I need help with this combinatorial proof: $1\cdot1!+2\cdot2!+\cdots+n\cdot n!=(n+1)!-1$ So far I came up with this: Let S be a set of numbers $1, 2, \ldots, n+1$ So LHS could be: How many permutations of length $k$ there is over set $S$ such that number $1$ is not the first in permutation. $k$ can be $2, 3,\ldots, n+1$. But the RHS gives me trouble... Thanks. (Sorry for the poor formating, I am not good with HTML)
-
2it's a duplicate of a question on this site please search before asking – RE60K Feb 20 '15 at 17:20
-
3@ADG: If you know a duplicate, then please reveal what it is instead of being smug and unhelpful. – hmakholm left over Monica Feb 20 '15 at 17:21
-
Do you need a bijective proof/combinatoiral proof or just whatever proof? – quid Feb 20 '15 at 17:21
-
@HenningMakholm what if you have seen it a while ago and can't seem to find it easily? – RE60K Feb 20 '15 at 17:23
-
4@ADG: Then you shouldn't imply that it's the asker's fault that they can't locate it better than you. – hmakholm left over Monica Feb 20 '15 at 17:24
-
@ADG: That was brilliant! Lol! – Laars Helenius Feb 20 '15 at 17:27
-
possible duplicate of Prove by Mathematical Induction: $1(1!) + 2(2!) + \cdot \cdot \cdot +n(n!) = (n+1)!-1$ – Jack D'Aurizio Feb 20 '15 at 17:47
-
Duplicate question found. By the way, we really need to improve the search system - already pointed that out on Meta, a while ago. – Jack D'Aurizio Feb 20 '15 at 17:49
-
3@Jack: The purported duplicate asks for a proof by induction; this one wants a combinatorial proof. – hmakholm left over Monica Feb 20 '15 at 17:58
-
1Such a waste of power. Anyway, I got your point, I am retracting my close vote. – Jack D'Aurizio Feb 20 '15 at 17:59
-
@Jack: You can do that now? Sigh, I've been away for too long. – hmakholm left over Monica Feb 20 '15 at 18:01
-
@HenningMakholm: I've been a wafer too long (https://www.youtube.com/watch?v=Y0A71tITqe0) – Jack D'Aurizio Feb 20 '15 at 18:03
4 Answers
$n\cdot n!$ counts the number of ways to order $\{1,2,3,\ldots,n+1\}$ without $n+1$ being first.
$(n-1)\cdot (n-1)!$ counts the number of ways to order $\{1,2,3,\ldots,n+1\}$ with $n+1$ being first and $n$ not second.
$(n-2)\cdot (n-2)!$ counts the number of ways to order $\{1,2,3,\ldots,n+1\}$ with $n+1$ being first and $n$ second and $n-1$ not third.
Etc...
The RHS is all the ways to order $\{1,2,3,\ldots,n+1\}$ without all the elements in descending order.
- 8,115
- 1
- 24
- 35
-
1
-
and yes, I needed a combinatorial proof only, not algebraic, nor by induction :-) – Lara Ljumović Feb 24 '15 at 13:14
-
-
We want to show that the permutations of $n+1$ elements, except for one of them, are in bijective correspondence with the pairs $(j,\tau)$ where $\tau$ is a permutation of $k\le n$ elements and $1\le j\le k$, for some $k$.
Here's a correspondence. We start with a permutation $\sigma$ of $\{1,2,3,\ldots,n+1\}$, for example
i 1 2 3 4 5 6 7 8 9 10
sigma(i) 1 6 3 2 7 5 4 8 9 10
If the permutation is the identity, then it doesn't map anywhere! Otherwise let $k+1$ be the largest index such that $\sigma(k+1)\ne k+1$. In the example $k+1=7$. We have $k\ge 1$ because index $1$ can't be the latest difference between $i$ and $\sigma(i)$.
Let the first output $j$ be the number such that $\sigma(k+1-j)=k+1$. In other words, $j$ is the number of elements to the right of the $k+1$ in the lower line of the permutation, before the trailing sequence of unchanged positions. In the example above we have $j=2$ because the two elements 5 4 come between 7 and the final 8 9 10.
The second output $\tau$ is now defined by $\tau(i) = \begin{cases} \sigma(i) & i < j \\ \sigma(i+1) & i \ge j \end{cases}$
i 1 2 3 4 5 6
tau(i) 1 6 3 2 5 4
This correspondence is bijective -- we can do it in reverse starting with $(j,\tau)$. First write down $\tau$, then insert the number $k+1$ before the $j$th-to-last element, and finally append the numers $k+2, k+3, \ldots, n+1$ at the end of the permutation.
- 291,611
Proof by induction works here. This is not a combinatorial proof though (sorry!).
At $n=1$, the statement $1\cdot 1!=2!-1$ is true; both sides equal $1$.
Assume the statement for $n=k$ is true: $$1\cdot 1!+2\cdot 2!+\cdots+k \cdot k!=(k+1)!-1$$
For $n=k+1$, we have \begin{align} 1\cdot 1!+2\cdot 2!+\cdots+(k+1) \cdot (k+1)!&=1\cdot 1!+2\cdot 2!+\cdots+k\cdot k!+(k+1) \cdot (k+1)! \\ &=(k+1)!-1+(k+1)\cdot(k+1)! \\ &=(k+1)!(1+(k+1))-1 \\ &=(k+1)!(k+2)-1 \\ &=(k+2)!-1 \end{align}
- 13,952
-
3This might not answer the question since it asks for a combinatorial proof. – quid Feb 20 '15 at 17:32
hint: $n\cdot n! = (n+1)!-n!$. Can you continue?
- 78,689
-
3That settles it algebraically, but it's not really a combinatorial proof ... – hmakholm left over Monica Feb 20 '15 at 17:30