Ivan Shishkin, Rye (1878)

Discussions

Integers as consecutive sums

0 messages

Solution

Solution by Ancient Tree · EN

Starting to check with some examples, we notice quickly that it seems like only the powers of two cannot be expressed this way.

  • To prove this, suppose first that our integer a=2ka=2^k can be written as a sum of consecutive integers :

2k=n+(n+1)++(n+m)=mn+(m1)m22^k=n+(n+1)+\ldots+(n+m)=mn + \frac{(m-1)m}{2}

with m2m\geq 2. Then by multiplying by 2 to get rid of the denominator :

2k+1=2mn+(m1)m2^{k+1}=2mn + (m-1)m

which shows that mm divises 2k+12^{k+1}, hence mm is a power of 2 itself, say, m=2rm=2^r, with r1r\geq 1.
Plugging this back into the equation gives :

2k+1=2r+1n+(2r1)2r=2r(2n+2r1)2^{k+1}=2^{r+1} n+\left(2^r-1\right) 2^r=2^r\left(2 n+2^r-1\right)

But this is impossible : the term 2n+2r12n+2^r-1 is odd, so it cannot be a factor of 2k+12^{k+1}.
This contradictions means that 2k2^k can never be written as a sum of consecutive integers.

  • Now suppose that our integer aa is not a power of 2. This means that aa has an odd factor, say qq, such that a=2kqa=2^k q for some k0k\geq 0. Let’s show that it can be written as a sum of consecutive integers. We work backwards from the answer : assume that it can be written this way, that is, there is a positive integer nn and m2m\geq 2 such that :

a=2kq=n++(n+m1)=mn+m(m1)2a=2^{k}q=n+\ldots+(n+m-1)=mn+\frac{m(m-1)}{2}

And again multiplying by 2 on each side to get rid of the denominator :

2k+1q=m(2n+m1)2^{k+1} q=m(2 n+m-1)mm and 2n+m12n+m-1 are of opposite parity. Now, a natural choice is to suppose one of the two following :

  1. 2k+1=m2^{k+1}=m and 2n+m1=q2n+m-1=q. We get that n=q+122kn=\frac{q+1}{2}-2^{k}. This is positive for q+122k+1\frac{q+1}{2}\geq 2^{k}+1. If this condition is verified, let’s check that this choice of mm and nn yields the right decomposition :

n++(n+m1)=mn+m(m1)2=2k+1(q+122k)+2k+1(2k+11)2=2kq=an+\ldots+(n+m-1)=m n+\frac{m(m-1)}{2}=2^{k+1}\left(\frac{q+1}{2}-2^k\right)+\frac{2^{k+1}\left(2^{k+1}-1\right)}{2}=2^k q=a2) If 2k+1=2n+m12^{k+1}=2n+m-1 and q=mq=m, we get that n=2kq12n=2^{k}-\frac{q-1}{2}. This is positive for q+122k\frac{q+1}{2} \leqslant 2^k. Again, it is straightforward to check that if this condition is verified, then this choice of nn and mm are correct.

Because one of the two conditions has to be true, then the property has been shown.

No messages yet.