HOME

HW 11 Mathematics 503, Mathematical Modeling, CSUF , July 20, 2007

Nasser M. Abbasi

June 15, 2014

Contents

1 Problem 3 page 257 section 4.4 (Green Functions)
2 Problem 8, page 258 section 4.5

1 Problem 3 page 257 section 4.4 (Green Functions)

problem:

Consider boundary value problem u′′− 2xu′ = f (x)  ,0< x < 1  ,  u(0) = u′(1) = 0  . Find Green function or explain where there isn't one.

answer:

We see that p(x)= − 1

First, lets see if λ = 0  or not. Since if λ = 0  since by theorem 4.19 (page 248) Green function does not exist, and I do not need to try to find it.

Let

u′′− 2xu′ = λ u

If λ = 0   then solve the homogeneous equation  ′′     ′
u − 2xu = 0  . Let        ′
y(x)= u (x)  , hence we obtain  ′
y − 2xy= 0  , by separation of variables, we then have

pict

Hence

lny= x2+ C

Which leads to y(x)= Aex2   . But since y = u′ , then du= Aex2
dx   or

        ∫ x
u (x) = A   et2dt+ B
         0

Therefore,

        ∫ x
u1(x) = A   et2dt
         0

and

u2(x)= B

At x = 0  we have u(0)= 0  , hence         ∫   2
u (0)= A 00etdt+ B or 0 = B so now         ∫  2
u(x)= A  x0 etdt . Now lets see if this satisfies the second boundary condition u′(1)=  0  . First note that

  (  ∫      )
d--    x t2        x2
dx  A 0 e dt  = Ae

hence   at x= 1  we obtain 0 = Aexp (1 )  which means A= 0  , but this means trivial solution since both A,B are zero. Hence λ ⁄= 0  OK, so now I try to find Green function:

Now we need to find 2 independent solutions as combinations of   ∫x t2
A  0 e dt and B such that each will satisfies at least one of the boundary conditions.

We need u(0)=  0  , hence if we take

        ∫ x
u1(x) =   et2dt
         0

which will be zero at x= 0  , and if we take

u2(x)= 1

then we see that u′2(1)= 0. Now find the Wronskian

       ⌊      ⌋      ⌊         ⌋
       |      |      |∫x t2    |
W = det|u1  u2| = det| 0 e dt 1| = − ex2
       |⌈      |⌉      |⌈    2    |⌉
        u′1  u′2          ex    0

Hence using equation 4.46 we obtain, noting that p = − 1

pict

Hence

|--------------------------------------------------|
|              2(        ∫   2            ∫  2  )  |
| g(x,ξ )= − e−ξ  H (ξ − x) x0 etdt+ H (x− ξ )0ξetdt |
----------------------------------------------------

and

       ∫
        x
u(x)=   0 g (x,ξ )f(ξ)dξ

I used the Green function I derived, and used it to plot the solution (for f (x) = 1  ) and compare the plot with the analytical solution.

PIC

This is a plot of just the impulse response (green function) due to an impulse at x = 0.5

PIC

This is another method to solving this problem by using properties of Green function

From above we found u  = ∫xet2dt
 1   0 , u = 1
 2  , but

pict

and

pict

At x = ξ , due to continuity, we require that  

     ∫ ξ
A (ξ )   et2dt = B (ξ)
      0
(1)

and to impose the discontinuity condition on the first derivative we have

pict

From (1) we then obtain that

          ∫
       −-1  ξ t2
B(ξ) = eξ2 0 e dt

Hence

pict

and

pict

Hence

|------------------------------------------------|
|            (                                )  |
| g(x,ξ)= -−12 H (ξ − x)∫ xet2dt+ H (x − ξ)∫ξet2dt   |
|         eξ           0                0        |
--------------------------------------------------

Compare this solution to the one found above using the formula method we see they are the same.

2 Problem 8, page 258 section 4.5

Problem:

Find the inverse of the differential operator          2 ′′
Lu = − (x u) on 1<  x< e subject to u(1) = u(e)= 0

solution:

This is SLP problem with p = x2,q= 0  . First find if λ = 0  is possible eigenvalue.

       (   )′
λ u= −  x2u′

Let λ = 0  , hence we have − (x2u′)′=  0  or − (2xu′+ x2u′′)= 0  or

u′′+ 2-u′ = 0
    x

Use separation of variables. First let y = u′ , hence y′+ 2xy=  0  or 1yddyx = − 2x  hence

pict

But y = u′ , hence du= A 1x2dx or      ∫
u= A   1x2dx

hence u= − A1+ B
      x or

|--------------|
|       A      |
| u(x)=  x + B |
----------------

where the minus sign is absorbed into A . Hence we have 2 independent solutions A
x  and B , so we need combination of these 2 solutions to satisfy the BV. At x= 1  we have u = 0  , hence if we take u =  1− 1
 1   x  then it will satisfy this condition. At x = e we need u= 0  , hence take      1
u2 = x − exp(− 1)

Then

       ⌊      ⌋      ⌊                 ⌋
       |u   u |      |1− 1   1− exp(− 1)|
W = det|| 1   2|| = det||x      x         || = − ex2
       ⌈      ⌉      ⌈                 ⌉
        u′1  u′2         −1x2        −x12

Hence

     1− e−1
W =  --x2--

Then green function is

pict

But the inverse L−1   is ∫
  g(x,ξ )f(x)dx where g(x,ξ)  is the green function given aboive.

Another way to solve the problem:

From above we found u1 = 1x − 1  , u2 = 1x − e−1   , but

pict

and

pict

At x = ξ , due to continuity, we require that  

     (     )         (       )
A (ξ)  1− 1  =  B(ξ)   1− e−1
       ξ               ξ
(1)

and to impose the discontinuity condition on the first derivative we have

pict

Solve (1) and (2) for B(ξ ),A (ξ)

From (2) we haveB(ξ) = 1+ A (ξ )  , substitute into (1) we have A (ξ )(1 − 1 )= (1+ A (ξ)) ( 1− e−1)
      ξ                   ξ or

pict

Hence

pict

Then

pict
pict

Hence

------------------------------------------------------------------------
|             (        (        )                 (        )         )  |
|          -1-           -1−ξe−1-  (1   )            --1−ξ-- (1    −1)    |
| g (x,ξ )= eξ2  H (ξ − x) ξ(e−1−1)   x − 1 + H (x− ξ) ξ(e−1−1)    x − e     |
------------------------------------------------------------------------|

Which agree with the formula method.

This a plot of Green function for ξ = 2

PIC