S n \mathfrak{S}_n is amazing (part 1)

Let S n \mathfrak{S}_n denote the set of all permutations of the set [ n ] = { 1 , 2 , 3 n } [n]=\{1,2,3 \cdots n\} .

Any permutation w S n w\in \mathfrak{S}_n can be written in what is called cycle notation, described here . This can be made a unique representation by requiring that the maximum element of each cycle goes first and that cycles are ordered in increasing order by their largest elements. This is known as the standard representation of a permutation. As an example, in S 4 \mathfrak{S}_4 the permutation 2314 2314 has a standard representation of ( 312 ) ( 4 ) (312)(4) .

The map ϕ n : S n S n \phi_n: \mathfrak{S}_n \to \mathfrak{S}_n takes a permutation w w and maps it to its standard representation without the parentheses. For how many permutations w S 10 w\in \mathfrak{S}_{10} does w = ϕ 10 ( w ) w= \phi_{10}(w) ?


The answer is 89.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

1 solution

Dylan Pentland
Jun 14, 2016

Let w ^ \hat{w} denote ϕ n ( w ) \phi_{n}(w) . For this problem in S n \mathfrak{S}_n , the claim in that the number of permutations that are left unchanged by ϕ n \phi_n is F n + 1 F_{n+1} or the n+1st Fibonacci number. Since Fibonacci numbers obey a nice recurrence relation, an easy way to approach this problem is to simply show that this is obeyed using induction.

First, let w = ( C 1 ) ( C 2 ) ( C l ) w = (C_1)(C_2) \cdots (C_l) . Divide this into two cases: C l = 1 , C l 1 |C_l|=1, |C_l|\neq 1 . If the size of C l C_l is one, then since this is the last cycle it must be larger than all other elements of w w - that is, it must be n n . For this first case, we have an unchanged permutation every time ( C 1 ) ( C l 1 ) = w S n 1 (C_1) \cdots (C_{l-1})=w'\in \mathfrak{S}_{n-1} is fixed, and there are F n F_n ways to do this by the induction hypothesis.

For the other case, we have C l = ( n , k ) C_l = (n, \cdots k) . Note that w ^ ( n ) = w ( n ) = k \hat{w}(n)=w(n)=k , meaning that C l C_l is a 2-cycle of the form ( n , k ) (n,k) . But we also have w ( k ) w(k) is n n and since we know exactly where n n is we see k = n 1 k=n-1 . In the same way as the last case, we get F n 1 F_{n-1} ways.

Now simply verify the first few cases to see that the relationship is in fact true.


Alternative solution : Suppose that w = ( C 1 ) ( C 2 ) ( C l ) w=(C_1)(C_2)\cdots(C_l) has C i 2 |C_i|\le 2 . It is not difficult to see that the two conditions on the standard form force the every cycle to contain the actual indices it covers in w ^ \hat{w} : for example if C 1 = 2 |C_1|=2 then C 1 = ( 21 ) C_1=(21) . When we go from cycle notation to the permutation, no change occurs because 1-cycles simply map to themselves and so do 2-cycles. This means that this class of permutations remains unchanged by ϕ n \phi_n .

Suppose that there was a cycle C = ( w i , w j ) C= (w_i, \cdots w_j) over the indices i j i \cdots j in some w ^ = w \hat{w}=w of size greater than two. We know that w k w_k is at index w k 1 w_{k-1} (this is cyclic) from w ^ = w \hat{w}=w , and what this tells us is that since w i w_i is the largest w i + 1 w_{i+1} is at the largest index. This implies C 2 |C|\le 2 , a contradiction.

Using the above, the only viable class of permutation is those with 1 or 2-cycles. This is a well known problem, and has a solution of F n + 1 F_{n+1} .

First of all, this problem intrigued me; thanks for posting it.
I am having a little trouble with the solution though.
I didn't understand your proof for "the only viable class of permutation is those with 1 or 2-cycles." Why is (4123) not a viable cycle?
I think that the cyclic condition is not that w k w_k is at w k 1 w_{k-1} but rather w k w_k is at w k + 1 w_{k+1} except for k=i of course. I think the answer should consequentially be 2 n 1 2^{n-1} .
Consider the case of n=3.
(1)(2)(3), (1)(32), (21)(3) and (312) all seem to be acceptable. hence the answer seems to be 2^2=4 rather than the 4th fibonacci number. Am I missing something here? Thanks for your consideration.


Adit Mohan - 4 years, 12 months ago

Log in to reply

In the permutation 312 312 , we have w ( 1 ) = 3 , w ( 3 ) = 2 , w ( 2 ) = 1 w(1)=3, w(3)=2, w(2)=1 . This gives the cycle ( 321 ) (321) , which is different. ( 312 ) (312) on the other hand, gives w ( 3 ) = 1 , w ( 1 ) = 2 , w ( 2 ) = 3 w(3)=1, w(1)=2, w(2)=3 or 231 231 . I think what tripped you up was reading a cycle like ( 312 ) (312) as '3 is at index 1, 1 at index 2, 2 at index 3' when it's really '3 is at index 2, 1 at index 3, 2 at index 1'. It goes the opposite way: if w ( i ) = j w(i)=j then we have ( i j ) (\cdots ij \cdots) in cycle notation. That's why if we have ( w k 1 w k ) (\cdots w_{k-1}w_k\cdots) then w ( w k 1 ) = w k w(w_{k-1})=w_k .

Hopefully this cleared things up.

Dylan Pentland - 4 years, 12 months ago

Log in to reply

Thanks again. But then shouldn't the standard notation for the example in your question i.e. 2314 be (312)(4) instead of (321)(4)?

Adit Mohan - 4 years, 12 months ago

Log in to reply

@Adit Mohan Fixed that, my bad. I guess the conclusion is that it's really easy to mess up cycle notation :p

Dylan Pentland - 4 years, 12 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...