For a domain $\Omega \subset \mathbb{C}$ and a $\rho: \Omega \rightarrow \mathbb{R}_0^+$, define the distance by: $$d_\rho (z_1,z_2) = \inf\limits_{\gamma \in C^1 (z_1,z_2)} \int_0^1 \rho(\gamma(t))\cdot|\dot{\gamma}(t)| \,dt,$$ where $C^1(z_1,z_2)$ is the space of smooth paths $\gamma: [0,1] \rightarrow \Omega$ such that $\gamma(0)=z_1$ and $\gamma(1)=z_2$.

I need to show that this satisfies $d_\rho (z_1,z_2) \leq d_\rho (z_1,w) + d_\rho (w,z_2)$.

I'm not very confident in what I did, which was to choose two paths $\gamma_1 \in C^1(z_0,w)$ and $\gamma_2 \in C^1(w,z_1)$, with $\dot{\gamma}_1 (1) = \dot{\gamma}_2(0)$ and $$\int_0^1 \rho(\gamma_1(t))\cdot|\dot{\gamma}_1(t)| \,dt = d_\rho (z_1,w) + \varepsilon/2,\\ \int_0^1 \rho(\gamma_2(t))\cdot|\dot{\gamma}_2(t)| \,dt = d_\rho (w,z_2) + \varepsilon/2$$ for some arbitrary $\varepsilon>0$.

Then a path $\gamma\in C^1(z_1,z_2)$ which is a concatenation of $\gamma_1$ and $\gamma_2$ has length $$\int_0^1 \rho(\gamma(t))\cdot|\dot{\gamma}(t)| \,dt \leq d_\rho(z_1,w) + d_\rho(w,z_2) + \varepsilon.$$ Thence $$d_\rho (z_1,z_2) \leq d_\rho(z_1,w) + d_\rho(w,z_2) + \varepsilon,$$ and since $\varepsilon$ is arbitrary, $$d_\rho (z_1,z_2) \leq d_\rho(z_1,w) + d_\rho(w,z_2).$$


(a) Minor issue: it's better to write $\dots < d_\rho(z_1,w)+\varepsilon/2$ (instead of $=$) because this is what the definition of infimum tells you.

(b) Major issue: How do you know you can arrange $\dot \gamma_1(1)=\dot \gamma_2(0)$? Not clear at all. One way to fix this is to take $\gamma_1$ and $\gamma_2$ so that the inequalities of the form $\int \dots < d_\rho(z_1,w)+\varepsilon/3$ hold, and then insert a connecting path in between. This extra path - call it $\gamma_{3/2}$ -- will begin and end at $w$ and will satisfy $\dot\gamma_{3/2}(0)=\dot \gamma_1(1)$ and $\dot\gamma_{3/2}(1)=\dot \gamma_2(0)$. It should also be smooth and very short -- so short that the integral over it will be $<\epsilon/3$.

One way to cook up this connector it to modify the familiar "polar flower" $r=\sin n\theta$ from calculus.


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.