Transcript Slide 1

Mutigrid Methods
for Solving Differential Equations
Ferien Akademie ’05 – Veselin Dikov
Multigrid Methods
Agenda
● Model problem
● Relaxation. Smoothing property
● Elements of Multigrid
● Multigrid schemes
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Model Problem
● 1D boundary problem of steady state
temperature a long a uniform rod
 u" ( x)  u ( x)  f ( x),
u (0)  u (1)  0
0  x  1,
 0
● Discretization in n points, step h = 1/n
 v j 1  2v j  v j 1
2
h
v0  v n  0
 v j  f j ,
Ferien Akademie ’05
1  j  n  1,
Veselin Dikov
Multigrid Methods
Model Problem
T
● Av = f, where v  v1 ,...,vn1  and f   f1 ,..., fn1 
T
2  h 2

1

1
A 2 
h 



● Stencil notation
1
2  h 2
.


1


.
.

.
.
1 
 1 2  h 2 
1
A  2 (1 2  h 2  1)
h
● A is Symmetric positive definite
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Agenda
● Model problem
● Relaxation. Smoothing property
● Elements of Multigrid
● Multigrid schemes
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Iterative Methods
● Iterative vs Direct methods
More about iterative methods
● Jacobi and Gauss-Seidel methods
● Smoothing property
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Smoothing Property
● Error along the domain
After 35 sweeps with
weighted Jacobi
Error was smoothed
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Smoothing Property
● Smoothing property explained in four steps
1. Fourier modes
vk , j
 jk 
 sin
,
 n 
0  j  n,
1  k  n 1
● k – wave number
Ferien Akademie ’05
Veselin Dikov
Smoothing Property
Multigrid Methods
● Smoothing property explained in four steps
1. Fourier modes
k=1
k=2
k=7
k = 12
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Smoothing Property
● Smoothing property explained in four steps
1. Fourier modes
● smooth modes -
n
1 k 
2
n
 k  n 1
● oscillatory modes 2
Ferien Akademie ’05
Veselin Dikov
Smoothing Property
Multigrid Methods
● Smoothing property explained in four steps
1. Fourier modes
2. Modified model problem
● f = 0, σ = 0
Au = 0
● exact solution: u = 0
● error: e = u – v = -v
we can trace the error!
Ferien Akademie ’05
Veselin Dikov
Smoothing Property
Multigrid Methods
● Smoothing property explained in four steps
1. Fourier modes
2. Modified model problem
3. Weighted Jacobi relaxation
● wJacobi step
v(k 1) R v(k )  c
 2 1

 1 2  1





R  I  A  I  
.
.
.
2
2

.
.

1



 1 2 
● error
e( m) R me(0)
Ferien Akademie ’05
Veselin Dikov
Smoothing Property
Multigrid Methods
● Smoothing property explained in four steps
1. Fourier modes
2. Modified model problem
3. Weighted Jacobi relaxation
4. Three experiments
● we relax with wJacobi with ω = 2/3 on
initial guesses respectively:
( 0)
( 0)
( 0)
v

v
,
v

v
and
v
 v6
1
3
e
# iterations
Ferien Akademie ’05
Veselin Dikov
Smoothing Property
Multigrid Methods
● Smoothing property explained in four steps
1. Fourier modes
2. Modified model problem
3. Weighted Jacobi relaxation
4. Three experiments
● repeat the experiment with:
e

ω = 2/3 and initial guess
1
1
(0)
v  v1  v3 v6
2
2
# iterations
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Smoothing Property
● Smoothing property explained in four steps
1. Fourier modes
2. Modified model problem
3. Weighted Jacobi relaxation
4. Three experiments
● Explanation
● Rω has the same eigenvectors as A and
they are the same as the wave vectors
● Recall that for the error e(m) = Rme(0)
● Eigenvalues of Rω ?
Ferien Akademie ’05
Veselin Dikov
Smoothing Property
Multigrid Methods
● Smoothing property explained in four steps
1. Fourier modes
2. Modified model problem
Eigenvalue
3. Weighted Jacobi relaxation
4. Three experiments
● Explanation
wavenumber k
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Smoothing Property
● Smoothing property explained in four steps
1. Fourier modes
2. Modified model problem
3. Weighted Jacobi relaxation
4. Three experiments
● Explanation
● Smoothing property
● Fast damping of oscillatory error modes
● Common for all iterative methods
● How to overcome the bad performance
effect over smooth error modes?
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Agenda
● Model problem
● Relaxation. Smoothing property
● Elements of Multigrid
● Multigrid schemes
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Elements of Multigrid
● Element I: A smooth wave looks more
oscillatory on a coarser grid
● Aliasing: k looks like (n-k)
Ferien Akademie ’05
Veselin Dikov
Elements of Multigrid
Multigrid Methods
● Element II: Nested Iterations
finest grid
Relax
Relax
coarsest grid
Relax
transfer the coarse grid
result to the finer grid for
the initial guess
● Problems?
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Elements of Multigrid
● Element III: Correction scheme
● Residual equation: Ae = r
● The scheme:
» Relax on Au = f on  to obtain an
h
approximation v .
» Compute r  f  Avh .
2h
» Relax on Ae = r on  to obtain an
2h
approximation to the error, e .
h
h
2h
v

v

e
» Correct the approximation
.
h
Ferien Akademie ’05
Veselin Dikov
Elements of Multigrid
Multigrid Methods
● Element IV: Interpolation and restriction
● Interpolation I 2hh :
v2h j  v 2j h ,


1 2h
v
 v j  v 2j h1 ,
2
● Restriction I h2 h :
h
2 j 1
n
0  j   1.
2
2h
h
Injection: v j  v2 j .


1 h
h
h
v

v

2
v

v
Full weighting:
2 j 1
2j
2 j 1 ,
4
● Variational property:
2h
j
I
h
2h
 ,
c I
2h T
h
Ferien Akademie ’05
n
1  j   1.
2
cR
Veselin Dikov
Multigrid Methods
Agenda
● Model problem
● Relaxation. Smoothing property
● Elements of Multigrid
● Multigrid schemes
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Two-Grid
● Two-Grid = Corr.Scheme+Interpolation+Restriction

v h  MG v h , f h

h
» Relax  1 times on Ah u h  f h on  with initial
h
guess v
» Compute r h  f h  Ah v h and restrict r 2h  I h2h r h.
2h 2h
2h
» Solve A e  r on  2 h.
h
h
h
» Interpolate e h  I 2hh e 2h and correct v  v  e .
h
» Relax  2 times on Ah u h  f h on  with initial
h
v
guess
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Two-Grid -> V-Cycle
● Two-Grid Scheme
● V-Cycle = Recursive Two-Grid Scheme
V-Cycle
Ferien Akademie ’05
W-Cycle
Veselin Dikov
Multigrid Methods
Full Multigrid(FMG)
● FMG = V-Cycle + nested iterations
FMG
Ferien Akademie ’05
Veselin Dikov
Multigrid Methods
Costs
● V-Cycle costs
d
1
1
1
2
n


Storage  2n d 1  d  2 d  ... L  
2
2  1  2d
 2
1
1
1
2
Computational cost  2
1



...

WU

d
2d
L 
d
2
2  1 2
 2
● FMG computational costs
1
1
2
 2  1
1



...

WU

2
 d 
d
2d
L
2
2  1  2 d
 1  2  2


● Speedup because working on smaller domains
Ferien Akademie ’05
Veselin Dikov