3

I'm working my way through The Mathematical Olympiad Handbook by A. Gardiner.

In the section A little useful mathematics right at the start the following basic identities are laid out

  • $x^n -1 = (x-1)(x^{n-1} + ... + 1)$

  • $x^n - y^n = (x-y)(x^{n-1} + ... + y^{n-1})$

  • $x^n + 1 = (x+1)(x^{n-1} + ... + 1)$

  • $x^n + y^n = (x-y)(x^{n-1} + ... + y^{n-1})$

In this context the following two exercises are posed

  1. Prove $2^{55} + 1$ is divisible by $33$
  2. Prove $1900^{1990} -1$ is divisible by $1991$

So for (1) the strong implication seems to be to somehow use the fact that $33=2^5+1=(2+1)(2^4-2^3+2^2-2^1+1)$ and use that, but other than realising that I am slightly lost.

I can see using Fermat's Little Theorem allows an attack ($2^{55}=(2^5)^{11}$ etc.) but how to do it with just those identities I can't see it as Fermat's Little Theorem is used later as a second proof of this problem in the number theory section.

Can anyone see how to do it?

Oussema
  • 641

3 Answers3

5

Hint:

$$2^{55} +1 =(2^5)^{11} +1 = 32^{11} +1 = (32+1)(\dots) $$ using your third identity.

Vishu
  • 14,469
  • Thanks seems so obvious now. I am struggling to use a similar technique to crack $1900^{1990}-1$ divisible by 1991. Any idea? – Gau55man Dec 19 '20 at 14:40
  • 1
    @Gau55man Have you copied down the problem correctly? Wolfram says that is not true https://www.wolframalpha.com/input/?i=+Is+%281900%5E%281990%29-1%29%2F1991+an+integer – Vishu Dec 19 '20 at 14:53
  • Yes I've triple checked $1900^{1990} -1$ is divisible by 1991. Most sensible pattern would mean what was intended was $1990^{1990}-1$ is divisible by $1991$ I'm guessing. – Gau55man Dec 19 '20 at 16:38
  • @Gau55man What have you checked it with? – Vishu Dec 19 '20 at 18:01
  • I meant I tripled checked that is what the questions says so there is a typo/mistake in my book. I'm assuming intention is $1990^{1990} -1$ which admits solution by recognising as others have that $(1990^2)^{955} -1 = (1990^2 - 1)(...)$ then using identity $n^2 -1=(n+1)(n-1)$ which gives $1990^2 -1=1991 \cdot 1989$ showing that 1991 is a factor. – Gau55man Dec 19 '20 at 22:35
  • @Gau55man Yes, that is probably the case. – Vishu Dec 20 '20 at 09:54
3

$$(2^{5})^{11} + 1 = 32^{11} + 1 \overset * \equiv (-1)^{11} + 1 = 0 \pmod{33}$$ $(*)$: $32 = 33 - 1 \equiv -1 \pmod {33}$

VIVID
  • 11,604
1

For divisibility of $2^{55}+1$ by $33,$ we can write $$2^{55}+1=(2^{5})^{11}+1=(33-1)^{11}+1,$$ and apply the third identity.

Isn't there a typo in the second exercise? I suggest to prove divisibility of $1990^{1990} -1$ by $1991$ instead. We have
$$1990^2 -1=1991\times 1989$$ and use the first identity.

user376343
  • 8,311