I've been working through Marsden's Complex Analysis book, and I've come to a question where I'm not quite sure if I've got it. I would love some help.

question: Let f be analytic on a region, A, and let g be a closed curve in A. For any $z_0$ in A not on g, show that:

$\int_{g}\frac{f^{'}(a)}{{a - z_0}}da$ = $\int_{g}\frac{f(a)}{{(a - z_0)}^2}da$

and the followup question: how can you generalize this result?

my answer:

I thought I could use Cauchy's integral formula for derivatives, with the fact that given the function is analytic, that all of the derivatives of f exist on A and all are analytic.

So using the formula

$f^{k}(z_0) * I(g, z_0)$ = $\frac{k!}{2\pi i}$ $\int_g$$\frac{f(a)}{(a-z)^{k+1}}da;$ k = 1, 2, 3, ...

I just substituted in $f$ = $f^{'}$ to get the left hand side, since $f^{'}$ is analytic as well, and used k = 0; then I used $f = f$ and k = 1 to get the right hand side. This makes the equality hold.

I was not sure exactly how to generalize this, as I was kind of just figuring out a way to make the equality hold. This makes me think I'm doing it wrong.

many thanks.

  • $\begingroup$ A generalization can be made with the same two steps but using the $i+1$ derivative and $k=0$ in the first step and using $k=i+1$ in the second. $\endgroup$ – Rafa Budría Mar 1 '17 at 9:42
  • $\begingroup$ clever, thank you very much. does the rest look correct? In particular, cauchy's integral formula works since f is analytic on A. is that enough to show that f's derivatives are all analytic as well? or do we need something else, like f is homotopic to a point in A? My book uses that language $\endgroup$ – mac5 Mar 1 '17 at 18:30
  • $\begingroup$ edit: here i means imaginary number. is that what you meant as well in your comment? or did you mean an index? $\endgroup$ – mac5 Mar 1 '17 at 18:47
  • $\begingroup$ Analytic in $A$ is enough $\endgroup$ – Rafa Budría Mar 1 '17 at 18:49
  • $\begingroup$ also, using the formula you said above, wouldnt it be off by a k! ? since the left side has k = 0 but the right side has k = i + 1 $\endgroup$ – mac5 Mar 1 '17 at 18:52

Let's re-start. The formula is about derivatives. You didn't write it correctly (but your idea is good) The lhs has the k-th derivative

$$f^{(k)}(z_0) * I(g, z_0)=\frac{k!}{2\pi i}\oint_g\frac{f(a)}{(a-z)^{k+1}}da\tag 1$$

being $f^{(0)}(z_0)=f(z_0)$. Now, define $h(z)=f'(z)$, then:

$$h(z_0) * I(g, z_0)=\frac{1}{2\pi i}\oint_g\frac{h(a)}{a-z}da\tag 2$$

$$f'(z_0) * I(g, z_0)=\frac{1}{2\pi i}\oint_g\frac{f'(a)}{a-z}da$$

Taking $(1)$ With $k=1$:

$$f'(z_0) * I(g, z_0)=\frac{1}{2\pi i}\oint_g\frac{f(a)}{(a-z)^2}da$$

And you are done:


The generalization. Let be $h(z)=f^{(j)}(z)$ and using $(2)$

$$f^{(j)}(z_0) * I(g, z_0)=\frac{1}{2\pi i}\oint_g\frac{f^{(j)}(a)}{a-z}da$$

Now, by $(1)$ with $k=j$

$$f^{(j)}(z_0) * I(g, z_0)=\frac{j!}{2\pi i}\oint_g\frac{f(a)}{(a-z)^{j+1}}da$$

Leading to:



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.