15
$\begingroup$

I was playing around with hypergeometric probabilities when I wound myself calculating the binomial coefficient $\binom{10}{3}$. I used the definition, and calculating in my head, I simplified to this expression before actually calculating anything $$ \frac {8\cdot9\cdot10}{2\cdot3} = 120 $$ And then it hit me that $8\cdot9\cdot10 = 6!$ and I started thinking about something I feel like calling generalized factorials, which is just the product of a number of successive naturals, like this $$ a!b = \prod_{n=b}^an = \frac{a!}{(b-1)!},\quad a, b \in \mathbb{Z}^+, \quad a\ge b $$ so that $a! = a!1$ (the notation was invented just now, and inspired by the $nCr$-notation for binomial coefficients). Now, apart from the trivial examples $(n!)!(n!) = n!$ and $a!1 = a!2 = a!$, when is the generalized factorial a factorial number? When is it the product of two (non-trivial) factorial numbers? As seen above, $10!8$ is both.

$\endgroup$
  • 3
    $\begingroup$ You need to look at rising factorial and falling factorial and n-permutations for some of the notation used. $\endgroup$ – Henry Feb 24 '12 at 0:41
  • $\begingroup$ You could restate your question as asking when is a factorial equal to the non-trivial product of two factorials as in $10! = 7! \times 6!$ and when is it the non-trivial product of three factorials as in $10! = 7! \times 5! \times 3!$. $\endgroup$ – Henry Feb 24 '12 at 1:05
  • $\begingroup$ As for the notation, I was aware of the possibility someone had come up with it before me, it's not like I'm THAT good. And, you are completely right. I could've thought that one more through. It was late at night. $\endgroup$ – Arthur Feb 24 '12 at 13:24
  • 5
    $\begingroup$ The notation $a!b$ has a well-established and familiar meaning: namely the product of $a!$ and $b$. $\endgroup$ – John Bentin Sep 15 '13 at 12:56
  • $\begingroup$ Similar question was answered at MathOverflow. $\endgroup$ – Alexey Ustinov Mar 30 '18 at 11:47
23
$\begingroup$

See Chris Caldwell, The diophantine equation $A!B!=C!$, J. Recreational Math. 26 (1994) 128-133. $9!=7!3!3!2!$, $10!=7!6!=7!5!3!$, and $16!=14!5!2!$ were the only known non-trivial examples of a factorial as a product of factorials as of the 3rd edition of Guy, Unsolved Problems In Number Theory (Problem B23).

$\endgroup$
  • $\begingroup$ Funny I should stumble upon half of the known solutions to an unsolved problem. Thanks for the references, though. I'll look them up. $\endgroup$ – Arthur Feb 24 '12 at 13:25
11
$\begingroup$

Immediately we see that any $n$ that's a power of 2 should work. e.g.

$4! = 3!2!2!$, $8! = 7!2!2!2!$, $16! = 15!2!2!2!2!$

Similarly, by combining whatever small factorials we want, we can take $n!$ and $(n-1)!$ to have that ratio. For instance, suppose we wanted to use $3!5!7! = 3628800$ somewhere; we can then make $3628800! = 3628799!7!5!3!$. You can obviously generate an infinite number of solutions this way.

However, the instances of $n!/(n-1)!$ are probably what Caldwell (see other answer) referred to as "trivial". The non-trivial instances of $n!/(n-k)!$ would be given by

$n(n-1)(n-2) ... (n-k) = A_1!A_2!...A_i!$ which -- in general -- I know no way of tackling the general case, but the case where $i=1$ (that is, $n!=(n-k)!A!$) can be attacked:

$n(n-1)(n-2) ... (n-k) = P(n) = A!$

Which is referred to as a polynomial-factorial diophantine equation. This is still largely open, but has some interesting results proven about it -- namely, that the set of solutions is finite for any polynomial P. An extensive treatment of the matter is given in http://www.ams.org/journals/tran/2006-358-04/S0002-9947-05-03780-3/S0002-9947-05-03780-3.pdf

$\endgroup$
  • $\begingroup$ Very interesting reference, thank you! $\endgroup$ – Jose Brox Feb 18 '16 at 20:43
3
$\begingroup$

Consider the general equation $$a_1!a_2!\cdots a_n! = b!.$$ People like Paul Erdős worked on this kind of equations. So, they are rather serious problems. Even for the simplest case where $n=2$, namely the equation $A!B!=C!$, we don't know whether there is any other non-trivial solution other than $10!=7!6!$.

Erdős proved that in the equation $A!B!=C!$ if we assume $1< A \leq B$, then for large enough values of $C$, the difference of $B$ and $C$ does not exceed $5 \log \log n$. It is clear that $5 \log \log n$ grows to infinity as $n$ increases. But in fact, this is a very slowly growing function. For instance, the first $n$ for which $5 \log \log n = 50$ has more than $9566$ digits. Caldwell proved (1994) that the only non-trivial solution with $C < 10^6$ is $10!=7!6!$. Florian Luca proved in this paper (2007) that considering the $abc$ hypothesis, we have $C-B=1$ for large enough $n$. Read also a remark (2009) on this paper if you're interested.

Suppose that $P(m)$ denotes the largest prime factor of $m$. Erdős also showed in one of his many papers that the assertion $$ P(n(n+1)) > 4 \log n $$ would imply that the general equation $(1)$ has finitely many non-trivial solutions.

The last paragraph is taken from a recent work of Hajdua, Pappa, and Szakács, who prove proved in this paper (2018) that writing $k=B-A$ for all non-trivial solutions of the equation $A!B!=C!$ different from $(A, B, C) =(6, 7, 10)$, we have $C<5k$. Further, if $k \leq 10^6$, then the only non-trivial solution is given by $(A, B, C) =(6, 7, 10)$. I presented their paper in our MathSciNet seminar at UBC and you can find the slides here.

More references:

  1. Discussion on ArtofProblemSolving
  2. Luca, Florian. "The Diophantine equation $P (x)= n!$ and a result of M. Overholt." Glasnik matematički 37.2 (2002): 269-273.
  3. Berend, Daniel, and Jørgen Harmse. "On polynomial-factorial diophantine equations." Transactions of the American Mathematical Society 358.4 (2006): 1741-1779.
$\endgroup$

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.