PROCEEDINGS OF THE
AMERICAN MATHEMATICAL SOCIETY
Volume 45, Number 2, August, 1974
ZERO-ONE LAWS FOR STABLE MEASURES
R. M. DUDLEY1 AND MAREK KANTER
ABSTRACT. For any stable measure p on a vector space, every
measurable linear subspace has measure 0 or 1.
1. Introduction. It is known that for any Gaussian probability measure,
a linear subspace has measure 0 or 1. This result has been extended to
additive subgroups by Kallianpur [2]. Here we extend the zero-one law in a
different direction, replacing "Gaussian" by "stable". We begin with some
definitions.
Definition. Let S be a vector space over R and let S be a (7-algebra
of subsets of S. We call (S, S) a measurable vector space iff both the follow-
ing hold:
(a) addition is jointly measurable from S x S into S,
(b) scalar multiplication is jointly measurable from R x S into S, fot
completed Lebesgue measure À on R.
Let S be a topological vector space and let J be the ff-algebra of
Borel sets (generated by the open sets). Then if S is metrizable and sep-
arable, (S, J ) is a measurable vector space, but it need not be so in general.
If (S, S) is a measurable vector space and p and v are finite, counta-
bly additive measures on S, then we have the convolution p * v defined as
usual by
(u*v)(A)= (px v)\{x, y): x + y e A\.
For any (5-valued) random-variable Z, let its probability distribution
(law), defined on o, be denoted by 5l(Z).
Given any vector space S and c £ R, let 772 (x) = ex fot all x £ S, and
0s(x) = x + s for any s £ S.
Definition. Given a measurable vector space (5, o), a probability
Received by the editors October 9, 1973.
AMS (MOS) subject classifications (1970). Primary 60B15; Secondary 60E05,
60 F 20, 28A40.
Key words and phrases. Stable measure, strictly stable, zero-one law, measur-
able vector space.
This research was partially supported by National Science Foundation
Grant GP-29072.
Copyright © 1974, American Mathematical Society
245
License or copyright restrictions may apply to redistribution; see http://www.ams.org/journal-terms-of-use