The Finite Difference Method for the Helmholtz Equation with

Download Report

Transcript The Finite Difference Method for the Helmholtz Equation with

The Finite Difference Method for
the Helmholtz Equation with
Applications to Cloaking
Li Zhang
Introduction
•
In the past few years, scientists have made
great progress in the field of cloaking.
Cloaking involves making an object invisible or
undetectable to electromagnetic waves.
•
In this paper, we will use the numerical
solutions to visualize the effect that the
cloaking constructions have on a traveling
waves.
2
The Helmholtz Equation
• The wave equation
2w
2

c
w
2
t
(1)
models the propagation of a wave travelling
through a given medium at a constant speed
c.
2
2
 2  2
x
y
3
• If we assume the solution w is separable,
then we can write
w( x, y, t )  u ( x, y )v(t )
Substituting this into (1) gives
 2v
u 2  c 2 vu
t
which can be rewritten as
1  v u

2
2
c v t
u
2
( 2)
4
• If we assume both sides are equal to the
constant  s 2 , then solving (1) reduces to
solving the two equations
  2v 2 2
 2 c s v 0
 t
2


u

s
u0

(3)
( 4)
•
• Equation (3) has solutions of the form
v(t )  c1 cos(t )  c2 sin( t )
• where   cs
5
Approximating Derivatives
• To apply the finite difference method, we will
need to estimate the derivatives of a function.
• Then from Taylor’s formula,
h 2 ''
h 3 ' ''
f ( x  h)  f ( x)  hf ( x) 
f ( x) 
f ( x)  
2!
3!
(5)
h 2 ''
h3 '''
f ( x  h)  f ( x)  hf ( x) 
f ( x) 
f ( x)  
2!
3!
( 6)
'
'
6
• Subtracting them and so we can approximate
the derivative of f using the formula
f ( x  h)  f ( x  h)
f ( x) 
2h
'
(7 )
• Furthermore , adding (5) and (6) gives the
approximate
f ( x  h)  2 f ( x )  f ( x  h)
f ' ' ( x) 
h2
(8)
7
In the similarly way, we can approximate the
partial derivatives of f as follows
u
u ( x  h, y )  u ( x  h, y )
( x, y ) 
x
2h
u
u ( x, y  k )  u ( x, y  k )
( x, y ) 
y
2k
 2u
u ( x  h, y )  2u ( x, y )  u ( x  h, y )
( x, y ) 
x 2
h2
 2u
u( x, y  k )  2u ( x, y)  u ( x, y  k )
(
x
,
y
)

y 2
k2
(9 )
(10)
(11)
(12)
8
The Finite Difference Method
• Let R  a, b c, d  be a rectangle in R 2 , and
consider the boundary value problem
2

u  s u  0
•
in R
(13)

•
u ( x, y)  f ( x, y) on R
• Then apply the finite difference method to
solve this problem.
9
k2
r 2
h
Letting
,this we can get the following
five-point formula,
ru ( x  h, y )  ru ( x  h, y )  u ( x, y  k )
 u( x, y  k )  (s k  2r  2)u( x, y)  0
2
2
10
Consider the boundary value problem
•
in R

 u  u  0x
u ( x, y )  sin( )
•
on R

6

• Using a 70 70 lattice resulted in the solution
pictured in Figure (1)
11
• Similarly, the solution to
 u  u  0
•
in R

u ( x, y )   1
•
on R

x y
• is pictured in Figure (2)
2
2
12
Transformations
• A divergence form operator acting on function
u  C 2 (R2 )
• is a differential operator L of the form
Lu  div ( Au )
r x
r y
F ( x, y )  ((1  ) , (1  ) )
2 r
2 r
• with r  x 2  y 2
13
Then for a differentiable function, define the matrix
•
 F1
 x ( x, y )
DF  
 F2 ( x, y )
 x
A( x, y) 
F1

( x, y ) 
y

F2
( x, y ) 

y
1
det DF ( x, y)
where
F ( x, y)  ( F1 ( x, y), F2 ( x, y))
DF ( x, y)DF ( x, y)
T
( x , y )  F 1 ( x , y )
14
A Generalization of the Helmholtz
Equation
• In this section, we are interested in a general
Helmholtz equation of the form
2
Lu  s u  0
• Given a 2 2 matrix A, we wish to apply the
finite difference method to solve a boundary
value problem
 Lu  s 2u  0
•
in R

u ( x, y)  f ( x, y)
•
on R
15
• where R  a, b c, d  as before, and
• For simplicity, let  u  u and  u  uy
x
• If we assume
1
• we can write
Lu  div ( Au )
2
 a11 a12 
A

a
a
 21 22 
 a11 a12   1u   a111u  a12 2u 
Au  





a
a

u
a

u

a

u
22 2 
 21 22   2   21 1
16
• Then we can express Lu as
2
2
Lu  div ( Au)    i (aij  j u)
(14)
i 1 j 1
• In the same manner as before from the Taylor
series, we can show that
1
1 ( 2u )( x, y ) 
( 2u ( x  h, y )   2u ( x  h, y ))
2h

1 u ( x  h, y  k )  u ( x  h, y  k ) u ( x  h, y  k )  u ( x  h, y  k )
(

)
2h
2k
2k

u ( x  h, y  k )  u ( x  h, y  k )  u ( x  h, y  k )  u ( x  h, y  k )
4hk
(15)
17
• Utilizing (9)-(12)along with(15) in (14) allows
us to approximate the equation Lu  s 2u  0
• with the nine-point formula
(
(
a12  a21
a
b
a  a21
)u ( x  h, y  k )  ( 222  2 )u ( x, y  k )  ( 12
)u ( x  h, y  k ) 
4hk
k
2k
4hk
(
a11 b1
2a11 2a22
a11 b1
2

)
u
(
x

h
,
y
)

(
s


)
u
(
x
,
y
)

(
 )u ( x  h, y ) 
2
2
2
2
h
2h
h
k
h
2h
a12  a21
a
b
a  a21
)u ( x  h, y  k )  ( 222  2 )u ( x, y  k )  ( 12
)u ( x  h, y  k )  0
4hk
k
2k
4hk
• with
2
b j    i aij
i 1
18
Applications to Cloaking
• We use a specific example to solve the
boundary value problem.
• Consider the transformation
r x
r y
F ( x, y )  ((1  ) , (1  ) )
2 r
2 r
• where
r  x2  y2
19
• Using the transformation as discussed in last
section, we can calculate to conclude
 2  r x2
 3

r
DF ( x, y )   2r
  xy

r3



2  r y2 
 3
2r
r 
 r
(2r  1) x 2


3
r

1
(
r

1
)
r
A( x, y )  
  (2r  1) xy
3

(
r

1
)
r


xy
r3
(2r  1) xy 

(r  1)r 3 
r
(2r  1) y 3 

r  1 (r  1)r 3 

20
• For example, let R be the square  3,3  3,3
• Consider  Lu  u  0
u ( x, y )  sin( x )

6

• Using the MATLAB program , the numerical
solution is shown in Figure (3)
21
• For the equation
 Lu  u  0

u ( x, y )   1

x2  y2
• The numerical solutions are Figure (4).
22
• Compare Figure (1) with Figure (3)

 u  u  0x
u ( x, y )  sin( )

6


 Lu  u  0x
u ( x, y )  sin( )

6

23
• Figure (2) and Figure (4)
 u  u  0

u ( x, y )   1

x2  y2
 Lu  u  0

u ( x, y )   1

x2  y2
24
Indistinguishable waves
•
Two waves are indistinguishable, if their
boundary values and normal derivatives are
identical.
30 30 40 40 50 50 60 60 70 70 80 80
Figure(1)and Figure(3) 0.377
0.214
0.116
0.062
0.034
0.018
Figure(2)and Figure(4) 0.161
0.044
0.018
0.009
0.005
0.003
25
• We set up the initial value

 2v
v 0

2
t

v(0)  1, v (0)  0

t
26
27
Thank you!
28