Let $c:[0,1]\to\mathbb{R}^2\backslash\{\mathbf{0}\}$ be a closed path with winding number $k$. Let $\tilde{c}=\rho(t)c(t)$, where $\rho:[0,1]\to(0,\infty)$ is function satisfying $\rho(0)=\rho(1)$. Determine the winding number of $\tilde{c}$.

What I did is that: Let $\alpha_0 = \frac{-ydx+xdy}{x^2+y^2} = f_1dx+f_2dy$, and $\int_{\tilde{c}}\alpha_0 = \int_0^1 f_1(\rho(t)c_1(t))(\rho'(t)c_1(t)+\rho(t)c_1'(t))dt$. But I don't know how to keep going from here. Any advice?


Does not

$k_c = \dfrac{1}{2\pi i}\displaystyle \int_0^1 \dfrac{c'(t)}{c(t)} \; dt? \tag 1$

indeed it does; then

$k_{\bar c} = \dfrac{1}{2\pi i}\displaystyle \int_0^1 \dfrac{(\rho(t)c(t))'}{\rho(t) c(t)} \; dt = \dfrac{1}{2\pi i}\int_0^1 \dfrac{\rho'(t) c(t) + \rho(t)c'(t)}{\rho(t) c(t)} \; dt$ $= \dfrac{1}{2\pi i} \displaystyle \int_0^1 \dfrac{\rho'(t) c(t)}{\rho(t) c(t)} \; dt + \dfrac{1}{2\pi i} \int_0^1 \dfrac{\rho(t)c'(t)}{\rho(t)c(t)} \; dt$ $= \dfrac{1}{2\pi i} \displaystyle \int_0^1 \dfrac{\rho'(t)}{\rho(t)} \; dt + \dfrac{1}{2\pi i}\int_0^1 \dfrac{c'(t)}{c(t)} \; dt = \int_0^1 \dfrac{\rho'(t)}{\rho(t)} \; dt + k_c; \tag 2$

now since

$\rho:[0, 1] \to (0, \infty), \tag 3$

we have

$(\ln \rho)(t))' = \dfrac{\rho'(t)}{\rho(t)}, \tag 4$


$\displaystyle \int_0^1 \dfrac{\rho'(t)}{\rho(t)} \; dt = \int_0^1 (\ln \rho(t))' \; dt = \ln (\rho(1)) - \ln (\rho(0)) = 0 \tag 5$


$\rho(0) = \rho(1); \tag 6$

thus we are left with

$k_{\bar c} = k_c, \tag 7$


The point is, of course, that the winding number $k_c$ does not really depend on the radius $r(t) = \vert c(t) \vert$ of $c(t)$, as long as we have $r(t) \ne 0$ on $c(t)$.

Nota Bene: In fact, the above argument is just a re-write of the usual demonstration that the winding number of $c(t)$ makes any sense at all, since in polars we have

$c(t) = r(t)e^{i\theta(t)}, \tag 8$


$c'(t) = r'(t) e^{i\theta(t)} + r(t) i \theta'(t) e^{i\theta(t)}, \tag 9$

$\dfrac{c'(t)}{c(t)} = \dfrac{r'(t) e^{i\theta(t)} + r(t) i \theta'(t) e^{i\theta(t)}}{r(t)e^{i\theta(t)}} = \dfrac{r'(t)}{r(t)} + i\theta'(t), \tag{10}$

which yields

$\displaystyle \int_0^1 \dfrac{c'(t)}{c(t)} \; dt = \int_0^1 \dfrac{r'(t)}{r(t)} \; dt + i \int_0^1 \theta'(t) \; dt = \ln r(1) - \ln r(0) + i \Delta \theta = i \Delta \theta, \tag{11}$

where $\Delta \theta$ is the increment incurred by $\theta$ as we traverse the closed path $c(t)$; as such,

$\Delta \theta = 2\pi n, \; n \in \Bbb Z, \tag{12}$

so the winding number is indeed a sensible construct. End of Note.

  • 1
    $\begingroup$ Thanks. But I’m not in the complex plane. So is it the same in R^2? $\endgroup$ – QD666 Feb 17 at 20:25
  • $\begingroup$ @QD666: whoops! my bad; that's what I get for reading/answering questions before coffee! Hang tight, will edit. Cheers! $\endgroup$ – Robert Lewis Feb 17 at 20:38
  • $\begingroup$ @QD666: sorry this is taking so damn long; I actually had to break from MSE for awhile and do my real life; still at it, however . . . $\endgroup$ – Robert Lewis Feb 18 at 3:30
  • 1
    $\begingroup$ No worries, it’s fine $\endgroup$ – QD666 Feb 18 at 3:39

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.