8

At our college, a professor told us to prove by a semi-formal demonstration (without complete induction):

  • For every odd natural: $24\mid(u^3-u)$

He said that that example was taken from a high school book - maybe I din't get something, but I really have no idea to prove that without using complete induction.

Any idea of smart demonstration?

Thanks for your help. (Excuse my bad English.)

fridojet
  • 203

7 Answers7

21

First note that $$ u^3-u = (u-1)u(u+1)$$

Given that $u$ is odd.In this case, $u-1$ and $u+1$ are even and one of them is divisible by 4. This follows from the basic observation that one of any two consecutive even numbers is divisible by 4. So, $(u-1)(u+1)$ is divisible by $4 \times 2 =8$.

Also, one of any three consecutive natural numbers is divisible by 3. So, one of $u-1,u,u+1$ is divisible by 3.

So, $(u^3-u)$ is divisible by 8 and 3, which are co-prime. So, it is divisible by $8\times 3=24$

dexter04
  • 2,001
13

Factor it: $u^3-u=u(u^2-1)=u(u-1)(u+1)$. Now show that one of the three factors must be divisible by $4$ and another by $2$, and that one must be divisible by $3$.

Brian M. Scott
  • 631,399
6

Observe that $u^3-u=(u-1)u(u+1)$. Now $3$ consequtive numbers are always divisible by $3$. So $3\mid u^3-u.$ Since $u$ is odd $\Rightarrow$ $u-1, \ u+1$ are even. Prove that one of $u-1, \ u+1$ is divisible by $4$ and the other by $2$. Conclude that $24\mid u^3-u.$

P..
  • 15,189
6

Another proof.

I assume you know the following formula:

$$1^2+2^2+...+k^2 = \frac{k(k+1)(2k+1)}{6}$$

for all integer $k\geq 0$.

If $u$ is odd then $u=2k+1$ for some integer $k\geq 0$. Expanding: $$\begin{eqnarray}u^3-u &=& (2k+1)^3-(2k+1) \\ &=& 8k^3 + 12k^2 + 6k + 1 - (2k+1) \\ &=& 8k^3 + 12k^2 + 4k \\ &=& 4k(2k+1)(k+1) \\ &=& 24\frac{k(k+1)(2k+1)}{6}\end{eqnarray}$$

So if $u$ is odd, then $\frac{u^3-u}{24}$ is the sum of the first $k=\frac{u-1}{2}$ squares, and hence is an integer.

Chris Watson
  • 275
  • 1
  • 11
Thomas Andrews
  • 186,215
1

Here is a slightly different proof, showing $u(u^2 - 1)$ is divisible by $24$ when $u$ is odd.

Since $u$ is odd, $u$ is of the form $2k + 1$, where $k$ is an integer.
Rewriting, we have: $$(2k + 1)((2k + 1)^2 - 1)$$ $$(2k + 1)(4k^2 + 4k)$$ $$(2k + 1)4k(k + 1)$$ Let $m$ denote this last expression.

Every integer (and thus $k$) is of one of the three forms: $3q$, $3q - 1$, or $3q - 2$, where $q$ is an integer.

Case $k$ = $3q$:
$k$ is divisible by $3$ and either $k$ or $k + 1$ is even (and thus divisible by $2$).
Thus $m$ is of the form $4(3)(2)h$ = $24h$ and therefore divisible by $24$.

Case $k$ = $3q - 1$:
$k + 1$ is divisible by $3$ and either $k$ or $k + 1$ is even (and thus divisible by $2$).
Thus $m$ is of the form $4(3)(2)h$ = $24h$ and therefore divisible by $24$.

Case $k$ = $3q - 2$:
$2k + 1$ = $6q - 3$ is divisible by $3$ and either $k$ or $k + 1$ is even (and thus divisible by $2$).
Thus $m$ is of the form $4(3)(2)h$ = $24h$ and therefore divisible by $24$.

Q.E.D.

1

Hint $ $ Induction step: $\rm\:24\:|\:f(n) = n^3\!-n\:\Rightarrow\:24\:|\:f(n\!+\!2) = f(n) + 6(n\!+\!1)^2\:$ by $\rm\:n\!+\!1\:$ is even.

Remark $\ $ If we telescsope this we obtain a nice representation showing the factor of $24$.

$$\rm\begin{eqnarray} f(2n\!+\!1)\! -\! f(1)\, &=&\,\rm f(2n\!+\!1)\!&&\rm-\ f(2n\!-\!1) &&\rm+\ \,\cdots\, + f(5)\!&&\rm-\ f(3) &&\rm +\ \ f(3)\!&&\rm-\ f(1) \\ \,&=&\,\rm &&\!\!\!\!\!\rm6(2n)^2&&+\ \,\cdots\, + &&\!\!\!6\cdot4^2&& +&&\!\!\!6\cdot2^2\end{eqnarray}$$

So, using $\rm\:f(1) = 0,\:$ and pulling out a factor of $\,4\,$ from the RHS via $\rm\,(2k)^2 = 4\cdot k^2$ we obtain

$$\rm\ f(2n\!+\!1)\, =\, 24\, (n^2 + (n\!-\!1)^2 +\, \cdots\, + 2^2 + 1^2)$$

making the factor of $24$ clear. This is essentially the result in Thomas's answer except that we have derived it mechanically (requiring no insight or special knowledge), using only the very simple idea of telescopy - about which you can find much more written in many of my prior posts on telescopy.

Bill Dubuque
  • 282,220
  • 1
    Poster said, "without using complete induction." Not sure what he meant by "complete," but I think he means without induction – Thomas Andrews Nov 09 '12 at 00:52
  • @Thomas But your answer is essentially the same as this - see my added remark. "Without induction" is meaningless, does a proof not "use induction" if it invokes a lemma that uses induction? (as you do) – Bill Dubuque Nov 09 '12 at 01:22
0

This very old question recently popped up, and none of the answers given are the one that occurred to me first. Here it is: $u^3-u=u(u^2-1)$. It is well known that the square of any odd number is $\equiv 1 \bmod 8$, meaning that $8\mid (u^2-1)$. It is also well known that if $3\not \mid u$ then $u^2 \equiv 1 \bmod 3$. Hence, either $3\mid u$ or $3\mid (u^2-1)$. Therefore, $3\mid u(u^2-1)$ and $8\mid u(u^2-1)$, meaning $24\mid u(u^2-1)=u^3-u$ QED.

  • This is essentially the same as many other ($10$ year prior) answers, e.g. those of dexter04 and Brian and P... Please don't duplicate answers. – Bill Dubuque Sep 18 '23 at 16:06
  • @Bill Dubuque The other answers you refer to break $u^3-u$ into three factors and examine the modular properties of those three factors. I break $u^3-u$ into two factors and examine the modular properties of those two factors. The prior answers make no mention of squares either $\bmod 8$ or $\bmod 3$. dexter04 asserts that one of the even factors is divisible by $4$, and the other respondents direct the reader to prove as much, in order to achieve divisibility by $8$. Even if you think I cover the same ground, I do so more economically, and hence (I believe) distinctly. – Keith Backman Sep 18 '23 at 22:05
  • If you justify your "well-known" claims you'll get proofs equivalent to the other answers (or many other dupes of this FAQ). One shouldn't add new answers to old FAQs unless they add significant novelty. – Bill Dubuque Sep 18 '23 at 22:17
  • Odd numbers are $\equiv 1,3,5,7 \bmod 8$. Odd squares are $\equiv 1,9,25,49 \equiv 1 \bmod 8$. There is no resort to arguing about how many factors of $2$ are present in consecutive even numbers. Apologies if that is insignificant novelty. – Keith Backman Sep 19 '23 at 14:44
  • Of course that method (and all common methods) have already been posted here many tens (if not hundreds) of times, e.g. here. – Bill Dubuque Sep 19 '23 at 15:06
  • First you criticize me for not justifying what I characterized as "well known" and when I show the justification you criticize me because the justification is so well known! Whether or not it is thought to be well known by me or you or anyone else, it is not pointed to by any of the previous answers, most of which rely on a distinct but equally well-known mathematical truth; without, I note, drawing any criticism from you. – Keith Backman Sep 19 '23 at 17:14