In recent years, the problem of embedding the projective spaces in Euclidean spaces was studied very much, by different methods. Usually, the negative results on the embedding problem are proved by using suitable homotopy invariants. The best known example of such homotopy invariants is given by the Stiefel–Whitney classes.

In an earlier paper (3), polycyclic groups in which every subgroup can be generated by d, or fewer, elements were studied. In this paper we investigate the structure of those polycyclic groups G such that every abelian normal subgroup of F(G), the Fitting subgroup of G, can begenerated by at most d elements.

The purpose of this note is to describe some algebraic conditions on a Banach algebra which force it to be finite dimensional. We shall assume throughout that we are dealing with Banach algebras over the field of complex numbers, C.

In this note we consider the question: If R is a right Noetherian ring and I is an invertible ideal of R, how do the Krull dimensions of various modules, factor rings and over-rings of R, connected with I, compare with the Krull dimension of R? This question is prompted by results in (5) and (6). In comparing the Krull dimension of the ring R with that of the ring R/I, the best result would be that the Krull dimension of the ring R is exactly one greater than that of the ring R/I. This result is not true in general; however, we see, in Theorem 2.4, that if the invertible ideal is contained in the Jacobson radical the result holds. In the general case we find it is necessary to introduce an over-ring T of R generated by the inverse I−1 of I. We then see that the Krull dimension of R is the larger of two possibilities: (a) Krull dimension of R/I plus one or (b) Krull dimension of T. In order to prove this result we construct a strictly increasing map from the poset of right ideals of R to the cartesian product of the poset of right ideals of T with a poset of certain infinite sequences of right ideals of R/I.

To establish our notation N will always denote a (left) near-ring without any type of multiplicative identity (unless the contrary is stated) satisfying On = 0 for each n ∈ N where 0 is the additive identity of N. A group M, written additively, which admits N as a set of right multipliers is a (right) N-module if a ∈ M, n1, n2 ∈ N implies a(n1 + n2) = an1 + an2 and a(n1n2) = (an1)n2. When N has a two-sided identity, 1, we suppose that a 1 = a for each a ∈ M. A subgroup X of M is an N-subgroup of M if it is an N-module; X is a submodule of M if it is a normal subgroup of M and a ∈ M, x ∈ X. n ∈ N implies (a + x)n – an ∈ X. We denote by SL(M) the set of N-subgroups and by L(M) the set of submodules of M. Since N may be regarded as an N-module we can talk about N-subgroups and submodules of N although we usually call the submodules of N right ideals of N. Other definitions can be found in (6).

In this note we point out that certain algebraic-topological constructions are particular cases of one construction, namely double-negation sheafification. The principal cases we have in mind are concerned with booleanpowers, completions of boolean algebras, and maximal rings of quotients.We conjecture that several other constructions—particularly completion-type constructions—will turn out also to be examples of double-negation sheafification.

A semigroup S with identity is (left) perfect if every unitary left S-system has a projective cover. This is the semigroup analogue of the definition of left perfect rings introduced in (1). The investigation of perfect semigroups was initiated by Isbell (4), who proved that a semigroup is perfect if and only if it satisfies two conditions referred to as conditions A and D.

In this paper we study a property which we call residual -commutativity. This idea was kindled by a paper of Ayoub (1) and one of Stanley (6). Durbin has defined a similar property in (2) which has also been studied by Slotterbeck in (4). Durbin's property implies residual -commutativity but we have not been able to decide if they are equivalent. However, we have shown that they coincide in certain circumstances. The notation, unless otherwise stated, is that of Robinson (3).

This note extends to categories Fountain's theorem (2) that for a perfect monoid S, every flat S-set is projective. (The converse is known (4).)

We firstly introduce some notation. Let a(t) ∈ Rn then we denote by α(t) a rearrangement of a(t) in non-decreasing order. We write a(1) ≺ a(2) if

and

The generalised Stieltjes transform of a function f(t) ∈ L(0, ∞) is defined by

The integral converges for complex s in the region Ω, where Ω is the s-plane cut from the origin along the negative real axis.

Two n-planes Γ and Δ in real Euclidean r-space Rr are called isoclinic with parameter λ if the angle θ between any x in Γ and its orthogonal projection Px on Δ is unique, with cos2 θ = λ. Let vλ(n, r) denote the maximum number of equi-isoclinic (i.e. pairwise isoclinic) n-planes in Rr with parameter λ.

Let E be a nuclear space provided with a topology different from the weak topology. Let {Ai: i ∈ I} be a fundamental system of equicontinuous subsets of the topological dual E' of E. If {Fi: i ∈ I} is a family of infinite dimensional Banach spaces with separable predual, there is a fundamental system {Bi: i ∈ I} of weakly closed absolutely convex equicontinuous subsets of E'such that is norm-isomorphic to Fi, for each i ∈ I. Other results related with the one above are also given.

Let F be an arbitrary topological vector space; we shall say that a subset S of F is quasi-convex if the set of continuous affine functionals on S separates the points of S. If X is a Banach space and T : X → F is a continuous linear operator, then T is quasi-convex if is quasi-convex, where U is the unit ball of X.

It is a well-known fact that any normed algebra can be represented isometrically as an algebra of operators with the operator norm. As might be expected from the very universality of this property, it is little used in the study of the structure of an algebra. Far more helpful are representations on Hilbert space, though these are correspondingly hard to come by: isometric representations on Hilbert space are not to be expected in general, and even continuous nontrivial representations may fail to exist. The purpose of this paper is to examine a class of representations intermediate in both availability and utility to those already mentioned—namely, representations on reflexive spaces. There certainly are normed algebras which admit isometric representations of the latter type but have not even faithful representations on Hilbert space: the most natural example is the algebra of all continuous linear operators on E where E = lp with 1 < p ≠ 2 < ∞, for Berkson and Porta proved in (2) that if E, F are taken from the spaces lp with 1 < p < ∞ and E ≠ F then the only continuous homomorphism from into is the zero mapping. On the other hand there are also algebras which have no continuous nontrivial representation on any reflexive space—for example the algebra of finite-rank operators on an irreflexive Banach space (see Berkson and Porta (2) or Barnes (1) or Theorem 3, Corollary 1 below).

In this paper an asymptotic formula is obtained for the number of primes representable as the sum of two square-free squares. The precise result is:

Theorem 1.Let N(x) be the number of primes not exceeding x represented by the quadratic formy2 + z2, where y and z are square-free. Let w be a fixed arbitrarily large number. Then

where

and

We assume the reader to be familiar with the basic definitions of near-rings, N-subgroups etc. as presented, for example, in (4). Throughout, N will denote a left near-ring (i.e. a, b, c ∈ N imply a(b + c) = ab + ac) in which 0n = 0 for each n ∈ N. We say that N is strictly semiprime if A2 implies A = (0) where A is an N-subgroup of N. An N-subgroup A is nilpotent if An for some positive integer n and an element a ∈ N is nil if an = 0 for some n. An element a ∈ N is regular if ax = 0 or xa = 0 implies x = 0.

Harnack inequalities are known to be of great importance in the theory of quasilinear elliptic partial differential equations. In the case of such equations defined over a domain Ω in Rn, inequalities of this type have been proved for solutions of second-order equations in divergence form which are of either elliptic or degenerate elliptic structure. More recently Bombieri and Giusti (2) have proved a Harnack inequality for solutions of linear elliptic equations on a minimal surface in Rn+1. The equations are of the form

where summation over i, j = 1, …, n+l is understood, and δ = (δ1, …, δn+1) is the tangential derivative on S. In (2), the inequality is used to give much simplified proofs of some classical results on minimal surfaces, and to generalise some more recent ones.

Let w = w(a0, a1, an–1) be a word in the free group freely generated by a0, a1, …, an–1; let wi, denote the word w(ai, ai+1, …, ai+n–1), where the subscriptsj in aj are reduced modulo n; and let

The well-known algebraic concept of tensor product exists for any variety of algebras.The tensor product of groups and of rings have been studied extensively. For other varieties, such as the variety of semigroups, the tensor product has been investigated more recently (5). In this paper we investigate the tensor product of distributive lattices.