I have a set of points in the complex plane that the parameters of the real and imaginary values are integers $P=\{a+bi,\enspace a,b\in\mathbb{Z}\}$ so the set P makes up a grid. There is a second grid $V$ offsetted by $\frac{1+i}{2}$ and it is defined by $V=\{p+\frac{1+i}{2},\enspace p\in P\}$. An element of $P$ and $V$ are called $p$ and $v$ respectively, formally: $p\in P,\enspace v\in V$.

The point $v$ represents a cylinder which can be rotated clockwise or counterclockwise by $90^{\circ}$, we can map the rotations $\{\text{right},\text{up},\text{left},\text{down}\}$ to $\{0,1,2,3\}$. Every cylinder starts at rotation $0$.

A person starts in position $0$ or $0+0i$ and each step he moves up, down, left or right $1$ unit at a time. Whenever the person passes through two $v_1$ and $v_2$ next to each other, $v_1,v_2\in V$, the one on the left is rotated $90^{\circ}$ counterclockwise and the one on the right is rotated $90^{\circ}$ clockwise.


Is it possible to make a path so that the initial state is the same as the end state? If yes, what conditions should $T$ have?

My attempt:

There are infinitely many trivial solutions such as $T=(1,-1)$ or $T=(1,1,-1,-1)$ or $T=(i,i,i,-i,-i,-i)$ which doesn't count. The general form of these solutions is $T=(t_1,\dots, t_n,-t_1,\dots ,-t_n)$ or $T=(t_1,\dots, t_n,-t_n,\dots ,-t_1)$

The path $T=(t_1,\dots,t_n),\enspace t_i\in\{1,i,-1,-i\}$.

The position of the person after $n$ steps is given by $p_n=\sum\limits_{i=1}^{n}t_i$.

At any step $n$ the position of the cylinder in the left is given by $$V_L(n)=p_n+\frac{1+i}{2}t_{n+1}=\frac{1+i}{2}t_{n+1}+\sum_{i=1}^{n}t_i$$ and the one on the right by $$V_R(n)=p_n+\frac{1-i}{2}t_{n+1}=\frac{1-i}{2}t_{n+1}+\sum_{i=1}^{n}t_i$$

$$V_L:\mathbb{N_0}\rightarrow V ,\quad V_R:\mathbb{N_0}\rightarrow V$$ $C_L(v)$ and $C_R(v)$ tells us the number of times the path has passed from the left and right, respectively, in any given cylinder at position $v$. Formally, $$C_L:V\rightarrow\mathbb{N_0} ,\quad C_R:V\rightarrow\mathbb{N_0}$$ $$C_L(v)=|\{k\mid v=V_L(k),\enspace k=1,\dots,\vert T\vert-1\}|$$ $$C_R(v)=|\{k\mid v=V_R(k),\enspace k=1,\dots,\vert T\vert-1\}|$$

The total rotation of the cylinder, is given by the number of path segments in the left minus the ones on the right modulo $4$.

$$C(v)= (C_L(v)-C_R(v))\mod{4}$$ $$\text{or}\quad C= (C_L-C_R)\mod{4}$$

Now I have to prove

$$\Large\quad\exists T\neq \emptyset,\sum\limits_{v\in V}C=0$$


By experiment, any path $T=(t_1,\dots, t_n,t_1,\dots, t_n,t_1,\dots, t_n,t_1,\dots, t_n)$ is a valid solution for this problem, but I don't know how to get there or even if these are the only valid solutions. For this $T$,

$$\sum\limits_i t_i=0$$

which means that the path started in the same position as it has ended.

Does this mean that for the solutions to be valid the only way is to make a path that starts and ends at the same point and go through it $4$ times one after another?

$$4C\equiv 4(C_L-C_R)\mod{4}$$ $$4C\equiv 0\mod{4}$$ I think this is why passing through a closed path 4 times resets everything, but is it the only solution? Is it possible for a path which doesn't end in the start position?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.