2

Let $P(a;b)$ be the set of all sequences generated by permutation of $a$ zeroes and $b$ ones. $$|P(a;b)|=\binom{a+b}{a}$$

I want to prove that if $gcd(a,b)=1$, every sequence in $P(a;b)$ will have $a+b$ unique circular shifts.

For the sake of brevity, I am defining an indexing set $\Lambda = \{ 0,1,\ldots,a+b-1 \}$ to order the elements of the sequences.

What is a circular shift? To apply circular shift on a sequence, remove the first element of the sequence and append it to the end. E.g. $a_1,a_2,\ldots, a_n$ becomes $a_2,a_3,\ldots, a_n, a_1$. Repeat this process iteratively. The sequences, hence obtained, are circular shifts of the original sequence.

Suppose we have a sequence, $\alpha=\left(\alpha_j : j\in \Lambda\right)$, its circular shifts are the sequences $\alpha^{(i)}=\left(\alpha_{i+j}: j\in \Lambda\right)$ taking subscripts modulo $a+b$.

After $a+b$ iterations of circular shifts, we get back $\alpha$ again i.e., $\alpha^{(a+b)}= \alpha$.

Applying circular shifts iteratively is similar to generating rotational symmetries of a regular polygon by rotating it at certain angle, (say, $90^\circ$ for squares).

Considering $\alpha$ as the correspondence to the identity element $e$ and applying circular shift $r$ iteratively, we obtain a group $\langle r\rangle$ where $r^{a+b}=e$.

From group theory, we know that order of $\langle r\rangle$ is either $a+b$ or some divisor of $a+b$.

Let $d$ be order of $\langle r\rangle$ where $a+b=d\times k$ for some $k\in \mathbb Z^+$.

Then, $r^d=e$ i.e., $\alpha^{(d)}=\alpha$. This is possible only if corresponding elements of these sequences are same.

$\alpha_0=\alpha_d=\alpha_{2d}=\ldots=\alpha_{(k-1)d}$

$\alpha_1=\alpha_{d+1}=\alpha_{2d+1}=\ldots=\alpha_{(k-1)d+1}$

and so on till

$\alpha_{d-1}=\alpha_{2d-1}=\alpha_{3d-1}=\ldots=\alpha_{kd-1}$.

We have partitioned the set $\Lambda$ into $d$ disjoint equivalence classes (each of size $k$) based on the relation $i\sim j$ iff $\alpha_i=\alpha_j$.

Any digit (be it zero or one) which occurs in $\alpha$ must do so in multiples of $k$. Thus, we need both $a$ and $b$ to be multiples of $k$. However, $gcd(a,b)=1$ so, the only possibility is $k=1$ and hence, $d=a+b$.

Thus, order of $\langle r\rangle$ is $d=a+b$ and hence, each of these $a+b$ circular shifts generated from $\alpha$ are unique. $\blacksquare$

Is my proof correct? Is there anything I need to clarify? I would love to see any alternate elementary proofs or improvement to this one.

I tried to look up this topic in this post, but couldn't understand it because I don't know anything about group actions yet.

  • 2
    There are language cleanups you could make, but the logic is correct. I will say that Servaes’ answer in the linked post might be readable with your background. It doesn’t quite spell out all the details, though, so you’ll need to make essentially this extension anyway. – Eric Nathan Stucky Sep 25 '23 at 12:12

1 Answers1

1

Yes, your proof is correct.

(I am writing this CW answer to help move this question out of the Unanswered queue; you can help by accepting or upvoting to +1.)