Give a combinatorial argument to prove that the number of derangements satisfies the following relation:
$$d_n = (n − 1)(d_{n−1} + d_{n−2})$$
for $n \geq 2$.
I am able to prove this algebraically but not able to see the combinatorial example.
I am able to prove this algebraically but not able to see the combinatorial example. |
|||||
|
|
Suppose that $f:[n]\to[n]$ is a bijection with no fixed points. Then $f(n)\in\lbrace 1,\ldots,n-1\rbrace$. All these cases are the same up to relabelling, so suppose $f(n)=n-1$. Now define $g:[n-1]\to[n-1]$ by $g(x)=f(x)$, unless $f(x)=n$ in which case define $g(x)=n-1$. If $g$ has no fixed points, fine. Otherwise, since $f$ has no fixed points, it must be that $g(n-1)=n-1$, i.e., $f(n-1)=n$. Thus $f$ just swaps $n$ and $n-1$ and $f|_{[n-2]}$ must have no fixed points. Now just check that each of the terms in your formula are accounted for and nothing is counted twice. |
|||
|
|
|
This is directly from Wikipedia and I think is understandable enough, and hence I post it here again.
|
||||
|
|
|
This is simply the first method given in this answer. In that answer, three methods for computing the number of derangements are given. |
|||
|
|