The purpose of this paper is to show how a sieve method which has had many applications to problems involving rational primes can be modified to derive new results on Gaussian primes (or, more generally, prime ideals in algebraic number fields). One consequence of our main theorem (Theorem 2 below) is the following result on rational primes.

In this paper is considered the average size of the 2-Selmer groups of a class of quadratic twists of each elliptic curve over ℚ with ℚ-torsion group ℤ2 × ℤ2. The existence is shown of a positive proportion of quadratic twists of such a curve, each of which has rank 0 Mordell-Weil group.

Voronoĭ conjectured that every parallelotope is affinely equivalent to a Voronoĭ polytope. For some m, a parallelotope is defined by a set of m facet vectors pi, and defines a set of m lattice vectors ti, for 1≤i≤m. It is shown that Voronoĭ's conjecture is true for an n-dimensional parallelotope P if and only if there exist scalars γi, and a positive definite n × n matrix Q such that γipi = Qti for each i. In this case, the quadratic form f(x) = xTQx is the metric form of P.

The Čech compactification of the set of integers is known [6] to have the remarkable property that it has no closed subsets of cardinal ℵ0 or c, every infinite closed subset of it having 2c points. The main object of the present paper is to investigate whether similar gaps in the cardinals of closed subsets can occur in metric spaces. We shall see that the situation there is rather different; if the generalized continuum hypothesis is assumed, there are no gaps, and in any case the missing cardinals, if any, must be big rather than small. The main results are obtained in §3; in particular, we completely determine the cardinals of the closed subsets of complete metric spaces, and also how many closed subsets of each cardinal there are. The methods depend on a study of the discrete subsets of metric spaces, which is carried out in §2, and which may be of independent interest. In conclusion, we briefly consider some fragmentary results for non-metric spaces, in §4. Throughout, we assume the axiom of choice but not the continuum hypothesis.

§1. Introduction and notation. In [1] and [2], Besicovitch demonstrated that there exist plane sets of measure zero containing line segments (and indeed entire lines) in all directions in the plane. It is natural to ask about the existence of analogous sets in Euclidean spaces of higher dimensions, and in [3] we defined an (n, k)-Besicovitch set to be a subset A of Rn, of n-dimensional Lebesgue measure zero, such that for each k-dimensional subspace Π of Rn, some translate of Π intersects A in a set of positive k-dimensional measure. (Thus Besicovitch's original constructions were for (2,1)-Besicovitch sets.) Recently, Marstrand [5] has shown (by approximating to sets by unions of cubes) that no (3, 2)-Besicovitch sets exist, and simultaneously the author [3] proved using Fourier transform methods that (n, k)- Bsicovitch sets cannot exist if k > ½n.

It is natural to say that a set S in a metric space has infinite generalized Hausdorff dimension if there is no Hausdorff measure Λh with Λh(S) = 0. In this note we study such sets. We first need some definitions.

We say that h(x) is a Hausdorff measure function if it satisfies the conditions:

In the preceding paper of the same title (cf. [1]) I defined the notion of the principal genus GK of a finite number field K as the least ideal group, which contains the group IK of totally positive principal ideals and is characterized rationally. The quotient group of the group AK of ideals in K modulo GK is the genus group, its order (Ak: GK) = gK is the genus number, which is thus a factor of the class number hK (in the narrow sense). Associated with the genus group is the genus-field, of K, which is defined as the maximal non-ramified extension of K composed of K and of some absolutely Abelian field.

Helly's theorem asserts that if W is a family of compact convex sets in a j-dimensional linear space, and if any j + 1 members of ℱ have a non-empty intersection, then there is a point common to all members of ℱ. If one attempts to generalise this result to the case when ℱ consists of sets which are expressible as the union of at most n disjoint compact convex sets then, in general, one finds that there is no number h(n, j) such that if any h(n, j) members of ℱ have a non-empty intersection, then there is a point common to all members of ℱ. The difficulty lies in the fact that, in general, the intersections of members of ℱ are more complicated in structure than are the members of ℱ.

We say that a set of closed circular discs of radii r1r2, …, all lying in a Euclidean plane, is saturated if and only if r = inf ri > 0 and any circle of radius r has at least one point in common with a circle of the set. For any set X we use α(X) to denote the area of X. If X denotes the point set union of the discs and X(k) the part of X inside the disc whose centre is the origin and radius k then by the lower density of the covering we mean . The problem is to find the exact lower bound of the lower density for any saturated set of circles. We show that it is φ/(6√3) provided the circles are disjoint. The general case, when they may overlap, remains unsolved.

If G is a topological group then we can think of G acting on itself by multiplying on the left. We would like to know when this action has the property that whenever g and h are distinct elements of G, then the element xg does not get arbitrarily close to xh as x varies in G. It is natural to say that this is the case if {(xg, xh): x∈G} is separated from the diagonal of G × G by a uniform neighbourhood of the diagonal.

Let D(f) denote the discriminant of a binary cubic form

having integral coefficients. We restrict ourselves to forms with D(f) ≠ 0, and further (as a matter of convenience) to forms which are irreducible in the rational field. It is a problem of some interest, in connection with the approximation properties of transcendental numbers, to estimate the sum

as H→∞, where ‖f‖ = max(|a|, |b|, |c|, |d|). Since |D(f)| ≪ H4 when ‖f‖ < H, there is the trivial lower bound

One of the problems of solid state physics is to explain why the atoms of certain elements (such as iron) arrange themselves in a body-centred cubic lattice, rather than in the much denser face-centred cubic lattice packing [8]. Recently, we discovered the geometric significance of these body-centred lattice packings: they are fragile in the sense that, assuming we have a sphere of fixed radius at each lattice point, any perturbation which does not alter the set of nearest neighbours of any sphere results in a denser configuration [3]. In other words, in-these packings it is the density of the interstitial void between the spheres that is being maximized. This fact might seem only to deepen the mystery of why such arrangements ever appear in nature at all. Our purpose here is to demonstrate that these lattice packings are in fact quite “stable”, provided one seeks a more subtle form of stability, one prompted by physical rather than purely geometrical considerations. More precisely, we will show for those lattices which possess a high degree of symmetry–and for the bodycentred cubic lattice in particular–that the EPSTEIN ZETA FUNCTION of the lattice, ∑ |r|-2s (where |r| is the distance of the r-th lattice point from the origin and s is any fixed number greater than n/2), is locally minimized, in the sense that, any lattice, obtained by a perturbation which does not alter the set of nearest neighbours of any lattice point, has an Epstein zeta function of larger value, for any s > n/2.

We study the average value of the divisor function $\unicode[STIX]{x1D70F}(n)$ for $n\leqslant x$ with $n\equiv a~\text{mod}~q$ . The divisor function is known to be evenly distributed over arithmetic progressions for all $q$ that are a little smaller than $x^{2/3}$ . We show how to go past this barrier when $q=p^{k}$ for odd primes $p$ and any fixed integer $k\geqslant 7$ .

In [3] the authors introduced the notion of a completely 0-simple semigroup of quotients. This definition has since been extended to the class of all semigroups giving a definition of semigroups of quotients which may be regarded as an analogue of the classical ring of quotients. When Q is a semigroup of quotients of a semigroup S, we also say that S is an order in Q.

The isometries of the space of convex bodies of Ed with respect to the symmetric-difference metric are precisely the mappings generated by measurepreserving affinities of Ed.

Answering a question of Füredi and Loeb [On the best constant for the Besicovitch covering theorem. Proc. Amer. Math. Soc.121(4) (1994), 1063–1073], we show that the maximum number of pairwise intersecting homothets of a $d$-dimensional centrally symmetric convex body $K$, none of which contains the center of another in its interior, is at most $O(3^{d}d\log d)$. If $K$ is not necessarily centrally symmetric and the role of its center is played by its centroid, then the above bound can be replaced by $O(3^{d}\binom{2d}{d}d\log d)$. We establish analogous results for the case where the center is defined as an arbitrary point in the interior of $K$. We also show that, in the latter case, one can always find families of at least $\unicode[STIX]{x1D6FA}((2/\sqrt{3})^{d})$ translates of $K$ with the above property.

For α≥0 and β≥0 we denote by K (α, β) the Kaplan classes of functions f analytic and non-zero in the open unit disk U = {z: |z| < 1} such that f ∈ K(α, β), if, and only if, for θ1 < θ2 < θ1 + 2π and 0 < r < 1,

Where

Assuming the Riemann Hypothesis, Soundararajan [Ann. of Math. (2) 170 (2009), 981–993] showed that $\int _{0}^{T}|\unicode[STIX]{x1D701}(1/2+\text{i}t)|^{2k}\ll T(\log T)^{k^{2}+\unicode[STIX]{x1D716}}$. His method was used by Chandee [Q. J. Math.62 (2011), 545–572] to obtain upper bounds for shifted moments of the Riemann Zeta function. Building on these ideas of Chandee and Soundararajan, we obtain, conditionally, upper bounds for shifted moments of Dirichlet $L$-functions which allow us to derive upper bounds for moments of theta functions.

We show that for $p\geqslant 1$, the $p$th moment of suprema of linear combinations of independent centered random variables are comparable with the sum of the first moment and the weak $p$th moment provided that $2q$th and $q$th integral moments of these variables are comparable for all $q\geqslant 2$. The latest condition turns out to be necessary in the independent and identically distributed case.

It is shown that the discrete fractional Fourier transform recovers the continuum fractional Fourier transform via a limiting process whereby inner products are preserved.