# Riemann–Hilbert-type problem

Let $$P$$ be a fixed pentagon in the hyperbolic plane $$\mathbb H^2$$ with all the angles equal to $$\pi / 3$$. Let $$w_1, w_2, \dots, w_5$$ be the sides of $$P$$ going in the counterclockwise order. We are looking for five holomorphic functions $$f_1, f_2, \dots f_5 \colon \mathbb H^2 \to \mathbb C$$ such that $$\sum_{j=1}^5 f_j(z) \equiv 1$$ and if $$z \in w_j$$ for some $$j$$ then the following numbers are real

$$f_j(z),\quad \tau f_{j+1} (z),\quad \overline \tau f_{j-1} (z),\quad f_{j+1}(z) + f_{j-2}(z), \quad f_{j-1}(z) + f_{j+2}(z) \in \mathbb R,$$ where $$\tau = \exp(2\pi i / 3)$$. Is it true that the space of solutions $$(f_1, f_2, f_3, f_4, f_5)$$ has finite dimension? Is it true that there is at most one solution?

## This question had a bounty worth +50 reputation from Nikita Kalinin that ended 8 hours ago. Grace period ends in 15 hours

This question has not received enough attention.

• I am confused with your notation. You say $z\in\{ w_j\}$ and the following statements depend on $j$. Can we remove the curly braces? – Alexandre Eremenko Feb 8 at 21:20
• @AlexandreEremenko , thanks, I've corrected the question. I meant that the given condition holds for any $j$. The curly brackets were added by j.c. , I hope he doesn't mind I have removed them. – Misha Feb 9 at 21:54
• The function $g=f_1^3$ satisfies $g(Tz)=\overline{g(z)}$, where $T$ is a symmetry in $w_1,w_2$ or $w_5$. Thus the group generated by these three symmetries must be discrete, else $f_1\equiv 0$ which quickly yields a contradiction. This looks a rare event (say, if $w_2$ and $w_5$ intersect at a point belonging to $\mathbb{H}^2$ and the angle between them is not a rational multiple of $\pi$, the composition of symmetries w.r.t. them already generates a non-discrete group.) – Fedor Petrov Feb 14 at 18:46
• @FedorPetrov , 1) Those symmetries belong to the symmetry group of the corresponding tessellation of $\mathbb H^2$ wich seems to be discrete. <br><br> 2) Let us suppose that lines $w_1, w_2, w_5$ form a triangle $T$. If $T$ lies "below" $w_1$ then it has angles $(2\pi/3, 2\pi/3, \dots)$, so its area is negative. If $T$ lies "above" $w_1$ then it has angles $(\pi/3, \pi/3, \dots)$, so its area is at most $\pi/3$ and the pentagon $P$ of area $3\pi-5\pi/3 = \frac43\pi$ can not fit inside $T$. So I think the sad fact is that $w_2$, $w_5$ can not intersect each other. – Misha Feb 15 at 10:43
• @Misha is it discrete for any equiangular pentagon? Or for generic? Why? Well, they do not intersect of course indeed, but I guess the composition of two symmetries is in general an element of infinite order even when it is not a rotation. – Fedor Petrov Feb 15 at 10:46