# Interchanging between real part of cmplx exp and trig functions

1. Mar 20, 2014

### N00813

1. The problem statement, all variables and given/known data

Given u(x,t) = sum( e^(-at/2)*cos(n*pi*x/2L) * Re[A_n*e^(i*w_n*t)+B_n*e^(-i*w_n*t)], and the boundary conditions u(-L)=u(L)=0 for all t;
du/dt = 0 for all x at t = 0;
u(x,t=0) = e^(-|x|/l)

Find A_n and B_n.

2. Relevant equations
N/A

3. The attempt at a solution

I have attempted to turn the Real part into coefficients of cos and sin, i.e.:
Re[A_n*e^(i*w_n*t)+B_n*e^(-i*w_n*t)] = C_n cos(w_n*t) + D_n sin (w_n*t)
then taking advantage of cos orthogonality in x to get C and D.

I can't think of how to turn C and D into A and B. So far I figured out that C is the real part of A+B, but I can't figure out how to get the imaginary parts of A and B from C and D.

I'm sure it's simple, but I just can't seem to get it.

Any help is appreciated.

Last edited: Mar 21, 2014
2. Mar 20, 2014

### Dick

I don't think you need orthogonality or anything special. Just use Euler's formula. e^(i*w_n*t)=cos(w_n*t)+i*sin(w_n*t) and e^(-i*w_n*t)=cos(w_n*t)-i*sin(w_n*t). It's just D is the real part of i*A-i*B, isn't it?

3. Mar 21, 2014

### N00813

Yes, it is.
Maybe I've gone down the wrong way.

The question tells me to find A and B.

Using the non-zero initial condition given, I can find Re [A_n + B_n]. Using the zero initial condition (du/dt = 0 @ t = 0) I can find (a/2)*Re[A_n + B_n] = Re [iw_n (A_n - B_n)].

But I can't think of how to turn that into A_n and B_n by themselves.

4. Mar 21, 2014

### N00813

If C = Re[A+B]
and D = Re[i(A-B)]

Then is D/i = Re[A-B]

and C + D/i Re [2A]?

This way I can find the real parts of A and B.

For the imaginary, I was thinking iC = Im [ -B -A] , D = Im [ -A + B].

Is this correct?

5. Mar 21, 2014

### abitslow

You lost me as soon as you failed to close the parentheses of sum(. Then I lost interest when you failed to show ANY indication of what is being summed over.

Last edited by a moderator: Mar 21, 2014
6. Mar 21, 2014

### Dick

You are thinking about it too hard. You can't really 'solve' for A and B given C and D. There are lots of choices that work. Like A=C/2+iD/2 and B=C/2-iD/2. Or A=(C/2+1)+iD/2 and B=(C/2-1)+iD/2. And many others.

7. Mar 21, 2014

### N00813

OK.

Let's say I start again with the question. I find the expression for Re[A + B] and the expression for Re[iw_n*A-iw_n*B]. From this, you're saying it's impossible to get A and B uniquely?

8. Mar 21, 2014

### Dick

Of course it is. I already gave you two choices that work. Taking the real part of something cancels some components. Your right side has two real parameters. The left side has two complex parameters which is the same as four real parameters. You can't determine them all. Write out A=Re(A)+i*Im(A) and B=Re(B)+iIm(b) and see how many you can determine.

9. Mar 22, 2014

### N00813

Thanks. I suppose the question itself has holes in it.