Page 1
Entropy stable high-order schemes for systems ofconservation laws.
Siddhartha Mishra
Center of Mathematics for Applications (CMA),University of Oslo, Norway, (and)
Seminar for Applied Mathematics (SAM),ETH Zurich, Switzerland.
Siddhartha Mishra High order entropy stable schemes
Page 2
Joint work with
I PART 1: Entropy stable FV schemes:I Ulrik S. Fjordholm (SAM, ETH Zurich).I Eitan Tadmor (CSCAMM, U. Maryland, USA.)I Aziz Madrane (Bombardier Aerospace, Montreal, Canada.)
I PART 2: Entropy stable STDG schemes:I Andreas Hiltebrand (SAM, ETH Zurich).
I PART 3: Small scale dependent shock waves.I Phillipe LeFloch (U. Paris VI, France).I M. Castro, C. Pares (U. Malaga, Spain).I L. V. Spinolo (U. Zurich, Switzerland).
Siddhartha Mishra High order entropy stable schemes
Page 3
Systems of conservation laws
I Systems of conservation laws:
Ut + div(F(U)) = 0.
I Examples:I Shallow water equations (Geophysics)I Euler equations (Aerodynamics)I MHD equations (Plasma physics)
Siddhartha Mishra High order entropy stable schemes
Page 4
Mathematical Framework
I Shock waves ⇒ Weak (distributional) Solutions
I Uniqueness is an issue: (Entropy conditions).
I Standard paradigm for numerical schemes.
Siddhartha Mishra High order entropy stable schemes
Page 5
Ingredient I: Approximate Riemann solvers
I Finite volume scheme: ddt Uj(t) + 1
∆x (Fj+1/2 − Fj−1/2) = 0.
X j 1 Xj +1
t n
tn+1
Unj U
nj+1U
nj −1
Un+1j
Fj +1/2Fj −1/2
− 2/ /2
I Numerical flux Fj+1/2 : (approximate) Riemann solver.I Godunov (Roe).I HLL type.
Siddhartha Mishra High order entropy stable schemes
Page 6
Ingredient II: Non-oscillatory reconstructions
I Piecewise polynomial reconstructions ⇒ Higher order ofaccuracy.
XJ−1/2
XJ+ 1/2
VJ
+
VJ+ 1
−
VJ
JV
+ 1
I Non-Oscillatory reconstructions (control in BV ):I TVD limiters (Van Leer).I ENO reconstruction (Harten et. al.).I WENO reconstruction (Shu, Osher).I DG method (Cockburn,Shu).
Siddhartha Mishra High order entropy stable schemes
Page 7
Ingredient III: Time stepping
I Strong stability preserving (SSP) Runge-Kutta methods.I Gottlieb, Shu, Tadmor.
I Control in BV .
Siddhartha Mishra High order entropy stable schemes
Page 8
Standard Paradigm
I Highly succesful in practice.
I Ex: Waves in the sun (Fuchs,McMurry,SM,Waagan):
I Problems still remain !!!
Siddhartha Mishra High order entropy stable schemes
Page 9
Problem I: Lack of rigorous stability/convergence results
I Stability: continuous framework.I (Multi-dimensional) Scalar conservation laws:
I Existence: BV estimates.I Maximum principles.I Uniqueness: (Infinitely many) Entropy inequalities.
Siddhartha Mishra High order entropy stable schemes
Page 10
Systems of conservation laws: Entropy framework
I Consider 1-D system: Ut + Fx = 0.
I Assume there exist S (Convex), V and Q with
V∂U = ∂S , V∂F = ∂Q.
I Entropy identity for smooth solutions:
St + Qx = 0.
I Entropy dissipation at shocks ⇒
St + Qx” ≤ ”0.
I provides stability estimate:
d
dt
∫S(U)dx ≤ 0⇒ ‖U(., t)‖L2 ≤ C .
I Holds for several space dimensions.
Siddhartha Mishra High order entropy stable schemes
Page 11
Robust numerical scheme for conservation laws
• shouldI Entropy stable for non-linear systems.
I Discrete entropy inequalities.
I Convergent forI Linear symmetrizable systems.I Scalar conservation laws.
Siddhartha Mishra High order entropy stable schemes
Page 12
Existing globally stable schemes
I Scalar equationsI Monotone schemes (1st-order).• Harten, TVD bounds , Crandall, Majda, Entropy estimate.
I TVD limiter based schemes (2nd-order)• Sweby, VanLeer, BV bounds, Osher, Tadmor, Entropybounds.
I ENO schemes (arbitrary order)• No stability bounds, E.Tadmor’s talk
I WENO schemes (arbitrary order)• No (global) stability bounds.
I DG schemes (arbitrary order)• Cockburn, Shu, 1990., BV stability.
I Systems of equations,I Finite volume schemes, Tadmor, 1987, 2003,• Entropy stable Ist order FV scheme
Siddhartha Mishra High order entropy stable schemes
Page 13
Problem II: Observed numerical instabilites
I Strong shocks:
−1 −0.5 0 0.5 10
5
10
15
ReferenceECERoeRoe
I Vortex Dominated flows:
I Hypersonic flows:
Siddhartha Mishra High order entropy stable schemes
Page 14
Problem III: Small scale dependent shock waves
I Limit solutions of the hyperbolic-parabolic system:
Uεt + div (F(Uε)) = εdiv (B(Uε)∇Uε)
I Depend explicitly on B: limε→0
Uε,B = UB
I Failure of standard schemes: Boundary value problems
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
1
1.5
2
2.5
3
EDDY
ROE(100 pts)
ROE(1000 pts)
Siddhartha Mishra High order entropy stable schemes
Page 15
Problem III: Small scale dependent shock waves
I Non-conservative hyperbolic system: Ut + AUx = 0.
I Ex:Multi-layer shallow-water, Coupled Burgers:
−2 0 2 4 6 8 100
2
4
6
8
−2 0 2 4 6 8 100
5
10
15
Godunov
Exact
I Subtle interaction of entropy with viscosity mechanisms.
I Standard paradigm 7→ New paradigm.
I Basis of new paradigm: Entropy stability.
Siddhartha Mishra High order entropy stable schemes
Page 16
Ingredient I: Entropy conservative fluxes
I Consider the one-d conservation law: Ut + F(U)x = 0.
I Let S be entropy function, flux, define,
V = ∂US , Ψ = 〈V ,F〉 − Q,
I Then
〈V,U〉t + 〈V,Fx〉 = 0,
St + (〈V,F〉)x − 〈Vx ,F〉 = 0,
I Using Ψx = 〈Vx ,F〉, we get that
St + Qx = 0.
I Entropy identity for smooth solutions.
Siddhartha Mishra High order entropy stable schemes
Page 17
Finite difference scheme
I Conservative semi-discrete scheme:
d
dtUj(t) +
1
∆x(Fj+1/2 − Fj−1/2) = 0
I On the grid:
X j 1 Xj +1
t n
tn+1
Unj U
nj+1U
nj −1
Un+1j
Fj +1/2Fj −1/2
− 2/ /2
Siddhartha Mishra High order entropy stable schemes
Page 18
Entropy conservative flux: Tadmor,1987
I Assuming〈[[Vj+1/2]],F∗j+1/2〉 = [[Ψj+1/2]].
I We can mimic calculations and arrive at,
d
dtS(Uj(t)) = − 1
∆x(Qj+1/2 − Qj−1/2),
I Existence (Tadmor,1987):
F∗j+1/2 =
∫ 1/2
−1/2F(Vj+1/2(ξ)
)dξ.
I Recent explicit solutions increase computational efficiency:I Shallow water: Fjordholm, Mishra, Tadmor, 2009.I Euler: Roe, 2007.
Siddhartha Mishra High order entropy stable schemes
Page 19
High-order entropy conservative fluxes
I 2p-th order accurate finite difference generalization (LeFloch,Mercier, Rohde, 2001):
Fp,∗j+1/2 :=
p∑i=1
αpi F∗(Uj ,Uj+i ),
I Fourth-order entropy conservative flux:
F4,∗j+1/2 =
4
3F∗(Uj ,Uj+1)− 1
6(F∗(Uj−1,Uj+1) + F∗(Uj ,Uj+2))
FOURTH ORDER
SECOND ORDER
Siddhartha Mishra High order entropy stable schemes
Page 20
Comparison
Standard paradigm New paradigm
Approximate Riemann solvers Entropy conservative flux
BV reconstructions
SSP-RK time stepping
Siddhartha Mishra High order entropy stable schemes
Page 21
Computed heights, 2nd Order scheme
−1 −0.5 0 0.5 1
1.4
1.6
1.8
2
2.2
(a) EEC, 400 pts
−1 −0.5 0 0.5 1
1.4
1.6
1.8
2
2.2
(b) EEC,1600 pts
Siddhartha Mishra High order entropy stable schemes
Page 22
Oscillations
I Entropy preserving schemes at shocks.
I Oscillations at mesh scale.
I Entropy should be dissipated at shocks.
I Absence of dissipation mechanism ⇒ energy re-distribution tosmallest resolvable scales.
I Have to add Numerical diffusion.
Siddhartha Mishra High order entropy stable schemes
Page 23
Entropy stable fluxes (Fjordholm, SM, Tadmor, 2009)
I Add dissipation in terms of entropy variables.
I ERoe:
Fj+1/2 = F∗j+1/2 −1
2Rj+1/2|Λj+1/2|R>j+1/2[[Vj+1/2]].
I ERus:
Fj+1/2 = F∗j+1/2−1
2max{|λmax
j |, |λmaxj+1 |}Rj+1/2R
>j+1/2[[Vj+1/2]].
I Resulting semi-discrete schemes are entropy stable.
I Schemes are only first-order accurate.
Siddhartha Mishra High order entropy stable schemes
Page 24
Comparison
Standard paradigm New paradigm
Approximate Riemann solvers Entropy conservative flux+Num diff in entropy var
BV reconstructions
SSP-RK time stepping
Siddhartha Mishra High order entropy stable schemes
Page 25
Shallow water Dam break
−1 −0.5 0 0.5 11.4
1.6
1.8
2
ReferenceECRoeRoeRusanov
Siddhartha Mishra High order entropy stable schemes
Page 26
Normalized run times
Relative error 1 0.5 0.1
Rusanov 1.05 8.24 203.41Roe 1.15 8.43 208.29ERoe 1 7.36 171.7
Siddhartha Mishra High order entropy stable schemes
Page 27
A different dam-break problem: Strong shock
−1 −0.5 0 0.5 10
5
10
15
ReferenceECERoeRoe
Siddhartha Mishra High order entropy stable schemes
Page 28
Very-high order numerical diffusion operator
I For formal order ∆xk :
I Replace 〈Dj+1/2,Vj+1 − Vj〉 7→ 〈Dj+1/2,V−j+1 − V+
j 〉I Based on piecewise polynomial reconstruction of order (k − 1)
in each cell.
XJ−1/2
XJ+ 1/2
VJ
+
VJ+ 1
−
VJ
JV
+ 1
I Entropy stability needs to be ensured.
Siddhartha Mishra High order entropy stable schemes
Page 29
Ingredient III: Entropy stable reconstruction
I Formulated by Fjordholm, SM, Tadmor, 2011.
I Assumes existence of diagonal scaling matrix B ≥ 0 such that
〈V〉j+1/2 := (R>)−1j+1/2Bj+1/2R
>j+1/2[[V]]j+1/2.
⇒ Entropy stability.I Can be ensured if reconstruction
I is in scaled entropy variables: W = R>V.I preserves SIGN PROPERTY (componentwise)
sign(〈w〉j+1/2) = sign([[w ]]j+1/2).
Siddhartha Mishra High order entropy stable schemes
Page 30
Piecewise linear reconstructions
I Second-order of accuracy.
I Sign property for standard TVD limiter:
Limiter Sign property
Minmod YesSuperbee NoMC NoVan-Leer No
I Higher than second-order accuracy ??
Siddhartha Mishra High order entropy stable schemes
Page 31
ENO reconstruction: E. Tadmor’s lecture
I Theorem (Fjordholm, SM, Tadmor, 2011):• ENO reconstruction preserves the sign property at any order.
−0.4 −0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4−0.2
0
0.2
0.4
0.6
0.8
1
1.2
k = 4
Cell averages
Reconstruction
Siddhartha Mishra High order entropy stable schemes
Page 32
Arbitrary order entropy stable schemes
I TeCNO scheme:
d
dtUj(t) +
1
∆x(Fj+1/2 − Fj−1/2) = 0
I Numerical flux:
Fj+1/2 = Fp,∗j+1/2 −
1
2Dj+1/2〈V〉j+1/2.
I Entropy conservative flux:
Fp,∗j+1/2 :=
p∑i=1
αpi F∗(Uj ,Uj+i ),
I Explicit two-point entropy conservative flux F∗
I ENO reconstruction in scaled entropy variables.
Siddhartha Mishra High order entropy stable schemes
Page 33
Arbitrary order entropy stable schemes
I Theorem: Consider system Ut + Fx = 0 with entropy functionS . Then the arbitrary order TeCNO scheme satisfies a discreteentropy inequality:
d
dtSj +
1
∆x
(Qj+1/2 − Qj+1/2
)≤ 0.
and is entropy stable (independent of the order of thescheme).
I Theorem: If the system is linear symmetrizable, then thearbitrary order TeCNO schemes converges weakly in L2 to theunique solution.
I Straightforward to extend on multi-D on Logically rectangulargrids.
Siddhartha Mishra High order entropy stable schemes
Page 34
Comparison
Standard paradigm New paradigm
Approximate Riemann solvers Entropy conservative flux+Num diff in entropy var
BV reconstructions Sign property preserving recon
SSP-RK time stepping
Siddhartha Mishra High order entropy stable schemes
Page 35
Wave equation: convergence for Sine wave
102
103
10−8
10−7
10−6
10−5
10−4
10−3
10−2
Number of grid points
L1 e
rro
r in
h
Errors for wave equation with u0(x)=sin(4π x). Errors at t=1.
RusENO3
ERusENO3
RusENO4
ERusENO4
RusENO5
ERusENO5
Siddhartha Mishra High order entropy stable schemes
Page 36
Euler: Shock-turbulence interaction
−5 0 50.5
1
1.5
2
2.5
3
3.5
4
4.5
5
x
rho
(c) ENO3
−5 0 50.5
1
1.5
2
2.5
3
3.5
4
4.5
5
x
rho
(d) TeCNO3
Siddhartha Mishra High order entropy stable schemes
Page 37
Euler: Shock-turbulence interaction
−5 0 50.5
1
1.5
2
2.5
3
3.5
4
4.5
5
x
rho
(e) ENO4
−5 0 50.5
1
1.5
2
2.5
3
3.5
4
4.5
5
x
rho
(f) TeCNO4
Siddhartha Mishra High order entropy stable schemes
Page 38
Advection of Euler vortex: TeCNO2
2 4 6 8 100
1
2
3
4
5
6
7
8
9
10
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Siddhartha Mishra High order entropy stable schemes
Page 39
Advection of Euler vortex:TeCNO3
2 4 6 8 100
1
2
3
4
5
6
7
8
9
10
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Siddhartha Mishra High order entropy stable schemes
Page 40
Advection of Euler vortex:TeCNO4
2 4 6 8 100
1
2
3
4
5
6
7
8
9
10
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Siddhartha Mishra High order entropy stable schemes
Page 41
Advection of Euler vortex
0 2 4 6 8 10
0.5
0.6
0.7
0.8
0.9
1
(g) TeCNO2
0 2 4 6 8 10
0.5
0.6
0.7
0.8
0.9
1
(h) TeCNO3
0 2 4 6 8 10
0.5
0.6
0.7
0.8
0.9
1
(i) TeCNO4
Siddhartha Mishra High order entropy stable schemes
Page 42
Euler: Cloud-Shock interaction: TeCNO2
Siddhartha Mishra High order entropy stable schemes
Page 43
Euler: Cloud-Shock interaction: TeCNO3
Siddhartha Mishra High order entropy stable schemes
Page 44
Euler: Cloud-Shock interaction: TeCNO4
Siddhartha Mishra High order entropy stable schemes
Page 45
Extension to Unstructured meshes: Fjordholm, Madrane,SM, Tadmor.
I Entropy conservative flux: Fij = F1ijn
1ij + F2
ijn2ij
I Components satisfy:
[[V]]>ij Fkij = [[ψk ]]ij k = 1, 2,
I Entropy stable flux:
Fij = Fij −1
2Dij [[V]]ij .
I Example of Numerical diffusion operator:
Dij = R(nij )|Λ(nij )|R>(nij )
.
I First-order accurate.
Siddhartha Mishra High order entropy stable schemes
Page 46
Unstructured grids
U
Ui
j
Siddhartha Mishra High order entropy stable schemes
Page 47
Second-order numerical diffusion operator: Fjordholm, SM
I Reconstruction has to satisfy Sign property at edge midpoints.I Modification of the Barth-Jesperson limiter.
I Numerical experiments ongoing.
Siddhartha Mishra High order entropy stable schemes
Page 48
Outstanding issues
I Very-high order discretizations on unstructured meshes.I Sign property at quadrature points.I Vertex centered ENO formulation (Abgrall).
I Fully discrete high-order schemes:I Numerical experiments indicate SSP-RK3 is entropy stable.
Siddhartha Mishra High order entropy stable schemes
Page 49
Possible solution to both problems
I Space-time Discontinuous Galerkin methods.I Hiltebrand, SM, in progress.
I Combines ingredients due toI Johnson, Szepessy, Hansbo.I Hughes, Franca, Mallet.I Barth.
Siddhartha Mishra High order entropy stable schemes
Page 50
Space time element
n
v−v+
vn−
vn+1−
Siddhartha Mishra High order entropy stable schemes
Page 51
Shock capturing space time DG for ut + (f i(u))xi= 0
I DG formulation: BDG (v,w) + BSD(v,w) + BSC (v,w) := 0.
I DG quasiliner form:
BDG (vh,w) := −∑K ,n
∫K
∫ tn+1
tn
u(vh)wt + f i (vh)wxi dxdt
+∑n,K
∫K
(u(vh(tn+1
− ))w(tn+1− )− u(vh(tn
−))w(tn+))
dx
+∑n,K
∫∂K
∫ tn+1
tn
F∗(vh(x−), vh(x+), ν)w(x−)dσ(x)dt
−∑n,K
∫∂K
∫ tn+1
tn
1
2D(vh(x+)− vh(x−))w(x−)dσ(x)dt.
I DOFs are entropy variables.
Siddhartha Mishra High order entropy stable schemes
Page 52
Too little diffusion ⇒ Oscillations
−1 −0.5 0 0.5 1−0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Siddhartha Mishra High order entropy stable schemes
Page 53
Shock capturing space time DG for ut + (f i(u))xi= 0
I Streamline diffusion quasilinear form:
BSD(vh,w) :=
h∑K ,n
∫K
∫ tn+1
tn
uv(vh)wt + f iv(vh)wxi D(u(vh)t + f i (vh)xi
)dxdt
I Shock capturing operator:
BSC (vh,w) :=
h∑K ,n
∫K
∫ tn+1
tn
‖u(vh)t + f i (vh)xi‖‖∇v‖+ ε
uv(vh)∇vh · ∇wdxdt.
Siddhartha Mishra High order entropy stable schemes
Page 54
Properties: Hiltebrand, SM.
I Entropy stability of BI Arbitrary order accuracy.
I Maximum principle for scalar conservation laws.
I Convergence to Linear symmetrizable systems.
I Preliminary numerical results.
Siddhartha Mishra High order entropy stable schemes
Page 55
Convergence rates for smooth solutions: Wave
101
102
103
10−6
10−4
10−2
100
Nx
||u−
ue
xa
ct||
1/||u
exa
ct||
1
deg=0
deg=1
deg=2
deg=3
deg=4
Siddhartha Mishra High order entropy stable schemes
Page 56
Comparison of different orders:Burgers
−1 −0.5 0 0.5 1−0.2
0
0.2
0.4
0.6
0.8
1
1.2
x
u
deg=0
deg=1
deg=2
deg=3
exact
Siddhartha Mishra High order entropy stable schemes
Page 57
Comparison of different schemes:Burgers
−1 −0.5 0 0.5 1−0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
no SD/SC
SD
SD+SC
exact
Siddhartha Mishra High order entropy stable schemes
Page 58
Comparison of different schemes: Euler Sod shock tube
−5 0 50.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
x
ρ
no SD/SCSDSD+SC(p)exact
Siddhartha Mishra High order entropy stable schemes
Page 59
Comparison of different orders: Euler shock tube
−5 0 50.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
x
ρ
deg=0deg=1deg=2deg=3exact
Siddhartha Mishra High order entropy stable schemes
Page 60
Comparison of different orders: Euler Shu-Osher
−5 0 50.5
1
1.5
2
2.5
3
3.5
4
4.5
5
x
ρ
deg=0deg=1deg=2deg=3reference
Siddhartha Mishra High order entropy stable schemes
Page 61
Space time DG: Issues
I Choice of parameters.
I Positivity for systems (a la Zhang, Shu).
I Solutions of the resulting non-linear systems.
I Efficient Preconditioners.
Siddhartha Mishra High order entropy stable schemes
Page 62
Comparison
Standard paradigm New paradigm
Approximate Riemann solvers Entropy conservative flux+Num diff in entropy var
BV reconstructions Sign property preserving recon (SG)Shock-capturing STDG (UG)
SSP-RK time stepping SSP-RK time stepping ??(Implicit) space-time DG.
Siddhartha Mishra High order entropy stable schemes
Page 63
Comparison
High-order (RK) FV Space-time DG
No tuning Some tuningFast (Relatively) SlowStructured grids Unstrutured GridsOne time scale Multiple time scales(Difficult) space time adaptivity space-time adaptivity.Single processor Massively parallel platforms
Siddhartha Mishra High order entropy stable schemes
Page 64
Small scale dependent shock waves
I Limit solutions of the hyperbolic-parabolic system:
Uεt + div (F(Uε)) = εdiv (B(Uε)∇Uε)
I Depend explicitly on B: limε→0
Uε,B = UB
I Failure of standard schemes: boundary layers
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
1
1.5
2
2.5
3
EDDY
ROE(100 pts)
ROE(1000 pts)
Siddhartha Mishra High order entropy stable schemes
Page 65
Role of equivalent equation: LeFloch, SM, 2009
I Equivalent equation for scheme:
d
dtUi (t) +
1
h
(Fi+1/2(t)− Fi−1/2(t)
)= 0
I
Ut + Fx = h(B(U)Ux
)x
+R, R =∞∑
q=2p+1
C 2pq (h)q−1
q![Fq(U)]q,
I For standard schemes: B 6= B !!!I New approach: F = F∗ − 1
2D[[V]]i+1/2,I Entropy conservative fluxI Numerical diffusion operator: D = BUV.I Entropy stable scheme with correct equivalent equation (at
leading order).
Siddhartha Mishra High order entropy stable schemes
Page 66
Boundary value problems: SM, Spinolo, 2011
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
1
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
2.8
3
Reference
Roe
CND
Siddhartha Mishra High order entropy stable schemes
Page 67
Non-conservative hyperbolic systems: Castro, Fjordholm,SM, Pares, 2011
−2 0 2 4 6 8 100
5
10u
−2 0 2 4 6 8 100
5
10
15
v
ESPC
Exact
Godunov
ESPC
Exact
Godunov
Siddhartha Mishra High order entropy stable schemes
Page 68
Issues
I Residual in equivalent equation large at strong shocks
I
Ut + Fx = h(B(U)Ux
)x
+R, R =∞∑
q=2p+1
C 2pq (h)q−1
q![Fq(U)]q,
I High-order schemes reduce the residual LeFloch, SM,forthcoming.
I Modified shock capturing operators in space-time DG.
Siddhartha Mishra High order entropy stable schemes
Page 69
Comparison
Standard paradigm New paradigm
Approximate Riemann solvers Entropy conservative flux+Num diff in entropy var +Matches physical viscosity
BV reconstructions Sign property preserving recon (SG)Shock-capturing STDG (UG)
SSP-RK time stepping SSP-RK time stepping ??(Implicit) space-time DG.
Siddhartha Mishra High order entropy stable schemes