Discontinuous Galerkin and Finite Difference Methods for ...
Transcript of Discontinuous Galerkin and Finite Difference Methods for ...
![Page 1: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/1.jpg)
About Me
Mario BencomoCurrently 2nd year graduate student in CAAM departmentat Rice University.B.S. in Physics and Applied Mathematics (Dec. 2010).Undergraduate University: University of Texas at El Paso(UTEP).
1
![Page 2: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/2.jpg)
Discontinuous Galerkin and Finite DifferenceMethods for the Acoustic Equations with
Smooth Coefficients
Mario BencomoTRIP Review Meeting 2013
2
![Page 3: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/3.jpg)
Problem Statement
Acoustic Equations (pressure-velocity form):
ρ(x)∂v∂ t
(x, t) + ∇p(x, t) = 0 (1a)
1κ
(x)∂p∂ t
(x, t) + ∇ ·v(x, t) = f (x, t) (1b)
for x ∈ Ω and t ∈ [0,T ], where x = (x ,z) and Ω = [0,1]2.Boundary and initial conditions:
p = 0, on ∂ Ω× [0,T ]
p(x,0) = p0(x) and v(x,0) = v0(x)
Research focus:Analyze the computational efficiency of discontinuousGalerkin (DG) and finite difference (FD) methods in thecontext of the acoustic equations with smoothcoefficients.
3
![Page 4: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/4.jpg)
Problem Statement
Acoustic Equations (pressure-velocity form):
ρ(x)∂v∂ t
(x, t) + ∇p(x, t) = 0 (1a)
1κ
(x)∂p∂ t
(x, t) + ∇ ·v(x, t) = f (x, t) (1b)
for x ∈ Ω and t ∈ [0,T ], where x = (x ,z) and Ω = [0,1]2.Boundary and initial conditions:
p = 0, on ∂ Ω× [0,T ]
p(x,0) = p0(x) and v(x,0) = v0(x)
Research focus:Analyze the computational efficiency of discontinuousGalerkin (DG) and finite difference (FD) methods in thecontext of the acoustic equations with smoothcoefficients.
3
![Page 5: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/5.jpg)
Why Smooth Coefficients?
Relevant for seismic applications: smooth trends in realdata!Comparison has not been done before!Previous work with discontinuous coefficients(Wang, 2009).
efficiency of DG over 2-4 FD (dome inclusion)DG resolves discontinuity interface errors
4
![Page 6: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/6.jpg)
Focus of Talk
1 Discontinuous Galerkin (DG) Methodderivation of schemebasis functionsreference element
2 Locally conforming DG (LCDG) Methodtriangulationreference elementbasis functions
3 Summary and Future Work
5
![Page 7: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/7.jpg)
Discontinuous Galerkin (DG) Method
Why DG? (Cockburn, 2006; Brezzi et al., 2004; Wilcox et al., 2010)
explicit time-stepping, after inverting block diagonalmatrixcan handle irregular meshes and complex geometrieshp-adaptivity
Idea: Approximate solution by piecewise polynomials onpartitioned domain.Example: 1D piecewise linear approximation
6
![Page 8: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/8.jpg)
Discontinuous Galerkin (DG) Method
Why DG? (Cockburn, 2006; Brezzi et al., 2004; Wilcox et al., 2010)
explicit time-stepping, after inverting block diagonalmatrixcan handle irregular meshes and complex geometrieshp-adaptivity
Idea: Approximate solution by piecewise polynomials onpartitioned domain.Example: 1D piecewise linear approximation
6
![Page 9: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/9.jpg)
Derivation of DG Semi-Discrete Scheme
Let:τ ∈T , for some triangulation T on Ω
test function w ∈ C∞(Ω)
Then,
ρ∂vx
∂ t+
∂p∂x
= 0 =⇒∫
τ
ρ∂vx
∂ tw dx +
∫τ
∂p∂x
w dx = 0
I.B.P. and replace p with numerical flux p∗ in boundaryintegral, ∫
τ
ρ∂vx
∂ tw dx−
∫τ
p∂w∂x
dx +∫
∂τ
p∗w nx dσ = 0
IBP=⇒
∫τ
ρ∂vx
∂ tw dx +
∫τ
∂p∂x
w dx +∫
∂τ
(p∗−p)w nx dσ = 0 (2)
7
![Page 10: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/10.jpg)
Derivation of DG Semi-Discrete Scheme
Let:τ ∈T , for some triangulation T on Ω
test function w ∈ C∞(Ω)
Then,
ρ∂vx
∂ t+
∂p∂x
= 0 =⇒∫
τ
ρ∂vx
∂ tw dx +
∫τ
∂p∂x
w dx = 0
I.B.P. and replace p with numerical flux p∗ in boundaryintegral, ∫
τ
ρ∂vx
∂ tw dx−
∫τ
p∂w∂x
dx +∫
∂τ
p∗w nx dσ = 0
IBP=⇒
∫τ
ρ∂vx
∂ tw dx +
∫τ
∂p∂x
w dx +∫
∂τ
(p∗−p)w nx dσ = 0 (2)
7
![Page 11: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/11.jpg)
Nodal Basis Functions
Finite dimensional space Wh:
Wh = w : w |τ ∈ PN(τ),∀τ ∈T
Nodal DG: Use nodal basis, i.e.,
PN(τ) = span`τ
j (x)N∗j=1 ∀τ ∈T ,
whereLagrange polynomials `τ
j (xτ
i ) = δij for given nodal setxτ
i N∗
i=1 ⊂ τ
N∗ = 12(N + 1)(N + 2), a.k.a., degrees of freedom per
triangular element
8
![Page 12: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/12.jpg)
Nodal Basis Functions
Finite dimensional space Wh:
Wh = w : w |τ ∈ PN(τ),∀τ ∈T
Nodal DG: Use nodal basis, i.e.,
PN(τ) = span`τ
j (x)N∗j=1 ∀τ ∈T ,
whereLagrange polynomials `τ
j (xτ
i ) = δij for given nodal setxτ
i N∗
i=1 ⊂ τ
N∗ = 12(N + 1)(N + 2), a.k.a., degrees of freedom per
triangular element
8
![Page 13: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/13.jpg)
Nodal Sets
Example of nodal sets xτ
j N∗
j=1 :
(a) N = 1,N∗ = 3 (b) N = 2,N∗ = 6 (c) N = 3,N∗ = 10
9
![Page 14: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/14.jpg)
DG Semi-Discrete Scheme
Find vx ,p ∈Wh such that∫τ
ρ∂vx
∂ tw dx−
∫τ
∂p∂x
w dx +∫
∂τ
(p∗−p)w nx dσ = 0
for all w ∈Wh and τ ∈T .
Note:
vx ∈Wh =⇒ vx (x, t)|τ =N∗
∑j=1
vx (xτ
j , t)`τ
j (x),
where vx (xτ
j , t)N∗
j=1 are unknowns. Same for vz and p, andnumerical fluxes v∗x ,v∗z ,p∗.
10
![Page 15: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/15.jpg)
Nodal Coefficient Vectors
RN∗
vx (t) := [vx (xτ
1, t),vx (xτ
2, t), . . . ,vx (xτ
N∗ , t)]T
vz(t) := · · ·p(t) := · · ·
RN+1
vmn (t) := [vn(xτ
m1, t),vn(xτ
m2, t), . . . ,vn(xτ
mN+1, t)]T
pm(t) := · · ·(vm
n )∗(t) := [v∗n(xτm1
, t),v∗n(xτm2
, t), . . . ,v∗n(xτmN+1
, t)]T
(pm)∗(t) := · · ·
where vn(xτmj, t) = nm
x vx (xτmj, t) + nm
z vz(xτmj, t), and similar for
v∗n(xτmj, t).
11
![Page 16: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/16.jpg)
DG Semi-Discrete Scheme
From ∫τ
ρ∂vx
∂ tw dx +
∫τ
∂p∂x
w dx +∫
∂τ
(p∗−p)w nx dσ = 0,
to
MRddt
vx (t) + Sxp(t) +3
∑m=1
nmx Mm ((pm)∗−pm)(t) = 0,
where:
mass matrix (M)ij :=∫
τ
`τ
i `τ
j dx, in RN∗×N∗
mass matrix (Mm)ij :=∫
emτ
`τ
i `τmj
dσ , in RN∗×(N+1)
stiffness matrix (Sx )ij :=∫
τ
`τ
i
∂`τ
j
∂xdx, in RN∗×N∗
ρ-matrix (R)ij := δijρ(xτ
j ), in RN∗×N∗
12
![Page 17: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/17.jpg)
DG Semi-Discrete Scheme
From ∫τ
ρ∂vx
∂ tw dx +
∫τ
∂p∂x
w dx +∫
∂τ
(p∗−p)w nx dσ = 0,
to
MRddt
vx (t) + Sxp(t) +3
∑m=1
nmx Mm ((pm)∗−pm)(t) = 0,
where:
mass matrix (M)ij :=∫
τ
`τ
i `τ
j dx, in RN∗×N∗
mass matrix (Mm)ij :=∫
emτ
`τ
i `τmj
dσ , in RN∗×(N+1)
stiffness matrix (Sx )ij :=∫
τ
`τ
i
∂`τ
j
∂xdx, in RN∗×N∗
ρ-matrix (R)ij := δijρ(xτ
j ), in RN∗×N∗
12
![Page 18: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/18.jpg)
DG Semi-Discrete Scheme
System of ODE’s:
Rddt
vx (t) =−Dxp(t) +3
∑m=1
nmx Lm ((pm)∗−pm)(t)
for each τ ∈T , where
Dx = M−1Sx , Lm = M−1Mm.
Similar result for other equations:
Rddt
vz(t) =−Dzp(t) +3
∑m=1
nmz Lm ((pm)∗−pm)(t)
K −1 ddt
p(t) = f(t)−Dxvx (t)−Dzvz(t)−3
∑m=1
Lm ((vn)∗−vn)(t)
where
(K )ij = δijκ(xτ
j ), (f)i =∫
τ
f `τ
i dx, Dz = M−1Sz .
13
![Page 19: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/19.jpg)
DG Semi-Discrete Scheme
System of ODE’s:
Rddt
vx (t) =−Dxp(t) +3
∑m=1
nmx Lm ((pm)∗−pm)(t)
for each τ ∈T , where
Dx = M−1Sx , Lm = M−1Mm.
Similar result for other equations:
Rddt
vz(t) =−Dzp(t) +3
∑m=1
nmz Lm ((pm)∗−pm)(t)
K −1 ddt
p(t) = f(t)−Dxvx (t)−Dzvz(t)−3
∑m=1
Lm ((vn)∗−vn)(t)
where
(K )ij = δijκ(xτ
j ), (f)i =∫
τ
f `τ
i dx, Dz = M−1Sz .
13
![Page 20: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/20.jpg)
DG Semi-Discrete Scheme
Acoustic equations:
ρ∂
∂ tvx =− ∂
∂xp
ρ∂
∂ tvz =− ∂
∂zp
1κ
∂
∂ tp = f −∇ ·v
DG scheme:
Rddt
vx (t) =−Dxp(t) +3
∑m=1
nmx Lm ((pm)∗−pm)(t)
Rddt
vz(t) =−Dzp(t) +3
∑m=1
nmz Lm ((pm)∗−pm)(t)
K −1 ddt
p(t) = f(t)−Dxvx (t)−Dzvz(t)−3
∑m=1
Lm ((vn)∗−vn)(t)
14
![Page 21: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/21.jpg)
Visualizing DG
After time discretization (leapfrog):
vn+1/2x = vn−1/2
x −∆tR−1
[Dxpn +
3
∑m=1
nmx Lm ((pm)∗−pm)
n
]
tn+1/2
tn
tn−1/2
15
![Page 22: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/22.jpg)
Visualizing DG
After time discretization (leapfrog):
vn+1/2x = vn−1/2
x −∆tR−1[Dxpn+3
∑m=1
nmx Lm((pm)∗−pm)
n]
tn+1/2
tn
tn−1/2
16
![Page 23: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/23.jpg)
Visualizing DG
After time discretization (leapfrog):
vn+1/2x = vn−1/2
x −∆tR−1[Dxpn+3
∑m=1
nmx Lm((pm)∗−pm)
n]
tn+1/2
tn
tn−1/2
17
![Page 24: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/24.jpg)
Visualizing DG
After time discretization (leapfrog):
vn+1/2x = vn−1/2
x −∆tR−1[Dxpn+3
∑m=1
nmx Lm((pm)∗−pm)
n]
tn+1/2
tn
tn−1/2
18
![Page 25: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/25.jpg)
Use of Reference Element in DG
Reference triangle τ:
(-1,-1) (1,-1)
(-1,1)
r
s
Idea: Carry computations in τ
construct nodal set and basis functions in τ:
`jN∗
j=1 s.t. `j(ri) = δij for riN∗
i=1 ⊂ τ
mass matrix computations∫τ
`τ
j `τ
i dx = J(τ)∫
τ
`j`i dr =⇒M = J(τ)M
19
![Page 26: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/26.jpg)
Locally Conforming DG (LCDG) Method for Acoustics
Why LCDG method? (Chung & Engquist, 2006, 2009)locally and globally energy conservativeoptimal convergence rateexplicit time-step
My contribution: nodal DG implementation of LCDGbasis functions and nodal setsuse of reference element
20
![Page 27: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/27.jpg)
Locally Conforming DG (LCDG) Method for Acoustics
Why LCDG method? (Chung & Engquist, 2006, 2009)locally and globally energy conservativeoptimal convergence rateexplicit time-step
My contribution: nodal DG implementation of LCDGbasis functions and nodal setsuse of reference element
20
![Page 28: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/28.jpg)
Locally Conforming DG (LCDG) Method for Acoustics
LCDG Idea: Enforce continuity of the pressure field and thenormal component of the velocity field in a “staggered” manner.
Example: 1D piecewise linear polynomial approximation
21
![Page 29: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/29.jpg)
Triangulation
Algorithm: unif_tri1 given Nsub, subdivide Ω into Nsub×Nsub partition of
squares2 triangulate each square by adding a diagonal from top-left
to bottom-right3 Pick interior point at each triangle and re-triangulate
Figure: Example of unif_tri for Nsub = 2.
22
![Page 30: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/30.jpg)
Triangulation
Algorithm: unif_tri1 given Nsub, subdivide Ω into Nsub×Nsub partition of
squares2 triangulate each square by adding a diagonal from top-left
to bottom-right3 Pick interior point at each triangle and re-triangulate
Figure: Example of unif_tri for Nsub = 2.
22
![Page 31: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/31.jpg)
Triangulation
Algorithm: unif_tri1 given Nsub, subdivide Ω into Nsub×Nsub partition of
squares2 triangulate each square by adding a diagonal from top-left
to bottom-right3 Pick interior point at each triangle and re-triangulate
Figure: Example of unif_tri for Nsub = 2.
22
![Page 32: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/32.jpg)
Triangulation
Algorithm: unif_tri1 given Nsub, subdivide Ω into Nsub×Nsub partition of
squares2 triangulate each square by adding a diagonal from top-left
to bottom-right3 Pick interior point at each triangle and re-triangulate
Figure: Example of unif_tri for Nsub = 2.
22
![Page 33: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/33.jpg)
Triangulation
Algorithm: unif_tri1 given Nsub, subdivide Ω into Nsub×Nsub partition of
squares2 triangulate each square by adding a diagonal from top-left
to bottom-right3 Pick interior point at each triangle and re-triangulate
Figure: Example of unif_tri for Nsub = 2.
22
![Page 34: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/34.jpg)
LCDG Approximation Spaces
Pressure space Ph: q ∈Ph if
q|τ ∈ PN(τ)
q continuous on dashed edges
q = 0 on ∂ Ω
Velocity space Vh: u ∈ Vh if
u|τ ∈ PN(τ)2
u ·n continuous dashed edges
23
![Page 35: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/35.jpg)
LCDG Approximation Spaces
Pressure space Ph: q ∈Ph if
q|τ ∈ PN(τ)
q continuous on dashed edges
q = 0 on ∂ Ω
Velocity space Vh: u ∈ Vh if
u|τ ∈ PN(τ)2
u ·n continuous dashed edges
23
![Page 36: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/36.jpg)
LCDG: Basis for Ph
Two types of pressure elements:
boundary elements interior elements
Reference elements:
r
s
(-1,-1) (1,-1)
(-1,1)
r
s
(-1,-1) (1,-1)
(1,1)(-1,1)
24
![Page 37: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/37.jpg)
LCDG: Basis for Ph
Basis functions for boundary pressure elements(N = 1 example):
1
r
s
Basis functions for pressure interior elements(N = 1 example):
1
r
s1
r
s1
r
s1
r
s
25
![Page 38: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/38.jpg)
LCDG: Basis for Vh
Velocity elements and respective reference elements:
velocity elementsr
s
(−1,−13
√3) (1,−1
3
√3)
(0, 23√3)
(0, 0)
reference element
26
![Page 39: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/39.jpg)
LCDG: Basis for Vh
Basis functions, N = 1 example:
Figure: Basis vector fields φ γ,j for N = 1 (Nγ = 12).
27
![Page 40: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/40.jpg)
LCDG Semi-Discrete Scheme
For a triangulation T , find p ∈Ph and v ∈ Vh such that∫Ω
ρ∂v∂ t·u dx−B∗h(p,u) = 0
∫Ω
1κ
∂p∂ t
q dx + Bh(v,q) =∫
Ωfq dx
for all q ∈Ph and u ∈ Vh, where
B∗h(p,u) =−∫
Ωp ∇ ·u dx + ∑
e∈E 0p
∫e
p [u · n] dσ
Bh(v,q) =∫
Ωv ·∇q dx− ∑
e∈Ev
∫e
v · n[q] dσ .
28
![Page 41: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/41.jpg)
Summary
standard DGmotivationbasis functionsreference element
LCDGtriangulationspaces Ph, Vhreference elements and basis functions
29
![Page 42: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/42.jpg)
Pending Work and Future Directions
Pending work:implementation of LCDGabsorbing BC (Chung & Engquist, 2009)time discretization (leapfrog and Runge-Kutta)error analysis
Future directions:non-uniform triangulation3D acousticselasticity equations
30
![Page 43: Discontinuous Galerkin and Finite Difference Methods for ...](https://reader031.fdocuments.in/reader031/viewer/2022022314/6215dcbbebf94b5cc47c9eb6/html5/thumbnails/43.jpg)
References
Brezzi, F., Marini, L., and Sïli, E. (2004). “Discontinuous galerkin methods for first-order hyperbolic problems.”Mathematical models and methods in applied sciences, 14(12):1893-1903.
Chung, E. and Engquist, B. (2009). “Optimal discontinuous galerkin methods for the acoustic wave equation inhigher dimensions.” SIAM Journal on Numerical Analysis, 47(5):3820-3848.
Chung, E. T. and Engquist, B. (2006). “Optimal discontinuous galerkin methods for wave propagation.” SIAM Journalon Numerical Analysis, 44(5):2131-2158.
Cockburn, B. (2003). “Discontinuous Galerkin methods.” ZAMM-Journal of Applied Mathematics and Mechanics/Zeitschrift für Angewandte Mathematik und Mechanik, 83(11):731-754.
Wilcox, Lucas C., et al. “A high-order discontinuous Galerkin method for wave propagation through coupledelasticÐacoustic media.” Journal of Computational Physics 229.24 (2010): 9373-9396.
Hesthaven, J. S. and Warburton, T. (2007). “Nodal discontinuous Galerkin methods: algorithms, analysis, andapplications,” volume 54. Springer.
Wang, X. (2009). “Discontinuous Galerkin Time Domain Methods for Acoustics and Comparison with FiniteDifference Time Domain Methods.” Master’s thesis, Rice University, Houston, Texas.
Warburton, T. and Embree, M. (2006). “The role of the penalty in the local discontinuous Galerkin method forMaxwell’s eigenvalue problem.” Computer methods in applied mechanics and engineering, 195(25):3205-3223.
31