Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

What form should the minimal-length curve to $\int \sqrt{dr^2 +r^2d\theta^2\over{1-r^2}}$ take? I think I can use the Euler-Lagrange equations. So write the integral as $\int\sqrt{({dr\over d\theta})^2 +r^2\over{1-r^2}}d\theta$, then since the integrand is not explicitly dependent on $\theta$, we can say that $I-r'{\partial I\over \partial r'}=c$ for some (complex?) constant $c$. But I don't seem to be getting anything useful. The equation reduced to $r^2=c\sqrt{(r'^2+r^2)(1-r^2)}$ by my calculation. But I think I am expecting straight lines through the origin...

share|improve this question
    
what are the boundary conditions? or are you looking for closed curves? –  Willie Wong Feb 26 '12 at 20:14
    
@WillieWong: The domain is the unit disc. I am wishing to show that the shortest path joining 2 points is along a diameter. –  Don Feb 26 '12 at 20:41
    
So you are looking for a curve that passes through $(1,0)$ and $(0,1)$ that minimises the integral? Or do you allow arbitrary endpoints on the unit disc? (Because it looks like you are looking for geodesics on the Poincare disc.) –  Willie Wong Feb 26 '12 at 20:53
    
@WillieWong: You are right, I am indeed looking for geodesics on the disc, but the metric is one $1\over (1-r^2)$ factor different –  Don Feb 26 '12 at 21:03

1 Answer 1

up vote 0 down vote accepted

If you expect straight lines through the origin, it makes no sense to consider $r$ as a function of $\theta$, since your entire line is at a single $\theta$ value. (Note also that using the Beltrami identity instead of the Euler-Lagrange equation as you did introduces a spurious solution $r'=0$ that leads to circles that don't solve the Euler-Lagrange equation.)

You can consider $\theta$ as a function of $r$ instead. Then you have

$$\int\sqrt{\frac{1+r^2\theta'^2}{1-r^2}}\mathrm dr\;,$$

and since $\theta$ doesn't occur in the integrand, the Euler-Lagrange equation becomes

$$\frac{\partial\mathcal L}{\partial\theta'}=C\;,\\ r^2\theta'\sqrt{\frac{1-r^2}{1+r^2\theta'^2}}=C\;. $$

This equation is indeed solved by $\theta=\text{const.}$, so lines through the origin are solutions, but it also has further solutions:

$$ r^4\theta'^2(1-r^2)=C(1+r^2\theta'^2)\;,\\ \theta'=\sqrt{\frac C{r^4-r^6-Cr^2}}\;. $$

Wolfram|Alpha integrates this to a rather complicated function of $r$.

share|improve this answer
    
Thank you, joriki!! –  Don Feb 26 '12 at 21:36
    
@Don: You're welcome! –  joriki Feb 26 '12 at 21:43

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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