Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!utcsri!arvind
From: arvind@utcsri.UUCP
Newsgroups: ut.theory
Subject: THEORY NET: Counting Perfect Powers
Message-ID: <5290@utcsri.UUCP>
Date: Fri, 21-Aug-87 12:17:01 EDT
Article-I.D.: utcsri.5290
Posted: Fri Aug 21 12:17:01 1987
Date-Received: Sat, 22-Aug-87 18:04:04 EDT
Distribution: ut
Organization: CSRI, University of Toronto
Lines: 14

Date: 21 Aug 1987 09:24:06-EDT (Friday)
From: "Victor S. Miller" <VICTOR@yktvmx.bitnet>
Subject: Counting perfect powers

In the interval $I_x = [x-\sqrt{x},x+\sqrt{x}] I would like to estimate
the number of perfect powers: i.e. the number of integers of the form
$n^r$ with r>1.  Now it is easy to see that for each r>1 there is at most
1 integer of the form $n^r$ in the interval.  That crude estimate
gives an upper bound of $\log x / \log 2$.  If you work a little harder
you can get that down to $(1+\epsilon)\log x / \log \log \log x$. Namely,
count all the exponents up to $\log x / \log \log \log x$.  If r is
bigger than that, then you must have $n<=\log \log x$.  What is the
true asymptotic?
                      Victor Miller -- IBM, Research