Time Integration

More information

Positivity-preserving high order schemes for convection dominated equations

Positivity-preserving high order schemes for convection dominated equations Positivity-preserving high order schemes for convection dominated equations Chi-Wang Shu Division of Applied Mathematics Brown University Joint work with Xiangxiong Zhang; Yinhua Xia; Yulong Xing; Cheng

More information

Algebraic flux correction and its application to convection-dominated flow. Matthias Möller

Algebraic flux correction and its application to convection-dominated flow. Matthias Möller Algebraic flux correction and its application to convection-dominated flow Matthias Möller matthias.moeller@math.uni-dortmund.de Institute of Applied Mathematics (LS III) University of Dortmund, Germany

More information

An Introduction to the Discontinuous Galerkin Method

An Introduction to the Discontinuous Galerkin Method An Introduction to the Discontinuous Galerkin Method Krzysztof J. Fidkowski Aerospace Computational Design Lab Massachusetts Institute of Technology March 16, 2005 Computational Prototyping Group Seminar

More information

On the Comparison of the Finite Volume and Discontinuous Galerkin Methods

On the Comparison of the Finite Volume and Discontinuous Galerkin Methods Diploma Thesis Institute for Numerical Simulation, TUHH On the Comparison of the Finite Volume and Discontinuous Galerkin Methods Corrected version Katja Baumbach August 17, 2006 Supervisor: Prof. Dr.

More information

Dedicated to the 70th birthday of Professor Lin Qun

Dedicated to the 70th birthday of Professor Lin Qun Journal of Computational Mathematics, Vol.4, No.3, 6, 39 5. ANTI-DIFFUSIVE FINITE DIFFERENCE WENO METHODS FOR SHALLOW WATER WITH TRANSPORT OF POLLUTANT ) Zhengfu Xu (Department of Mathematics, Pennsylvania

More information

Finite Volume Method

Finite Volume Method Finite Volume Method An Introduction Praveen. C CTFD Division National Aerospace Laboratories Bangalore 560 037 email: praveen@cfdlab.net April 7, 2006 Praveen. C (CTFD, NAL) FVM CMMACS 1 / 65 Outline

More information

FUNDAMENTALS OF LAX-WENDROFF TYPE APPROACH TO HYPERBOLIC PROBLEMS WITH DISCONTINUITIES

FUNDAMENTALS OF LAX-WENDROFF TYPE APPROACH TO HYPERBOLIC PROBLEMS WITH DISCONTINUITIES 6th European Conference on Computational Mechanics (ECCM 6) 7th European Conference on Computational Fluid Dynamics (ECFD 7) 1115 June 018, Glasgow, UK FUNDAMENTALS OF LAX-WENDROFF TYPE APPROACH TO HYPERBOLIC

More information

Eulerian interface-sharpening methods for hyperbolic problems

Eulerian interface-sharpening methods for hyperbolic problems Eulerian interface-sharpening methods for hyperbolic problems Application to compressible multiphase flow Keh-Ming Shyue Department of Mathematics National Taiwan University Taiwan 11:55-12:25, March 05,

More information

Entropy stable schemes for degenerate convection-diffusion equations

Entropy stable schemes for degenerate convection-diffusion equations Entropy stable schemes for degenerate convection-diffusion equations Silvia Jerez 1 Carlos Parés 2 ModCompShock, Paris 6-8 Decmber 216 1 CIMAT, Guanajuato, Mexico. 2 University of Malaga, Málaga, Spain.

More information

Projection Dynamics in Godunov-Type Schemes

Projection Dynamics in Godunov-Type Schemes JOURNAL OF COMPUTATIONAL PHYSICS 142, 412 427 (1998) ARTICLE NO. CP985923 Projection Dynamics in Godunov-Type Schemes Kun Xu and Jishan Hu Department of Mathematics, Hong Kong University of Science and

More information

Angular momentum preserving CFD on general grids

Angular momentum preserving CFD on general grids B. Després LJLL-Paris VI Thanks CEA and ANR Chrome Angular momentum preserving CFD on general grids collaboration Emmanuel Labourasse (CEA) B. Després LJLL-Paris VI Thanks CEA and ANR Chrome collaboration

More information

Implicit Solution of Viscous Aerodynamic Flows using the Discontinuous Galerkin Method

Implicit Solution of Viscous Aerodynamic Flows using the Discontinuous Galerkin Method Implicit Solution of Viscous Aerodynamic Flows using the Discontinuous Galerkin Method Per-Olof Persson and Jaime Peraire Massachusetts Institute of Technology 7th World Congress on Computational Mechanics

More information

RESEARCH HIGHLIGHTS. WAF: Weighted Average Flux Method

RESEARCH HIGHLIGHTS. WAF: Weighted Average Flux Method RESEARCH HIGHLIGHTS (Last update: 3 rd April 2013) Here I briefly describe my contributions to research on numerical methods for hyperbolic balance laws that, in my view, have made an impact in the scientific

More information

High-resolution finite volume methods for hyperbolic PDEs on manifolds

High-resolution finite volume methods for hyperbolic PDEs on manifolds High-resolution finite volume methods for hyperbolic PDEs on manifolds Randall J. LeVeque Department of Applied Mathematics University of Washington Supported in part by NSF, DOE Overview High-resolution

More information

Runge-Kutta discontinuous Galerkin method with a simple and compact Hermite WENO limiter

Runge-Kutta discontinuous Galerkin method with a simple and compact Hermite WENO limiter Runge-Kutta discontinuous Galerkin method with a simple and compact Hermite WENO limiter Jun Zhu, inghui Zhong, Chi-Wang Shu 3 and Jianxian Qiu 4 Abstract In this paper, we propose a new type of weighted

More information

Bound-preserving high order schemes in computational fluid dynamics Chi-Wang Shu

Bound-preserving high order schemes in computational fluid dynamics Chi-Wang Shu Bound-preserving high order schemes in computational fluid dynamics Chi-Wang Shu Division of Applied Mathematics Brown University Outline Introduction Maximum-principle-preserving for scalar conservation

More information

Numerical Methods for Hyperbolic Conservation Laws Lecture 4

Numerical Methods for Hyperbolic Conservation Laws Lecture 4 Numerical Methods for Hyperbolic Conservation Laws Lecture 4 Wen Shen Department of Mathematics, Penn State University Email: wxs7@psu.edu Oxford, Spring, 018 Lecture Notes online: http://personal.psu.edu/wxs7/notesnumcons/

More information

Finite volume method on unstructured grids

Finite volume method on unstructured grids Finite volume method on unstructured grids Praveen. C praveen@math.tifrbng.res.in Tata Institute of Fundamental Research Center for Applicable Mathematics Bangalore 560065 http://math.tifrbng.res.in/~praveen

More information

H. L. Atkins* NASA Langley Research Center. Hampton, VA either limiters or added dissipation when applied to

H. L. Atkins* NASA Langley Research Center. Hampton, VA either limiters or added dissipation when applied to Local Analysis of Shock Capturing Using Discontinuous Galerkin Methodology H. L. Atkins* NASA Langley Research Center Hampton, A 68- Abstract The compact form of the discontinuous Galerkin method allows

More information

AProofoftheStabilityoftheSpectral Difference Method For All Orders of Accuracy

AProofoftheStabilityoftheSpectral Difference Method For All Orders of Accuracy AProofoftheStabilityoftheSpectral Difference Method For All Orders of Accuracy Antony Jameson 1 1 Thomas V. Jones Professor of Engineering Department of Aeronautics and Astronautics Stanford University

More information

Entropy-stable discontinuous Galerkin nite element method with streamline diusion and shock-capturing

Entropy-stable discontinuous Galerkin nite element method with streamline diusion and shock-capturing Entropy-stable discontinuous Galerkin nite element method with streamline diusion and shock-capturing Siddhartha Mishra Seminar for Applied Mathematics ETH Zurich Goal Find a numerical scheme for conservation

More information

A Space-Time Expansion Discontinuous Galerkin Scheme with Local Time-Stepping for the Ideal and Viscous MHD Equations

A Space-Time Expansion Discontinuous Galerkin Scheme with Local Time-Stepping for the Ideal and Viscous MHD Equations A Space-Time Expansion Discontinuous Galerkin Scheme with Local Time-Stepping for the Ideal and Viscous MHD Equations Ch. Altmann, G. Gassner, F. Lörcher, C.-D. Munz Numerical Flow Models for Controlled

More information

A Bound-Preserving Fourth Order Compact Finite Difference Scheme for Scalar Convection Diffusion Equations

A Bound-Preserving Fourth Order Compact Finite Difference Scheme for Scalar Convection Diffusion Equations A Bound-Preserving Fourth Order Compact Finite Difference Scheme for Scalar Convection Diffusion Equations Hao Li Math Dept, Purdue Univeristy Ocean University of China, December, 2017 Joint work with

More information

CapSel Roe Roe solver.

CapSel Roe Roe solver. CapSel Roe - 01 Roe solver keppens@rijnh.nl modern high resolution, shock-capturing schemes for Euler capitalize on known solution of the Riemann problem originally developed by Godunov always use conservative

More information

A Basis for Discretizations

A Basis for Discretizations A Basis for Discretizations Robert Struijs 6 January Abstract. We discuss approximations of a given order of accuracy to first and higher derivatives on a structured grid in N dimensions using a basis

More information

A Central Compact-Reconstruction WENO Method for Hyperbolic Conservation Laws

A Central Compact-Reconstruction WENO Method for Hyperbolic Conservation Laws A Central Compact-Reconstruction WENO Method for Hyperbolic Conservation Laws Kilian Cooley 1 Prof. James Baeder 2 1 Department of Mathematics, University of Maryland - College Park 2 Department of Aerospace

More information

Comparison of cell-centered and node-centered formulations of a high-resolution well-balanced finite volume scheme: application to shallow water flows

Comparison of cell-centered and node-centered formulations of a high-resolution well-balanced finite volume scheme: application to shallow water flows Comparison of cell-centered and node-centered formulations of a high-resolution well-balanced finite volume scheme: application to shallow water flows Dr Argiris I. Delis Dr Ioannis K. Nikolos (TUC) Maria

More information

What is a flux? The Things We Does Know

What is a flux? The Things We Does Know What is a flux? Finite Volume methods (and others) (are based on ensuring conservation by computing the flux through the surfaces of a polyhedral box. Either the normal component of the flux is evaluated

More information

Anti-diffusive finite difference WENO methods for shallow water with. transport of pollutant

Anti-diffusive finite difference WENO methods for shallow water with. transport of pollutant Anti-diffusive finite difference WENO methods for shallow water with transport of pollutant Zhengfu Xu 1 and Chi-Wang Shu 2 Dedicated to Professor Qun Lin on the occasion of his 70th birthday Abstract

More information

ARTICLE IN PRESS Mathematical and Computer Modelling ( )

ARTICLE IN PRESS Mathematical and Computer Modelling ( ) Mathematical and Computer Modelling Contents lists available at ScienceDirect Mathematical and Computer Modelling ournal homepage: wwwelseviercom/locate/mcm Total variation diminishing nonstandard finite

More information

On Numerical Simulation of Flow Problems by Discontinuous Galerkin and Finite Volume Techniques

On Numerical Simulation of Flow Problems by Discontinuous Galerkin and Finite Volume Techniques On Numerical Simulation of Flow Problems by Discontinuous Galerkin and Finite Volume Techniques Ondřej Winter 1, Petr Sváček 1 1 CTU in Prague, Faculty of Mechanical Engineering, Department of Technical

More information

Advanced numerical methods for nonlinear advectiondiffusion-reaction. Peter Frolkovič, University of Heidelberg

Advanced numerical methods for nonlinear advectiondiffusion-reaction. Peter Frolkovič, University of Heidelberg Advanced numerical methods for nonlinear advectiondiffusion-reaction equations Peter Frolkovič, University of Heidelberg Content Motivation and background R 3 T Numerical modelling advection advection

More information

A New Class of High-Order Energy Stable Flux Reconstruction Schemes

A New Class of High-Order Energy Stable Flux Reconstruction Schemes DOI 10.1007/s10915-010-940-z A New Class of High-Order Energy Stable Flux Reconstruction Schemes P.E. Vincent P. Castonguay A. Jameson Received: May 010 / Revised: September 010 / Accepted: 5 September

More information

On the relationship of local projection stabilization to other stabilized methods for one-dimensional advection-diffusion equations

On the relationship of local projection stabilization to other stabilized methods for one-dimensional advection-diffusion equations On the relationship of local projection stabilization to other stabilized methods for one-dimensional advection-diffusion equations Lutz Tobiska Institut für Analysis und Numerik Otto-von-Guericke-Universität

More information

VISCOUS FLUX LIMITERS

VISCOUS FLUX LIMITERS VISCOUS FLUX LIMITERS E. F. Toro Department of Aerospace Science College of Aeronautics Cranfield Institute of Technology Cranfield, Beds MK43 OAL England. Abstract We present Numerical Viscosity Functions,

More information

AA214B: NUMERICAL METHODS FOR COMPRESSIBLE FLOWS

AA214B: NUMERICAL METHODS FOR COMPRESSIBLE FLOWS AA214B: NUMERICAL METHODS FOR COMPRESSIBLE FLOWS 1 / 10 AA214B: NUMERICAL METHODS FOR COMPRESSIBLE FLOWS Pseudo-Time Integration 1 / 10 AA214B: NUMERICAL METHODS FOR COMPRESSIBLE FLOWS 2 / 10 Outline 1

More information

Improved Seventh-Order WENO Scheme

Improved Seventh-Order WENO Scheme 48th AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Eposition 4-7 January 2, Orlando, Florida AIAA 2-45 Improved Seventh-Order WENO Scheme Yiqing Shen Gecheng Zha Dept.

More information

Numerical Methods for Conservation Laws WPI, January 2006 C. Ringhofer C2 b 2

Numerical Methods for Conservation Laws WPI, January 2006 C. Ringhofer C2 b 2 Numerical Methods for Conservation Laws WPI, January 2006 C. Ringhofer ringhofer@asu.edu, C2 b 2 2 h2 x u http://math.la.asu.edu/ chris Last update: Jan 24, 2006 1 LITERATURE 1. Numerical Methods for Conservation

More information

An Improved Non-linear Weights for Seventh-Order WENO Scheme

An Improved Non-linear Weights for Seventh-Order WENO Scheme An Improved Non-linear Weights for Seventh-Order WENO Scheme arxiv:6.06755v [math.na] Nov 06 Samala Rathan, G Naga Raju Department of Mathematics, Visvesvaraya National Institute of Technology, Nagpur,

More information

Sub-Cell Shock Capturing for Discontinuous Galerkin Methods

Sub-Cell Shock Capturing for Discontinuous Galerkin Methods Sub-Cell Shock Capturing for Discontinuous Galerkin Methods Per-Olof Persson and Jaime Peraire Massachusetts Institute of Technology, Cambridge, MA 39, U.S.A. A shock capturing strategy for higher order

More information

A Fifth Order Flux Implicit WENO Method

A Fifth Order Flux Implicit WENO Method A Fifth Order Flux Implicit WENO Method Sigal Gottlieb and Julia S. Mullen and Steven J. Ruuth April 3, 25 Keywords: implicit, weighted essentially non-oscillatory, time-discretizations. Abstract The weighted

More information

Lecture - Implementation and numerical aspects of DG-FEM in 1D

Lecture - Implementation and numerical aspects of DG-FEM in 1D Lecture - Implementation and numerical aspects of DG-FEM in 1D Ph.D. Course: An Introduction to DG-FEM for solving partial differential equations Allan P. Engsig-Karup Scientific Computing Section DTU

More information

Finite Element methods for hyperbolic systems

Finite Element methods for hyperbolic systems Finite Element methods for hyperbolic systems Eric Sonnendrücker Max-Planck-Institut für Plasmaphysik und Zentrum Mathematik, TU München Lecture notes Wintersemester 14-15 January 19, 15 Contents 1 Introduction

More information

Finite difference Hermite WENO schemes for the Hamilton-Jacobi equations 1

Finite difference Hermite WENO schemes for the Hamilton-Jacobi equations 1 Finite difference Hermite WENO schemes for the Hamilton-Jacobi equations Feng Zheng, Chi-Wang Shu 3 and Jianian Qiu 4 Abstract In this paper, a new type of finite difference Hermite weighted essentially

More information

Introduction to Finite Volume projection methods. On Interfaces with non-zero mass flux

Introduction to Finite Volume projection methods. On Interfaces with non-zero mass flux Introduction to Finite Volume projection methods On Interfaces with non-zero mass flux Rupert Klein Mathematik & Informatik, Freie Universität Berlin Summerschool SPP 1506 Darmstadt, July 09, 2010 Introduction

More information

Scientific Computing WS 2017/2018. Lecture 18. Jürgen Fuhrmann Lecture 18 Slide 1

Scientific Computing WS 2017/2018. Lecture 18. Jürgen Fuhrmann Lecture 18 Slide 1 Scientific Computing WS 2017/2018 Lecture 18 Jürgen Fuhrmann juergen.fuhrmann@wias-berlin.de Lecture 18 Slide 1 Lecture 18 Slide 2 Weak formulation of homogeneous Dirichlet problem Search u H0 1 (Ω) (here,

More information

High order finite difference Hermite WENO schemes for the Hamilton-Jacobi equations on unstructured meshes

High order finite difference Hermite WENO schemes for the Hamilton-Jacobi equations on unstructured meshes High order finite difference Hermite WENO schemes for the Hamilton-Jacobi equations on unstructured meshes Feng Zheng, Chi-Wang Shu and Jianxian Qiu 3 Abstract In this paper, a new type of high order Hermite

More information

FDM for wave equations

FDM for wave equations FDM for wave equations Consider the second order wave equation Some properties Existence & Uniqueness Wave speed finite!!! Dependence region Analytical solution in 1D Finite difference discretization Finite

More information

The one-dimensional equations for the fluid dynamics of a gas can be written in conservation form as follows:

The one-dimensional equations for the fluid dynamics of a gas can be written in conservation form as follows: Topic 7 Fluid Dynamics Lecture The Riemann Problem and Shock Tube Problem A simple one dimensional model of a gas was introduced by G.A. Sod, J. Computational Physics 7, 1 (1978), to test various algorithms

More information

On finite element methods for 3D time dependent convection diffusion reaction equations with small diffusion

On finite element methods for 3D time dependent convection diffusion reaction equations with small diffusion On finite element methods for 3D time dependent convection diffusion reaction equations with small diffusion Volker John and Ellen Schmeyer FR 6.1 Mathematik, Universität des Saarlandes, Postfach 15 11

More information

Adaptive TVD-RK Discontinuous Galerkin Algorithms for Shallow Water Equations

Adaptive TVD-RK Discontinuous Galerkin Algorithms for Shallow Water Equations Adaptive TVD-RK Discontinuous Galerkin Algorithms for Shallow Water Equations Thida Pongsanguansin, Khamron Mekchay and Montri Maleewong Abstract The adaptive Discontinuous Galerkin DG method for solving

More information

Numerical Oscillations and how to avoid them

Numerical Oscillations and how to avoid them Numerical Oscillations and how to avoid them Willem Hundsdorfer Talk for CWI Scientific Meeting, based on work with Anna Mozartova (CWI, RBS) & Marc Spijker (Leiden Univ.) For details: see thesis of A.

More information

Discontinuous Galerkin method for hyperbolic equations with singularities

Discontinuous Galerkin method for hyperbolic equations with singularities Discontinuous Galerkin method for hyperbolic equations with singularities Chi-Wang Shu Division of Applied Mathematics Brown University Joint work with Qiang Zhang, Yang Yang and Dongming Wei Outline Introduction

More information

A high order adaptive finite element method for solving nonlinear hyperbolic conservation laws

A high order adaptive finite element method for solving nonlinear hyperbolic conservation laws A high order adaptive finite element method for solving nonlinear hyperbolic conservation laws Zhengfu Xu, Jinchao Xu and Chi-Wang Shu 0th April 010 Abstract In this note, we apply the h-adaptive streamline

More information

Math 7824 Spring 2010 Numerical solution of partial differential equations Classroom notes and homework

Math 7824 Spring 2010 Numerical solution of partial differential equations Classroom notes and homework Math 7824 Spring 2010 Numerical solution of partial differential equations Classroom notes and homework Jan Mandel University of Colorado Denver May 12, 2010 1/20/09: Sec. 1.1, 1.2. Hw 1 due 1/27: problems

More information

YINGJIE LIU, CHI-WANG SHU, EITAN TADMOR, AND MENGPING ZHANG

YINGJIE LIU, CHI-WANG SHU, EITAN TADMOR, AND MENGPING ZHANG CENTRAL DISCONTINUOUS GALERKIN METHODS ON OVERLAPPING CELLS WITH A NON-OSCILLATORY HIERARCHICAL RECONSTRUCTION YINGJIE LIU, CHI-WANG SHU, EITAN TADMOR, AND MENGPING ZHANG Abstract. The central scheme of

More information

Extremum-Preserving Limiters for MUSCL and PPM

Extremum-Preserving Limiters for MUSCL and PPM arxiv:0903.400v [physics.comp-ph] 7 Mar 009 Extremum-Preserving Limiters for MUSCL and PPM Michael Sekora Program in Applied and Computational Mathematics, Princeton University Princeton, NJ 08540, USA

More information

Antony Jameson. AIAA 18 th Computational Fluid Dynamics Conference

Antony Jameson. AIAA 18 th Computational Fluid Dynamics Conference Energy Estimates for Nonlinear Conservation Laws with Applications to Solutions of the Burgers Equation and One-Dimensional Viscous Flow in a Shock Tube by Central Difference Schemes Antony Jameson AIAA

More information

ANALYSIS AND IMPLEMENTATION OF THE GAS-KINETIC BGK SCHEME FOR COMPUTATIONAL GAS DYNAMICS

ANALYSIS AND IMPLEMENTATION OF THE GAS-KINETIC BGK SCHEME FOR COMPUTATIONAL GAS DYNAMICS INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN FLUIDS, VOL. 25, 21±49 (1997) ANALYSIS AND IMPLEMENTATION OF THE GAS-KINETIC BGK SCHEME FOR COMPUTATIONAL GAS DYNAMICS CHONGAM KIM*, KUN XU, LUIGI MARTINELLI

More information