Moser and Abramson [4] proved: given m≥2 and 0≤kl, k2, …, kp<m, the number of p-combinations

satisfying

is

1

([x] denotes the greatest integer ≤x).

This note is concerned with the implication SepII(Q)→SepI(Q) where Q is a class of subsets of some set S.

where cZ denotes S—Z.

It is well-known that in general the above implication is false (e.g. let Q be the closed subsets of the reals).

In this note we consider the class of topological spaces whose every Hausdorff quotient is regular (or equivalently, normal). Sufficient conditions in order that a space belongs to this class have appeared in papers by Morita [3], and more recently by MacDonald and Willard [2]. In both cases, the authors impose conditions implying paracompactness. We shall derive additional sufficient conditions here which show that paracompactness is not necessary in general.

A bounded basis {xn} of a Banach space X is called perfectly homogeneous if every bounded block basic sequence {yn} of {xn} is equivalent to {xn}. By a result of M. Zippin [4], a basis in a Banach space is perfectly homogeneous if and only if it is equivalent to the unit vector basis of c0 or lp, 1 ≤ p < + ∞. A basis {xn} of a Banach space X is called symmetric, if every permutation {xσ(n)} of {xn} is a basis of X, equivalent to the basis {xn}. It is clear that every perfectly homogeneous basis is symmetric.

A simpler proof of the theorem on the spinor structure of space-time is given. Some geometrical insights are provided.

It is well known [1,3] that a commutative semigroup (S, +) can be embedded in a semigroup which is a union of groups if and only if S is separative (2a = a + b = 2b implies a = b). We extend this result to additively commutative semirings.

A semiring (S, +, ⋅) is a set S with associative addition (+) and multiplication (⋅), the latter distributing over addition from left and right. In what follows (S, +, ⋅) will denote a semiring in which the additive semigroup (S, +) is commutative. An element 0 can be adjoined, where s = s + 0, 0 = 0 ⋅ s = s ⋅ 0 for all s in S, to form S°.

The coefficients of the two series

(1)

are recursively given by Nielsen [1]: c0=g0=1 and

where s1 is the Euler constant γ and for n>1 sn = ζ(n).

Let A be an associative, but not necessarily commutative, ring with identity, and J = J(A) its Jacobson radical. A (unital) module is regular iff every submodule is pure (see (1)). The regular socle R(M) of a module M is the sum of all its submodules which are regular. These concepts have been introduced and studied in (2).

A theorem of M. Katětov asserts that a bounded uniformly continuous function f on a subspace Q of a uniform space P has a bounded uniformly continuous extension to all of P. In this note we give new proofs of two special cases of this theorem: (i) Q is totally bounded, and (ii) P is a locally compact group and Q is a subgroup, both P and Q having the left uniformity.

Il me fait un très grand plaisir de vous présenter aujourd'hui la conférence Jeffrey-Williams nommée ainsi en l'honeur des distingués fondateurs de la Société Mathématique du Canada.

Il y a beaucoup d'évidences indirectes d'un vif échange d'idées entre Herman Minkowski et David Hilbert, pendant la période de temps entre la soutenance de thèse de Hilbert (décembre 1884) et la mort prématurée de Minkowski (le 12 janvier 1909).

Given a matrix A = (ank) (n, k = 0, 1, 2, …), let (A) denote the set of all sequences x = {xk} such that {An(x)} ∊ c where An(x) = Σk=0∞ankxk (n≥0) and c denotes the set of all convergent sequences. It is well known (see e.g. Zeller [6] or Zeller and Beekmann [7], p. 48) that given an unbounded sequence x, there exists a regular (=permanent) matrix A with ank = 0 for k > n (and indeed with ann ≠ 0) such that (A) = c ⊕x, the linear space spanned by c and x. We call A an Einfolgenverfahren. (See [7].) In [4] Rhoades considered, inconclusively, the question whether there exists a Hausdorff matrix H such that (H)= c ⊕ x (for arbitrary unbounded sequence x).

Let Y be a linear space over the complex plane C, and let F be a mapping on the complex linear space Y ⊕ C into subsets of C with the following properties: for y ∊ Y, λ and μ ∊ C, F(y + μ) is a nonempty and bounded subset of C, F(λy + μ) = λF(y) + μ and F(μ) = {μ}. We shall write f(y + μ) = sup{|λ + μ|: λ ∊ F(y)}, the radius of F(y + μ), y ∊ Y and μ ∊ C. The convex hull (resp. the closure) of a subset M of C is denoted by conv M (resp. ).

The purpose of this note is to give a short proof of a generalisation of a theorem of Ryser, Theorem 10.2.3 of [1], concerning matrices that occur in the theory of symmetric block designs.

The two main results of matrix theory required in the proof given below are:

1. (1) If B, C are square matrices such that BC = zI where z is a non-zero complex number, then CB = zI

2. (2) A matrix S which is both symmetric (i.e. S' = S) and skew-symmetric (i.e. S' = —S)is zero.

In a recent article [1] vacuum field solutions of the Brans-Dicke [2] field equations were found, the space-time metric in each solution being of the Friedmann type. Most of these solutions existed only for specific values of the parameter ω and, in particular, the two largest sets of solutions corresponded to the values and . Peters [3, 4] has shown that when all solutions of the Brans-Dicke vacuum equations are conformai to space-times with vanishing Ricci tensor. The purpose of this note is to investigate the possible geometric consequences of the value .

Problems involving Egyptian fractions (rationals whose numerator is 1 and whose denominator is a positive integer) have been extensively studied. (See [1] for a more complete bibliography). Some of the most interesting questions, many still unsolved, concern the solvability of

where k is fixed.

In [2] Rav proved necessary and sufficient conditions for the solvabilty of the above equation, as a consequence of some other theorems which are rather complicated in their proofs. In this note we give a short, elementary proof of this theorem, and at the same time generalize it slightly.