(5 intermediate revisions by 2 users not shown)
Line 18: Line 18:
  
 
<math> 0 = \int\nabla u\ ds  </math>
 
<math> 0 = \int\nabla u\ ds  </math>
 +
 +
If we have a domain, <math>\Omega</math> and <math>\gamma</math> a curve in <math>\Omega</math>, where <math>A, B </math> are end points of <math>\gamma</math>, from vector calculus, we have
 +
 +
<math> u(B) - u(A) = \int_\gamma \nabla u \cdot ds </math>.
 +
 +
but the integral is zero since <math> \nabla u </math> is zero. Hence <math> u(A) = u(B) </math> for any arbitrary point <math>A</math> and <math>B</math> and <math>u</math> is constant in <math>\Omega</math>.
 +
 +
----
 +
Practice Problems: Suggested Solutions/Thoughts
 +
 +
[http://www.math.purdue.edu/~bell/MA425/prac1a.pdf| Link to the practice Problems]
 +
 +
==Problem 1.==
 +
 +
'''Daniel''': I interpret "an analytic function f has constant modulus on a domain" as a function f that maps all domain to some f(z) where |f(z)| = r. Then f is probably not constant, since this is a circle. So I am interpreting the problem wrong?
 +
 +
'''Hooram''': This is my attempt. It may or may not be a correct proof.
 +
 +
The problem is to show that if f is analytic and |f| is constant on <math>\Omega</math>, then f is constant on <math>\Omega</math>.
 +
First, write f=u+iv. Then <math>|f| = u^2+v^2 = c</math>, some constant. Now, we take the derivative of the modulus of f:
 +
 +
<math>\dfrac{\partial}{\partial x} |f| = 2u u_x + 2v u_x = 0</math>
 +
 +
<math>\dfrac{\partial}{\partial y} |f| = 2u u_y + 2v u_y = 0.</math>
 +
 +
We know the derivatives <math>u_x, u_y, v_x, v_y</math> exist because f is assumed to be analytic.
 +
 +
So we have
 +
 +
<math> u u_x + v u_x = 0</math>
 +
 +
<math> u u_y + v u_y = 0.</math>
 +
 +
Since f is analytic, by Cauchy-Riemann equations, we have <math>u_x = v_y </math> and <math>u_y = -v_x </math>.
 +
 +
Before we carry on, let's think about what we're gonna do. We're gonna do two things:
 +
 +
1) eliminate <math>u_y</math> and  conclude <math>u_x = 0</math> 
 +
 +
<math>u_y = \dfrac{-vu_x}{u}</math>
 +
 +
<math>uu_x + \dfrac{-v^2u_x}{u} = 0 </math>
 +
 +
<math>u_x (u^2+v^2) = u_x c = 0 </math>
 +
 +
<math> u_x = 0 </math>
 +
 +
2) Similarly eliminate <math>u_x</math> and conclude <math>u_y = 0 </math>
 +
 +
Then, we will have <math>u_x = u_y = 0</math>, which implies that <math>f'(z) = u_x + iu_y = 0</math>.
 +
 +
Hence f is constant. This completes the proof.
 +
 +
'''Daniel''': Thanks Hooram, although I am still unsure whether |f| = <math>u^2 + y^2 </math> is valid? Also isn't it
 +
 +
<math>\dfrac{\partial}{\partial x} |f| = \dfrac{\partial u^2}{\partial x} + \dfrac{\partial v^2}{\partial x} = 2uu_x + 2vv_x </math>
 +
 +
 +
 +
==Problem 2.==
 +
 +
 +
'''Daniel''': Both a and b are simple - I would simply invoke the Fund. Theorem of Calc. Problem 2 could have been tricky if it involved going over the branch cut, but since it did not, I think we are safe. (check out pg. 175)
 +
 +
 +
==Problem 3.==
 +
'''Daniel''': I saw a similar problem on pg. 164. Shouldn't be too tricky since t is a real variable. Remembering what actually is <math>e^{3i\pi} </math> may be the hard part.
 +
 +
 +
==Problem 4.==
 +
'''Daniel''': Will return to it.
 +
 +
 +
==Problem 5.==
 +
'''Daniel''': hmm
 +
 +
----

Latest revision as of 07:02, 2 October 2012

Practice material for Exam 1 collaboration space

You can easily talk about math here, like this:

$ e^{i\theta} = \cos \theta + i \sin \theta. $

Is this the Cauchy Integral Formula?

$ f(a)=\frac{1}{2\pi i}\int_\gamma \frac{f(z)}{z-a} \ dz $


This isn't directly related to the practice exam, but is concerning a fact discussed in class. In one of the first lessons an important fact was provided. Namely, Suppose u is continuously a differentiable function on a connected open set $ \Omega $ and that $ \nabla u \equiv 0 $ Then u must be constant on $ \omega $.

How/Why is

$ 0 = \int\nabla u\ ds $

If we have a domain, $ \Omega $ and $ \gamma $ a curve in $ \Omega $, where $ A, B $ are end points of $ \gamma $, from vector calculus, we have

$ u(B) - u(A) = \int_\gamma \nabla u \cdot ds $.

but the integral is zero since $ \nabla u $ is zero. Hence $ u(A) = u(B) $ for any arbitrary point $ A $ and $ B $ and $ u $ is constant in $ \Omega $.


Practice Problems: Suggested Solutions/Thoughts

Link to the practice Problems

Problem 1.

Daniel: I interpret "an analytic function f has constant modulus on a domain" as a function f that maps all domain to some f(z) where |f(z)| = r. Then f is probably not constant, since this is a circle. So I am interpreting the problem wrong?

Hooram: This is my attempt. It may or may not be a correct proof.

The problem is to show that if f is analytic and |f| is constant on $ \Omega $, then f is constant on $ \Omega $. First, write f=u+iv. Then $ |f| = u^2+v^2 = c $, some constant. Now, we take the derivative of the modulus of f:

$ \dfrac{\partial}{\partial x} |f| = 2u u_x + 2v u_x = 0 $

$ \dfrac{\partial}{\partial y} |f| = 2u u_y + 2v u_y = 0. $

We know the derivatives $ u_x, u_y, v_x, v_y $ exist because f is assumed to be analytic.

So we have

$ u u_x + v u_x = 0 $

$ u u_y + v u_y = 0. $

Since f is analytic, by Cauchy-Riemann equations, we have $ u_x = v_y $ and $ u_y = -v_x $.

Before we carry on, let's think about what we're gonna do. We're gonna do two things:

1) eliminate $ u_y $ and conclude $ u_x = 0 $

$ u_y = \dfrac{-vu_x}{u} $

$ uu_x + \dfrac{-v^2u_x}{u} = 0 $

$ u_x (u^2+v^2) = u_x c = 0 $

$ u_x = 0 $

2) Similarly eliminate $ u_x $ and conclude $ u_y = 0 $

Then, we will have $ u_x = u_y = 0 $, which implies that $ f'(z) = u_x + iu_y = 0 $.

Hence f is constant. This completes the proof.

Daniel: Thanks Hooram, although I am still unsure whether |f| = $ u^2 + y^2 $ is valid? Also isn't it

$ \dfrac{\partial}{\partial x} |f| = \dfrac{\partial u^2}{\partial x} + \dfrac{\partial v^2}{\partial x} = 2uu_x + 2vv_x $


Problem 2.

Daniel: Both a and b are simple - I would simply invoke the Fund. Theorem of Calc. Problem 2 could have been tricky if it involved going over the branch cut, but since it did not, I think we are safe. (check out pg. 175)


Problem 3.

Daniel: I saw a similar problem on pg. 164. Shouldn't be too tricky since t is a real variable. Remembering what actually is $ e^{3i\pi} $ may be the hard part.


Problem 4.

Daniel: Will return to it.


Problem 5.

Daniel: hmm


Alumni Liaison

Correspondence Chess Grandmaster and Purdue Alumni

Prof. Dan Fleetwood