27
$\begingroup$

I am somewhat new to summations and products, but I know that the sum of the first positive n integers is given by:

$$\sum_{k=1}^n k = \frac{n(n+1)}{2} = \frac{n^2+n}{2}$$

However, I know that no such formula is known for the most simple product (in my opinion) - the factorial:

$$n!=\prod_{k=1}^n k = \prod_{k=2}^n k$$

I don't know if that is how products work, but I would really like to know!

So my question is why is there no explicit formula (in terms of n, other than n(n-1)...2*1) for the product of the first n integers? Is there a proof that says that one cannot exist or is it that one has not been discovered?

By explicit formula I mean a non-functional equation that does not require n amount of steps to calculate - just like the summation formula does not require n additions.

$\endgroup$
7
  • $\begingroup$ Are you familiar with Stirling approximation? $\endgroup$
    – Kunnysan
    Jul 29, 2013 at 8:31
  • 11
    $\begingroup$ This really depends on what one would accept as explicit formula. $\endgroup$ Jul 29, 2013 at 8:36
  • 10
    $\begingroup$ Such a formula exists, but it involves integration :) $$n!=\int_0^{\infty}x^{n}e^{-x}dx.$$ $\endgroup$ Jul 29, 2013 at 8:36
  • 7
    $\begingroup$ @O.L. That's not an improvement—instead of finding it in $n$ steps, you're finding it after an infinite number of steps!! $\endgroup$ Nov 14, 2014 at 16:37
  • 1
    $\begingroup$ @columbus8myhw I never saw it that way until you pointed it out. Nice observation! $\endgroup$ Nov 15, 2014 at 2:38

5 Answers 5

24
$\begingroup$

(The following is a joke.)

Put $a_n:={1\over 6}n(n+1)(n+2)$ and $b_n:={1\over2}n(n+1)$, and define the magic constant $$\xi:=\sum_{n=1}^\infty {n!\over 2^{a_n}}\ =\ 0.630882266676063396815526621896\ldots\quad .$$ Then $$n!=\bigl\lfloor\> 2^{a_n}\xi\>\bigr\rfloor\quad \bigl({\rm mod}\ \ 2^{b_n}\bigr)\qquad(n\geq1)\ .$$ Try it out!

$\endgroup$
3
  • $\begingroup$ Is this an encoding of an infinite sequence of integers into a real number? Hahaha $\endgroup$
    – Memming
    Jul 29, 2013 at 11:52
  • 5
    $\begingroup$ I wonder if this magic constant has any other uses. $\endgroup$ Jul 29, 2013 at 13:55
  • $\begingroup$ How did you arrive at the formula? $\endgroup$
    – BigbearZzz
    Nov 14, 2015 at 2:31
14
$\begingroup$

Since nobody seems to be willing to answer, let me try to summarize the hints from the comments.

First, we can write factorial as $$n!=\int_0^{\infty}x^n e^{-x}dx.$$ One of the consequences of this formula is Stirling's approximation: as $n\rightarrow\infty$, $$n!\sim \left(\frac{n}{e}\right)^n\sqrt{2\pi n}.\tag{1}$$ At first sight, if there was an explicit algebraic formula for $n!$ in terms of $n$, in such expression there would be no place for $e$ or $\pi$; since they are both present in the asymptotics, one is tempted to conclude that no such algebraic formula can exist. Actually this is not quite true: as was pointed to me by Michalis, one can e.g. obtain $e$ as the asymptotics of $\left(1+\frac{1}{n}\right)^n$.

To show non-existence rigorously, you should first rigorously define what do you mean by "explicit formula" $n!=f(n)$, i.e. to define the class of "acceptable" functions $f$. For example, rational $f$ are immediately ruled out by the above asymptotics.

$\endgroup$
3
  • 2
    $\begingroup$ Hm. The existence of $e$ in asymtotics doesn't exclude an algebraic formula. For example $(n+1)^n\sim n^n e$. $\endgroup$
    – MichalisN
    Jul 29, 2013 at 9:58
  • $\begingroup$ @Michalis You are right. Let me think how to say this more correctly. $\endgroup$ Jul 29, 2013 at 10:06
  • 2
    $\begingroup$ The sum of the first $n$ integers takes us from addition up to the next operator, multiplication. So it seems reasonable to look for a formula for $n!$ involving exponentiation. $\endgroup$
    – Jack M
    May 8, 2015 at 0:40
2
$\begingroup$

I thought about Fermat theory and how there must be an algebraic version of

$n!=\frac{{\mathrm d}^n}{{\mathrm d}^nx}\,x^n$

and indeed it let me to a formula of the form

$n!=\sum_{k=0}^nR_{nk}\,k^n$

namely with $R_{nk}=(-1)^{n+k}{n\choose k}$. Of course, you would use factorials to compute the binomials, so there is no computational speedup, but it still makes for some nice looking equations:

0! = + 0^0 1! = - 0^1 + 1*1^1, 2! = + 0^2 - 2*1^2 + 2^2, 3! = - 0^3 + 3*1^3 - 3*2^3 + 3^3, 4! = + 0^4 - 4*1^4 + 6*2^4 - 4*3^4 + 4^4, 5! = - 0^5 + 5*1^5 - 10*2^5 + 10*3^5 - 5*4^5 + 5^5, 6! = + 0^6 - 6*1^6 + 15*2^6 - 20*3^6 + 15*4^6 - 6*5^6 + 6^6, 7! = - 0^7 + 7*1^7 - 21*2^7 + 35*3^7 - 35*4^7 + 21*5^7 - 7*6^7 + 7^7, 8! = + 0^8 - 8*1^8 + 28*2^8 - 56*3^8 + 70*4^8 - 56*5^8 + 28*6^8 - 8*7^8 + 8^8

The theorem underlying here is that, for all $n$

$\sum_{k=0}^n\dfrac{(-1)^k (-k)^n}{k!\,(n - k)!}=1$

$\endgroup$
1
$\begingroup$

enter image description here

enter image description here

These two formulas give n! This was discovered by Euler. Reference this link for further reading. http://eulerarchive.maa.org/hedi/HEDI-2007-09.pdf

$\endgroup$
1
-2
$\begingroup$

If I am not mistaken, Manjul Bhargava generalized the factorial function to include numbers such as (pi)! and (2+√3)!

$\endgroup$
4
  • 2
    $\begingroup$ This generalization is tightly related to so called gamma function, which has been discovered long before Bhargava was born, and this is in no way related to the question. en.m.wikipedia.org/wiki/Gamma_function $\endgroup$
    – Wojowu
    Nov 14, 2015 at 7:34
  • $\begingroup$ Oh interesting. I didn't see that the original poster (OP) asked for an explicit formula. $\endgroup$
    – Yunus Syed
    Nov 14, 2015 at 14:46
  • 1
    $\begingroup$ But this is the only thing which OP asks for. This is even said in the question title. $\endgroup$
    – Wojowu
    Nov 14, 2015 at 15:25
  • $\begingroup$ I misread it, sorry. $\endgroup$
    – Yunus Syed
    Nov 14, 2015 at 16:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.