Download Document

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
S
M V
工程數學教學經驗談
陳正宗
海洋大學 特聘教授
河海工程學系
Oct. 1, 2004, NTU, 13:30~13:50
2004/10/1
1
NTOU, MSVLAB
Outlines
S
M V







Introduction
ODE
Gaussian elimination
Double Lapalce transform for Euler-Cauchy
ODE.
Poisson integral formula
SVD technique
Conclusions
2004/10/1
2
NTOU, MSVLAB
Introduction
S
M V




Students: Quantity (OK) Quality (?)
100% -> 30% -> 15% (Past)
26% -> 52% (Current)
Attitude
Interest (Tool and Method)
Five demonstrative examples
2004/10/1
3
NTOU, MSVLAB
ODE
S
M V
Given a two order differential equation
Q : y '' 2 y ' y  0
A: y  et , tet .
why t occurs ? (Wronskian, variation of parameters,
L’Hospital rule……)
Special case:
Q : y ''  0
A : y  e0t , te0t  y  1, t
2004/10/1
4
NTOU, MSVLAB
Gaussian elimination
S
M V
. . . .
. . . .
1
Solve linear algebraic equation
0   p  0 
 5 4 1
 4 6 4 1   q  1 

     
 1 4 6 4   r  0 


 0 1 4 5   s  0 
1
Matrix operation for Guassian elimination
 5 4 1 0   p 
 4 6 4 1   q 
 a    a T
aT 
 1 4 6 4   r 


 0 1 4 5   s 
0 
1 
 
 
0 
0 
. . . .
. . . .
8
7
5
14
7
6
NASTRAN: DMAP (Direct Matrix Abstract Programming)
Bathe: Substructure (Superelement, substructure, Guyan
reduction, congruent transformation)
2004/10/1
5
NTOU, MSVLAB
S
M V
Gaussian elimination (Cont.)
. . . .
1
. . . .
1
. . . .
5
14
. . . .
7
6
8
7
2004/10/1
6
NTOU, MSVLAB
S
M V
Double Lapalce transform for
Euler-Cauchy ODE
Eurler-Cauchy ODE
at 2 y '' (t )  bty ' (t )  cy (t )  0
a, b, c are constants, y is the function of t
Higher order Eurler-Cauchy ODE
ant n y ( n ) (t )  an 1t n 1 y ( n 1) (t ) 
 a1ty ' (t )  a0 y (t )  0
LL (Euler-Cauchy ODE) = origin Euler-Cauchy ODE
L : Laplace transform
FF ( f (t ))  2 f ( t )
HH ( f (t ))   f (t )
F and H are Fourier and Hilbert transforms, respectively.
2004/10/1
7
NTOU, MSVLAB
S
Poisson integral formula
M V
Methods
G. E.:  2 u ( x)  0, x  
B. C. : u  f ( )
Image concept
Free of image concept
 2u ( x )  0
a

R
u( x) |xB  f ( )

2
0
Null-field integral
equation method
R'
Image source
1
u(  , ) 
2
Searching the
image point
Reciprocal radii
method
Traditional method


a2   2
 2
 f ( )d
2
a



2
a

cos(



)


2004/10/1
Degenerate
kernel
Poisson integral formula
8
NTOU, MSVLAB
Searching the image point by using degenerate
kernels
S
M V
.x
Fundamental solution:
 1 
 I
U F ( x, s )  ln R  
( ) m cos[m(   )],


m 1 m R
U F ( x, s )  ln( r )  ln x  s  
 1 R
U FE ( x, s )  ln   
( ) m cos[m(   )],

m

m

1

x

ln x  s  ln   
1 R m
( ) cos[m(   )],   R
m 

1  m
( ) cos[m(    )],   R 
m R
m1
ln x  s   ln R   
m1
R



R
 R 
2
R

  R,
  R,
a
x
s
.
B
 s   a
s
2
R
2UG ( x; s, s)   ( x  s)   ( x  s), x  
UG ( x; s, s)  0, x  B
a2
R
U G ( x; s, s)  ln | x  s |  ln | x  s |  ln a  ln R
1  m
( ) cos[m(   )]}
s
m 1 m R

 {ln R  
{ln(

a2
1  R
)   ( 2 ) m cos[m(   )]}  ln a  ln R
R
a
m 1 m

R
1 
 R
 ln( )   [( ) m  ( 2 ) m ]cos[m(   )], 0    R.
a
R
a
m 1 m
2004/10/1
9
NTOU, MSVLAB
Free of image point - null-field integral equation in
conjunction with degenerate kernels
S
M V
U F ( x, s)  ln r ,
2 u( x) 
T
B
0

B
c
Green’s
identity
Fundamental
solution
I
F ( s, x) u ( s) dB( s) 
U
B

I
F ( s, x) t ( s) dB( s),
x 
s  ( R,  )
xB
TFE ( s, x) u ( s) dB( s)
 U FE ( s, x) t ( s) dB(s),

u (s)  f ( )  a0 

t (s)  p0 

 (a
n
x 
B
c
cos(n )  bn sin(n ))
a
x  (,  )  B
specified
n 1
( pn cos(n )  qn sin(n )). unknown
n 1
Degenerate kernel
Unknown
coefficients
2004/10/1

 



m
1

2
(
)
cos[
m
(



)]
a

(
a
cos(
n

)

b
sin(
n

))




0
n
n

 d
0  m1 a

n 1

 
1 2 


1

2
( ) m cos[m(   )] f ( ) d



m 1 a
2 0 

u(  , ) 
1
2
2
10
NTOU, MSVLAB
SVD technique
S
M V
A matrix Amn, m is the number of function, n is the unknown number.
We can get
Am  n xn 1  bm 1
SVD
Amn  mm mn  Tnn
 n
 

   0

 
 0
 0
,
  
 1

 
 0  m  n
mn
where  and  are unitary matrices
2004/10/1
11
NTOU, MSVLAB
S
M V
SVD for Continuum Mechanics
F  VR  RU
F : deformation gradient
dx = F dX
dX
X
x
dx
F  VR  RU
2004/10/1
12
NTOU, MSVLAB
Principal directions
S
M V
undeformed
stretching
rotation
deformed
F  RU
undeformed
stretching
rotation
deformed
F  VR
2004/10/1
13
NTOU, MSVLAB
Meaning of  and 
S
M V
 
Spurious system
Deformed system
Degenerate system
Fictitious system
2004/10/1
(Chen et. al. Royal Society, 2001)
True system
(Chen et. al. IJCNAA, 2002)
Undeformed system
(Chen et. al. IJNME, 2004)
Normal system
(Chen et. al. JCA, Rev., 2004)
14
True system
NTOU, MSVLAB
Conclusions
S
M V




Five examples were demonstrated for the
teaching of engineering mathematics.
Teaching and research merge may have the
opportunity to merge together.
How to teach eng. math. for current students
is a challenge to us.
Not only tools but also technique should be
considered to strengthen our teaching.
2004/10/1
15
NTOU, MSVLAB
S
M V
歡迎參觀海洋大學力學聲響振動實驗室
烘培雞及捎來伊妹兒
URL: http://ind.ntou.edu.tw/~msvlab/
Email: [email protected]
2004/10/1
16
NTOU, MSVLAB
Related documents