Spring 2009 - Problem 11

Jordan curve theorem, winding number

Suppose that f ⁣:DC\func{f}{\D}{\C} is holomorphic and injective in some annulus {zr<z<1}\set{z \mid r < \abs{z} < 1}. Show that ff is injective in D\D.

Solution.

Suppose otherwise, and that there exist z1,z2Dz_1, z_2 \in \D such that f(z1)=f(z2)=wf\p{z_1} = f\p{z_2} = w. Picking R(0,1)R \in \p{0, 1} large enough, we have z1,z2B(0,R)z_1, z_2 \in B\p{0, R}. Thus, if we set γ=B(0,R)\gamma = \partial B\p{0, R} oriented counter-clockwise with domain [0,2π]\br{0, 2\pi}, the argument principle tells us that

Wfγ(w)2.W_{f \circ \gamma}\p{w} \geq 2.

But this implies that fγf \circ \gamma is not simple. Indeed, by the Jordan curve theorem, a simple closed curve has winding number in {1,0,1}\set{-1, 0, 1}, depending on the orientation. Thus, there exist 0t1<t2<2π0 \leq t_1 < t_2 < 2\pi such that f(γ(t1))=f(γ(t2))f\p{\gamma\p{t_1}} = f\p{\gamma\p{t_2}}. Since γ\gamma is in the annulus where ff is injective, we have γ(t1)=γ(t2)\gamma\p{t_1} = \gamma\p{t_2}. This can only happen if γ(t1)=γ(t2)=γ(0)\gamma\p{t_1} = \gamma\p{t_2} = \gamma\p{0}, which is impossible since 0<t2<2π0 < t_2 < 2\pi. Thus, ff must have been injective in D\D to begin with, as required.