Chapter5.08a PDF

Numerical Methods for Differential Equations
Chapter 5: Partial differential equations elliptic and parabolic
Gustaf Soderlind
and Carmen Arevalo
Numerical Analysis, Lund University
Textbooks: A First Course in the Numerical Analysis of Differential Equations, by Arieh Iserles
and Introduction to Mathematical Modelling with Differential Equations, by Lennart Edsberg
c Gustaf Soderlind,

Numerical Analysis, Mathematical Sciences, Lund University, 2008-09
Numerical Methods for Differential Equations p. 1/50
1. Brief overview of PDE problems

Classification: Three basic types, four prototype equations
Elliptic
u = 0
+ BC
Parabolic
ut = u
+ BC & IC
utt = u
+ BC & IC
Hyperbolic
ut + a(u)ux = 0
+ BC & IC
Classification of PDEs
Linear PDE with two independent variables
Auxx + 2Buxy + Cuyy + L(ux , uy , u, x, y) = 0
with L linear in ux , uy , u. Study
A B
= AC B 2
:= det
B C
Elliptic > 0
Parabolic = 0
Hyperbolic < 0
Standard PDEs. Prototypical problems
>0
=0
<0
Elliptic PDE
Laplace equation
uxx + uyy = 0
A = C = 1; B = 0
Parabolic PDE
Diffusion equation
ut = uxx
A = 1; B = C = 0
Hyperbolic PDEs
Wave equation
utt = uxx
A = 1; B = 0; C = 1
PDE method types
FDM
Finite difference methods
FEM
Finite element methods
FVM
Finite volume methods
BEM
Boundary element methods
We will mostly study FDM to cover basic theory

Industrial relevance: FEM
PDE methods for elliptic problems
Simple geometry
FDM or Fourier methods
Complex geometry
FEM
Special problems
FVM or BEM
Large sparse systems

Combine with iterative solvers such as multigrid methods
PDE methods for parabolic problems
Simple geometry
FDM or Fourier methods
Complex geometry
FEM
Stiffness
Always use Astable time-stepping methods
Need Newton-type solvers for large sparse systems
PDE methods for hyperbolic problems

FDM, FVM. Sometimes FEM
Shocks
Solutions may be discontinuous example: sonic boom
Turbulence
Multiscale phenomena
Hyperbolic problems have several complications and many

highly specialized techniques are often needed
2. Elliptic problems. FDM

Laplacian
2
2
2
=
+ 2+ 2
2
x
y
z
Laplace equation
u = 0
with boundary conditions u = u0 (x, y, z)
Poisson equation
u = f
with boundary conditions u = u0 (x, y, z)
x, y, z
x, y, z
Other boundary conditions also of interest ( = bdry of )
Elliptic problems. Some applications

Equilibrium problems
Structural analysis (strength of materials)
Heat distribution
Potential problems
Potential flow (inviscid, subsonic flow)
Electromagnetics (fields, radiation)
Eigenvalue problems
Acoustics
Microphysics
Poisson equation an elliptic model problem

2u 2u
+ 2 = f (x, y)
2
x
y
Computational domain = [0, 1] [0, 1] (unit square)
Dirichlet conditions u(x, y) = 0 on boundary
Uniform grid {xi , yj }N,M
i,j=1 with equidistant mesh widths
x = 1/(N + 1) and y = 1/(M + 1)
Discretization Finite differences with ui,j u(xi , yj )
ui1,j 2ui,j + ui+1,j ui,j1 2ui,j + ui,j+1
+
= f (xi , yj )
2
2
x
y
Equidistant mesh x = y
ui1,j + ui,j1 4ui,j + ui,j+1 + ui+1,j
= f (xi , yj )
2
x
Participating approximations and mesh points
yj+1
yj
yj1
xi1
xi
xi+1
Computational stencil for x = y

ui1,j + ui,j1 4ui,j + ui,j+1 + ui+1,j
= f (xi , yj )
2
x
Five-point operator
1
The FDM linear system of equations

Lexicographic ordering of unknowns partitioned system
0 ...
2
x
...
u,1
u,2
u,3
.
..
..
. I
I
with Toeplitz matrix T = tridiag(1
u,N
f (x , y1 )

f (x , y2 )

f (x , y3 )

=
..

.

f (x , yN )
1)
The system is N 2 N 2 , hence large and very sparse

3. Elliptic problems. FEM

Finite Element Method
PDE
Ansatz
Requirement
Lu = 0
P
u=
ci i
hi , Lui = 0
Lu =
ci Li
gives coefficients {ci }
FEM is a least squares approximation, fitting a linear

combination of basis functions {i } to the differential
equation using orthogonality
Simplest case Piecewise linear basis functions
Strong and weak forms

Strong form
u = f ;
u = 0 on
Take v with v = 0 on
Z
Z
u v =
f v
Integrate by parts to get weak form

Z
Z
u v =
f v
Strong and weak forms. 1D case

Recall integration by parts in 1D
Z 1
Z
1
u v = [u v]0 +
0
u v
or in terms of an inner product

hu , vi = hu , v i
Generalization to 2D, 3D uses vector calculus
Weak form of u = f
Define inner product
hv, ui =
vu d
and energy norm (note scalar product!)

Z
a(v, u) =
v u d
to get the weak form of u = f
as
a(v, u) = hv, f i
Galerkin method (Finite Element Method)

1. Basis functions
2. Approximate
3. Determine cj from
{i }
P
u = cj j
R
P R
cj i j = f i
The cj are determined by the linear system

Kc = F
The matrix K is called stiffness matrix
R
Stiffness matrix elements kij = i j = a(i , j )

R
Right-hand side Fi = i f = hi , f i
The FEM mesh. Domain triangulation

Piecewise linear basis {j } require domain triangulation
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
4. Parabolic problems
The prototypical equation is the
Diffusion equation ut = u
Nonlinear diffusion
ut = div (k(u)gradu)
Boundary and initial conditions are needed
Parabolic problems. Some applications

Diffusive processes
Heat conduction
ut = d uxx
Chemical reactions
Reactiondiffusion
ut = d uxx + f (u)
Convectiondiffusion
ut = ux +
1
u
Pe xx
Irreversibility ut = u is not well-posed!

Diffusion a parabolic model problem

Equation
Initial values
Boundary values
ut = uxx
u(x, 0) = g(x)
u(0, t) = u(1, t) = 0
Separation of variables u(x, t) := X(x)T (t)

ut = X T ,
T = Ce
uxx = X T
X
T
=
=:
T
X
X = A sin x + B cos x
Parabolic model problem. . .

Boundary values
X(0) = X(1) = 0 k = (k)2 , therefore
Xk (x) =
2 sin kx
Initial values
Fourier expansion g(x) =
P
1
(k)2 t
Tk (t) = e
ck 2 sin kx
X
(k)2 t
ck e
sin kx
u(x, t) = 2
k=1
5. Method of lines (MOL) discretization

In ut = uxx , discretize 2 /x2 by
uxx
ui1 2ui + ui+1
x2
System of ODEs (semidiscretization) u = Tx u
1
1 1 2
u
u =
...
x2
1 2
Full FDM discretization

Note that ui (t) u(x, t) along the line x = xi in (x, t) plane
Use time-stepping to solve the IVP; Explicit Euler with
ui,j u(xi , tj ) implies
ui,j+1 ui,j
ui1,j 2ui,j + ui+1,j
=
t
x2
With the Courant number = t/x2 we obtain recursion
ui,j+1 = ui,j + (ui1,j 2ui,j + ui+1,j )
Method of lines. The grid

Rectangular grid {(i x, j t), i = 0 : N + 1, j 0} with
x = 1/(N + 1)
8
ui,j u(i x, j t)
Method of lines. Computational stencil

Explicit Euler time stepping. Participating grid points
tj+1
tj
xi1
xi
xi+1
1
Courant number = t/x2
21
Method of lines. Stability and the CFL condition

Explicit Euler with x = 1/(N + 1) implies recursion
u,j+1 = u,j + t Tx u,j
k
Recall k [Tx ] = 4(N + 1) sin
for k = 1 : N
2(N + 1)
2
Stability requires t k S for all eigenvalues

1.5
0.5
Im

4t
2
t k
,
t
2
x
0.5
1.5
2.5
1.5
1
Re
0.5
0.5
The CFL condition

For stability we need 4t/x2 2
CFL condition (Courant, Friedrichs, Lewy 1928)
1
t
2
x
2
The CFL condition is a severe restriction on time step t

Stiffness The CFL condition can be avoided by using
A-stable methods, e.g. Trapezoidal Rule or Implicit Euler
Experimental stability investigation

N = 30 internal pts in [0, 1], M = 187 time steps on [0, 0.1]
Stable solution at CFL = .514
0.8
0.6
0.4
0.2
0
1
0.8
0.1
0.6
0.08
0.06
0.4
0.04
0.2
0.02
0
Violating the CFL condition. Instability

Unstable solution at CFL = .522
1
0.8
0.6
0.4
0.2
0
0.2
0.4
1
0.8
0.1
0.6
0.08
0.06
0.4
0.04
0.2
0.02
0
CrankNicolson method (1947)

CrankNicolson method Trapezoidal Rule for PDEs
The trapezoidal rule is
implicit
Astable
more work/step
no restriction on t
Theorem CrankNicolson is unconditionally stable

There is no CFL condition on the time-step t
CrankNicolson method. . .
Courant number = t/x2
recursion
(I T )u,j+1 = (I + T )u,j
2
2
with Toeplitz matrix T = tridiag(1 2 1)
Tridiagonal structure low complexity
Refactorize only if Courant number = t/x2 changes
6. Error analysis. Convergence

MOL with explicit Euler for ut = uxx
Global error ei,j = ui,j u(xi , tj )
Local error Insert exact solution to get
u(xi , tj+1 ) u(xi , tj )
=
t
u(xi1 , tj ) 2u(xi , tj ) + u(xi+1 , tj )
li,j
=
2
x
Expand local error li,j in Taylor series
Local error
Taylor expansion
li,j
t
x2
= ut uxx +
utt +
uxxxx + O(t2 , x4 )
2
12
Therefore
li,j
t
x2
=
utt +
uxxxx + O(t2 , x4 )
2
12
The Lax Principle

Conclusion
Consistency
Stability
Convergence
li,j 0
CFL condition
ei,j 0
as t, x 0
t/x2 1/2
as t, x 0
Theorem (Lax Principle)

Consistency + Stability Convergence
Note Choice of norm is very important
The order of the method

With local error
li,j
t
x2
=
utt +
uxxxx = O(t, x2 )
2
12
and stability in terms of CFL condition = t/x2 1/2

we have global error ei,j = O(t, x2 )
For fixed we have t x2 and it follows that
Global error ei,j = O(t, x2 ) = O(x2 )
Theorem The order of convergence is p = 2
Order of PDE discretizations

Discretization process from PDE to SD to FD
j
j
+
k
=
A
u
ut = uxx v x = Px vx + hx (t) uj+1

Suppose
order of SD scheme for spatial variables is p1
order of ODE time discretization is p2
Theorem If t = x2 the FD order of convergence is
p = min{p1 , 2p2 }
Example. The CrankNicolson method

From Trapezoidal rule yn+1 = yn + h2 [f (yn ) + f (yn+1 )]
uj+1
i
uji
t
j+1
j+1
j+1
j
j
j
+
u
+
2u
)
+
(u
+
u
2u
[(u
i+1 )]
i
i1
i+1
i
i1
2
2x
j+1
j+1 j
j
j+1
j
ui1 + (1 + )ui ui+1 = ui1 + (1 )ui + ui+1
2
2
2
2
Same order p = min{2, 4} = 2 as with Explicit Euler
tj+1
tj
xi1
xi
xi+1
CrankNicolson. Stability
1
A = (I T ) (I + T ) with usual Toeplitz matrix T

2
2
1 + 2 [T ]
Theorem The eigenvalues are [A ] =
1 2 [T ]
Note [T ] (4, 0) 1 < [A ] < 1 This implies that
there is no CFL stability condition on the Courant ratio !
The method is stable for all t > 0
Theorem CrankNicolson is unconditionally stable
Experimental stability investigation

Stable solution at CFL = 3.2
0.8
0.6
0.4
0.2
0
1
0.8
0.1
0.6
0.08
0.06
0.4
0.04
0.2
0.02
0
7. Parabolic problems. FEM

Consider diffusion problem in strong form ut uxx = 0 with
Dirichlet boundary conditions
Multiply by test function v and integrate by parts
Z 1
Z 1
vut dx +
v u dx = 0
0
In terms of inner product and energy norm

Weak form hv, ut i + a(v, u) = 0 for all v with v(0) = v(1) = 0
Galerkin method (Finite Element Method)

{i }
1. Basis functions
u(t, x) =
2. Approximate
3. Determine cj from
Note hi , ut i =
cj (t)j (x)
hi , ut i + a(i , u) = 0
cj hi , j i and a(i , u) =
cj hi , j i
We get an initial value problem

Bx c + Kx c = 0
for the determination of the coefficients cj (t) with c(0)
determined by the initial condition
Galerkin method
Simplest case
Piecewise linear basis functions on equidistant grid
Stiffness matrix elements kij = hi , j i
Kx
1
=
tridiag(1 2 1)
x
and mass matrix elements bij = hi , j i

Bx
x
=
tridiag(1 4
6
1)
Simplest Galerkin FEM. . .

Note that in the initial value problem
Bx c + Kx c = 0
the matrix Bx is tridiagonal no advantage from
explicit time stepping methods
Explicit Euler
Bx (cn+1 cn ) = t Kx cn
requires the solution of a tridiagonal system on every step
Simplest Galerkin FEM. . .

As the system is stiff, consider implicit A-stable method
t
Bx (cn+1 cn ) =
Kx (cn + cn+1 )
2
and solve tridiagonal system
t
t
(Bx +
Kx )cn+1 = (Bx
Kx )cn
2
2
on every step
Trapezoidal rule has same cost, but better stability
8. Well-posedness
Linear partial differential equation
ut = Lu + f , 0 x 1,
u(0, t) = 0 (t),
Suppose
u(1, t) = 1 (t)
wt = Lw + f,
vt
= Lv + f,
t 0,
u(x, 0) = h(x),
w(x, 0) = h(x)
v(x, 0) = h(x) + g(x)
Subtract to get homogeneous Dirichlet problem

ut = Lu,
u(x, 0) = g(x),
0 (t) 0,
1 (t) 0
Well-posedness. Time evolution

Suppose time evolution u(x, t) = E(t)g(x)
Definition The equation is well-posed if for every t > 0
there is a constant 0 < C(t ) < such that
kE(t)k C(t ) for all 0 t t
Theorem A well-posed equation has a solution that
depends continuously on the initial value (the data)
is uniformly bounded in any compact interval
ut = uxx is well posed

P
Fourier series expansion g(x) = 2 1 ck sin kx implies
X
(k)2 t
u(x, t) = 2
ck e
sin kx
k=1
kE(t)gk22 =
= 2
1
0
|u(x, t)|2 dx
ck cj e
(k2 +j 2 ) 2 t
k=1 j=1
X
k=1
2 2(k)2 t
ck e
X
k=1
Hence kE(t)k2 1 for every t 0
sin kx sin jx dx
0
c2k = kgk22

Chapter5.08a PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chapter5.08a PDF

Uploaded by

Copyright:

Available Formats

Numerical Methods for Differential Equations

Chapter 5: Partial differential equations elliptic and parabolic

Numerical Methods for Differential Equations p. 1/50

1. Brief overview of PDE problems

Numerical Methods for Differential Equations p. 2/50

Numerical Methods for Differential Equations p. 3/50

Standard PDEs. Prototypical problems

Numerical Methods for Differential Equations p. 4/50

PDE method types

Finite difference methods

Finite element methods

Finite volume methods

Boundary element methods

We will mostly study FDM to cover basic theory

Numerical Methods for Differential Equations p. 5/50

PDE methods for elliptic problems

FDM or Fourier methods

Large sparse systems

Numerical Methods for Differential Equations p. 6/50

PDE methods for parabolic problems

FDM or Fourier methods

Numerical Methods for Differential Equations p. 7/50

PDE methods for hyperbolic problems

Hyperbolic problems have several complications and many

Numerical Methods for Differential Equations p. 8/50

2. Elliptic problems. FDM

Other boundary conditions also of interest ( = bdry of )

Numerical Methods for Differential Equations p. 9/50

Elliptic problems. Some applications

Poisson equation an elliptic model problem

Computational stencil for x = y

The FDM linear system of equations

with Toeplitz matrix T = tridiag(1

The system is N 2 N 2 , hence large and very sparse

3. Elliptic problems. FEM

gives coefficients {ci }

FEM is a least squares approximation, fitting a linear

Numerical Methods for Differential Equations p. 15/50

Strong and weak forms

Integrate by parts to get weak form

Numerical Methods for Differential Equations p. 16/50

Strong and weak forms. 1D case

or in terms of an inner product

Numerical Methods for Differential Equations p. 17/50

and energy norm (note scalar product!)

to get the weak form of u = f

Galerkin method (Finite Element Method)

The cj are determined by the linear system

Stiffness matrix elements kij = i j = a(i , j )

The FEM mesh. Domain triangulation

Numerical Methods for Differential Equations p. 20/50

Boundary and initial conditions are needed

Numerical Methods for Differential Equations p. 21/50

Parabolic problems. Some applications

Irreversibility ut = u is not well-posed!

Diffusion a parabolic model problem

Separation of variables u(x, t) := X(x)T (t)

Numerical Methods for Differential Equations p. 23/50

Parabolic model problem. . .

Numerical Methods for Differential Equations p. 24/50

5. Method of lines (MOL) discretization

ui1 2ui + ui+1

System of ODEs (semidiscretization) u = Tx u