3
$\begingroup$

Let $k\ge1$. It is known that the number of lattice points on the $k$-sphere $S^k(0)$ (center at the origin, radius $R$), namely the size of $\mathbb{Z}^{k+1}\cap S^k(0)$, is bounded by $R^{k-1+\epsilon}$, $\forall \epsilon>0$.

Now I have a question about this result. We consider a $k$-sphere $S^k(a)$ embedded in $\mathbb{R}^d$ with arbitrary center $a\in\mathbb{R}^d$ and radius $R$, where $d\ge k+1$. Do we have the same estimates about the size of $\mathbb{Z}^{d}\cap S^k(a)$? Can we have uniform estimates independent of the center $a$ and dimension $d$? I don't know if there are any references about this generalization. Any help and comments will be greatly appreciated.

$\endgroup$
  • $\begingroup$ The earlier MO question, "What is the smallest sphere whose surface includes 100 integer points?", even though phrased for origin-centered spheres, contains some info on off-origin spheres. For example, "The sphere centred on $(1,1,1)/2$ and radius $\sqrt{131}/2\approx 5.723$ contains $120$ points with integer coordinates." $\endgroup$ – Joseph O'Rourke Dec 16 '18 at 0:22
  • $\begingroup$ Looks like your question is related to mine mathoverflow.net/questions/319130/… (I didn't see it before posting mine), though I think there is a larger hope a "number theory" argument will help with yours since you want the points to be exactly on the sphere. $\endgroup$ – Rodrigo Dec 20 '18 at 19:23
1
$\begingroup$

Yes, the estimate is uniform in $a$ and $d$, I think. It suffices to handle the two-dimensional case, by considering two-dimensional slices. So consider the circle $(x - a)^2 + (y-b)^2 = R^2$ with $a,b$ arbitrary. We may assume $0 \leq a,b < 1$. Then for any solution $(x,y)$, $x$ and $y$ must be $O(R)$. If there are more than just a few solutions, then we can solve for $a$ and $b$, proving that $a, b$ are algebraic of bounded degree and height $R^{O(1)}$ (this is an "effective nullstellensatz"). Now you use the divisor bound in the extension $K = \mathbf{Q}(i, a,b)$ to bound the number of solutions to $$(x-a)^2 + (y-b)^2 = (x - a + i (y-b))(x - a - i(y-b)) = R^2.$$

$\endgroup$
  • $\begingroup$ In fact, $a$ and $b$ must be rational, with denominators $O(R)$. $\endgroup$ – Sean Eberhard Jan 16 '19 at 7:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.