From - Sun Aug 6 01:14:00 2000 Path: rQdQ!sn-xit-03!supernews.com!nntp.primenet.com!nntp.gblx.net!europa.netcrusader.net!204.71.34.3!newsfeed.cwix.com!sjc-peer.news.verio.net!news.verio.net!nntp2.cerf.net!ns1.mathforum.com!forum.mathforum.com!gateway From: XXXakoro@hotpop.comXXX (Andres Koropecki) Newsgroups: sci.math Subject: Re: IMO problem 2 Date: 6 Aug 2000 00:25:46 -0400 Organization: The Math Forum Lines: 33 Sender: daemon@forum.mathforum.com Message-ID: References: NNTP-Posting-Host: forum.mathforum.com X-Unparseable-Date: 5 Xref: rQdQ sci.math:395308 On 4 Aug 00 11:28:44, Paul Harvey wrote: >Does there exist an elegant solution to IMO #2, that is >Let abc = 1, a,b,c are +ve reals >(a-1+1/b)(b-1+1/c)(c-1+1/a)<=1 This proof is probably less elegant that P. Montgomery's but it's a different one (and it can probably be simplified too): If one or three of a+1/b-1, b+1/c-1, c+1/a-1 are negative, the inequality trivially holds. If two are negative, say b+1/c<1 and c+1/a<1, then 2+b+1/a <= b+1/c+c+1/a <2. Hence b+1/a < 0, absurd. Suppose all the three multiplicands are positive. (**) In the following, we use the facts that abc=1, a=1/bc and such. (a+1/b-1)(b+1/c-1)(c+1/a-1) <= 1 iff [(a+1/b-1)/a][(b+1/c-1)/b][(c+1/a-1)/c] <= 1 iff (1+c-1/a)(1+a-1/b)(1+b-1/c) <= 1 (***_2) iff [(1+c-1/a)/c][(1+a-1/b)/a][(1+b-1/c)/b] <= 1 iff (1+1/c-b)(1+1/a-c)(1+1/b-a) <= 1 (***_1). Suppose 1/c + 1/a + 1/b - a - b - c <= 0 (*) Since the multiplicands are positive (by (**)) we can use the arithmetic-geometric means inequality which implies that (1+1/c-b)(1+1/a-c)(1+1/b-a) <= [(3 + 1/c + 1/a + 1/b - a - b - c)/3]^2 and we know that 3 + 1/c + 1/a + 1/b - a - b - c <= 3 by (*). Therefore, [(3 + 1/c + 1/a + 1/b - a - b - c)/3]^2 <= 1 and (***_1) holds. If (*) does not hold, we apply a similar argument to (***_2): The arithmetic-geometric means inequality impies that (1+c-1/a)(1+a-1/b)(1+b-1/c) <= [(3 + a + b + c - 1/a - 1/b - 1/c)/3]^3 But we have a + b + c - 1/a - 1/b - 1/c <= 0 so that (1+c-1/a)(1+a-1/b)(1+b-1/c) <= (3/3)^3 = 1 and (***_2) holds. Q.E.D.