6

I'm looking for a fast method. I just multiplied all the numbers together modulo ten and divided by $5^{12}$ and $2^{12}$ modulo 10, which gave me $2$.

2 Answers2

10

Here is a fast method to determine the last non-zero digit of $n!$.

Let the base $5$ representation of $n \in\mathbb{Z}^+$ be $\overline{a_{m}a_{m-1} \ldots a_{0}}_{5}$.

Prove that the last non-zero digit of $n!$ is $\equiv 2^{\sum\limits_{i=0}^{m}ia_{i}} \times \prod\limits_{i=0}^{m}a_{i}! \pmod{10}$.

This was a question I set for some quiz in the past, and can easily be done by induction. I leave this as an exercise.

This is fast in general since $a_i!$ is small and exploiting $2^{4k+1} \equiv 2 \pmod{10}$.


Applying this to $n=50$, $50=200_5$, so the last non-zero digit of $50!$ is $$\equiv 2^{0(0)+1(0)+2(2)}(2!0!0!) \equiv 2 \pmod{10}$$


Let's apply this to $n=2013$: $2013=31023_5$ so the last non-zero digit of $2013!$ is $$\equiv 2^{0(3)+1(2)+2(0)+3(1)+4(3)}3!1!0!2!3! \equiv 2^{17}(72) \equiv 2(2) \equiv 4 \pmod{10}$$


Motivation

Well I guess I should provide some motivation. Some of this will already be in the comments. The motivation will actually be quite close to a full proof, but I've left out the more boring/easy/straightforward parts.

We may easily determine the last digit of the product of all the non-multiples of $5$ which are $\leq n$, i.e.

$$\prod_{1 \leq k \leq n, 5 \nmid k}{k}$$

by using the fact that $(5i+1)(5i+2)(5i+3)(5i+4) \equiv 4 \pmod{10}$. We get $4^j(a_0!)$ where $n=5j+a_0$.

Intuitively, each power of $5$ sort of "removes a power of $2$" from this product, since they combine to form a power of $10$. So let's start with accounting for the multiples of $5$, and worry about multiples of $25, 125, \ldots$ later if need be.

For convenience, define $f(m)$ to be the last non-zero digit of $m$. By virtue of the fact that $n!$ is even for $n>1$, we may write at least for $n>1$

\begin{align} f(n!) \equiv f(6^jn!) & \equiv f(6^j\prod_{1 \leq k \leq n, 5 \nmid k}{k}\prod_{1 \leq k \leq j}{5k}) \pmod{10} \\ & \equiv f\left(30^j\left(\prod_{1 \leq k \leq n, 5 \nmid k}{k}\right)j!\right) \pmod{10} \\ & \equiv f(3^j[4^j(a_0!)]j!) \pmod{10} \\ & \equiv 2^ja_0!f(j!) \pmod{10} \end{align}

Already we can see that it is worthwile to consider the base $5$ expansion of $n$, and the term $\prod_{i=0}^{m}{a_i!}$ in the formula is apparent. What's left is to look at the product of the $2^j$ terms and express it in terms of $a_i$, and to prove the formula we get by strong induction.

Ivan Loh
  • 17,115
  • Very nice, but without further explanation it comes off as a bit too clever. How did you discover this crazy formula? – dfeuer Nov 16 '13 at 23:22
  • @dfeuer Start with a recurrence relation for the last non-zero digit of $n!$, relating the last non-zero digit of $n!$ to the last non-zero digit of $\lfloor \frac{n}{5} \rfloor!$. It isn't too hard from there to guess what the formula should be, and then to use the recurrence to prove the formula by induction. – Ivan Loh Nov 16 '13 at 23:30
  • And why would I think to relate the last non-zero digits like that? – dfeuer Nov 16 '13 at 23:33
  • 1
    @dfeuer note that $(5i+1)(5i+2)(5i+3)(5i+4) \equiv 4 \pmod{10}$ so we can group all the non-multiples of $5$ and determine their product $\pmod{10}$. Then we just need to account for the powers of $5$ which will sort of "remove" powers of $2$ in your product. – Ivan Loh Nov 16 '13 at 23:36
  • @dfeuer I just saw this question under the related list, which establishes a recurrence. My motivation was from a recurrence I got while playing around with the last nonzero digit which was quite similar to that question. – Ivan Loh Nov 16 '13 at 23:37
  • 1
    The issue is a common one, I think. Someone figures something out and proudly displays the proof, but throws out the hardest and most instructive part—the figuring out. I've surely been guilty of this myself. There's always a strong temptation to "clean up" a proof, getting rid of anything not logically necessary but making everyone wonder how you got there. – dfeuer Nov 16 '13 at 23:50
4

You have correctly figured that $5^{12}$ is the highest power of $5$ that divides $50!$. Similarly, $\lfloor\frac{50}{2}\rfloor+\lfloor\frac{50}{4}\rfloor+\lfloor\frac{50}{8}\rfloor+\lfloor\frac{50}{16}\rfloor+\lfloor\frac{50}{32}\rfloor=25+12+6+3+1=47$, so $2^{47}$ is the highest power of $2$ that divides $50!$. We can similarly figure out the other highest powers of primes dividing $50!$: $\lfloor\frac{50}{3}\rfloor+\lfloor\frac{50}{9}\rfloor+\lfloor\frac{50}{27}\rfloor=16+5+1=22$, and $\lfloor\frac{50}{7}\rfloor+\lfloor\frac{50}{49}\rfloor=7+1=8$, and so forth, so in fact $$50!=2^{47}\cdot3^{22}\cdot5^{12}\cdot7^{8}\cdot11^{4}\cdot13^{3}\cdot17^{2}\cdot19^2\cdot23^2\cdot29\cdot31\cdot37\cdot41\cdot43\cdot47.$$ Clearly, the number ends in exactly $12$ zeroes, so the desired digit is $50!/10^{12}\bmod10$, or $$2^{35}\cdot3^{22}\cdot7^{8}\cdot1^{4}\cdot3^{3}\cdot7^{2}\cdot9^2\cdot3^2\cdot9\cdot1\cdot7\cdot1\cdot3\cdot7\bmod10,$$ which is not too hard to compute using identities like $3^4=9^2\equiv1\bmod10$, $3\cdot7\equiv1\bmod10$. (It might not be super elegant, but I don't think there is a much quicker method.)

Harald Hanche-Olsen
  • 32,608
  • 2
  • 61
  • 87