So for every integer $k$ the point $(x,y)$ lies on the circle $x^2
+ y^2 = 1$. (Note $(k^2 + 1) \neq 0$.)
Assume that it is possible for rational points to lie on the circle
$x^2 +y^2 = 3$. For example $$\left(\frac{m}{n}\right)^2 +
\left(\frac{k}{l}\right)^2 = 3$$ where $m$, $n$, $k$, $l$ are
integers, $n$, $l \neq 0$, gcd $(m,n)$ = 1, gcd $(k,l)$ = 1. This
can be rearranged to give $(ml)^2 + (kn)^2 = 3(nl)^2$ and
hence
So the only possible way for $(ml)^2 + (kn)^2 \equiv 0$ (mod 3) is
for $(ml)^2 \equiv 0$ (mod 3) and $(kn)^2 \equiv 0$ (mod 3).
Case 1: suppose that $m \equiv 0$ (mod 3). Then $n$ not$\equiv 0$
(mod 3) because gcd $(m,n) = 1$, so that $k \equiv 0$ (mod 3). Then
$l$ not $\equiv 0$ (mod 3) because gcd $(k,l)$ = 1. We write $m =
3m_1$ and $k = 3k_1$ where $m_1$ and $k_1$ are integers; then as
$(ml)^2 + (kn)^2 = 3(nl)^2$ we have $$3(m_1^2l^2 +k_1^2n^2) =
(nl)^2.$$ This shows that $nl \equiv 0$ (mod 3) which is a
contradiction.
Case 2: suppose that $m$ not$\equiv 0$ (mod 3). Then $l \equiv 0$
(mod 3) and by the same reasoning as before we reach a
contradiction.
Our assumption must be wrong, therefore no rational points lie on
the circle $x^2 +y^2 = 3$.
In turn 4 people throw away three nuts from a pile and hide a
quarter of the remainder finally leaving a multiple of 4 nuts. How
many nuts were at the start?