Let
and
. If
is a solution to the problem, so is
, where
is an arbitrary integer. This is because
and if
then
.
This means we can limit our findings on an arbitrary interval of length 2 and extrapolate from there.
For sake of simplicity take the
. We split the problem into two cases:
Case 1: Interval [-1,0)
This means
Claim:
for all positive integers n.
Using strong induction, base case is trivial
. Assume
for all
Consider
: .
Now
The only way for
to divide
is for
. This concludes the proof.
Now, since
, . Thus
. As the problem's condition holds true for any positive integer n, it must hold for arbitrarily large n, i.e. on the limiting case as n tends to infinity. Taking the limit on our inequality yields
, with no
satisfying the double inequality.
Case 2: Interval [0,1)
This means
.
Claim:
for all positive integers n.
Using strong induction, base case is again trivial
. Assume
for all
Consider
: .
Now
The only way for
to divide
is for
. This concludes the proof.
Since
for any n, it implies
.
Thus the only reals
satisfying the problem are
, where m is an integer.