I was learning about Catalan numbers online. I have understood the combinatorial argument, recurrence relation and generating function based proofs of Dyck words and Dyck paths.
Let $S_{2n}$ be the symmetric group of permutations on integers $1$ through $2n$.
Consider the conjugacy class of the permutation $(1, 2)(3, 4)(5, 6)\ldots(2n-1, 2n)$ in $S_{2n}$. Let's denote it by $W_{2n}$. [N.B. Conjugacy preserves cycle structure, so what I mean is that $W_{2n}\ \left(\subset S_{2n}\right)$ is the set of all permutations which can be expressed as product of $n$ mutually disjoint transpositions.]
Check that: $$|W_{2n}|=\frac{1}{n!}\binom{2n}{2}\binom{2n-2}{2}\binom{2n-4}{2}\cdots \binom{2}{2}=\frac{(2n)!}{2^n\cdot n!}$$
Since disjoint transpositions commute, every ordering should be counted once. That accounts for the division by $n!$.
In bijective proof problems (problem 175, page 41), I came across the fact that the number of permutations $w\in W_{2n}$ such that the product $(1,2,\ldots,2n) \circ w$ has $(n+1)$ disjoint cycles is given by the $n$-th Catalan number.
How do I prove this?
For example, in the group $S_6$, there are $5$ permutations $w\in W_{6}$ (consisting of $3$ disjoint transpositions each) such that the product $(1,2,3,4,5,6)\circ w$ has $4$ disjoint cycles.
- $(1, 2, 3, 4, 5, 6)\circ(1, 2)(3, 4)(5, 6) = (1, 3, 5)(2)(4)(6)$
- $(1, 2, 3, 4, 5, 6)\circ(1, 2)(3, 6)(4, 5) = (1, 3)(2)(4, 6)(5)$
- $(1, 2, 3, 4, 5, 6)\circ (1, 4)(2, 3)(5, 6) = (1, 5)(2, 4)(3)(6)$
- $(1, 2, 3, 4, 5, 6)\circ (1, 6)(2, 3)(4, 5) = (1)(2, 4, 6)(3)(5)$
- $(1, 2, 3, 4, 5, 6)\circ (1, 6)(2, 5)(3, 4) = (1)(2, 6)(3, 5)(4)$
I have been trying to construct a bijective mapping of these permutations with Dyck words of length $2n$ but unable to make it. How do I progress with this approach? Is there a better alternate approach i.e., a clever combinatorial argument?
I am a beginner in abstract algebra, I understand order, compositions, etc. of the elements of symmetric group; I don't have much in depth knowledge on this topic. It is possible that I might be missing out on some obvious property of the permutations/cycles which can potentially help solve this problem.
Observe that the $5$ valid permutations share a property in common:
This hints us to regard each transposition as matched pairs and to consider similar combinatorial objects.
And you are right, one property of permutations/cycles/transposition really helps (me).
You should be familiar with this characterization of Catalan number:
And we can make a bijection between your problem and this characterization, as follows...
Example: $(4 \; {\color{red}{7}} \; 1 \; 5 \; {\color{blue}{2}} \; 6 \; 3) \cdot ({\color{blue}{2}} \; {\color{red}{7}}) = (1 \; 5 \; {\color{blue}{2}}) \cdot (4 \; {\color{red}{7}} \; 6 \; 3)$.
If we consider a cycle as a circle, this acts like cutting the circle into two pieces.
Above are my sketch of this approach. You may fill in the details by yourself.