From mboxrd@z Thu Jan 1 00:00:00 1970 X-Msuck: nntp://news.gmane.io/gmane.science.mathematics.categories/1320 Path: news.gmane.org!not-for-mail From: Peter Freyd Newsgroups: gmane.science.mathematics.categories Subject: Derivatives Date: Mon, 3 Jan 2000 10:28:18 -0500 (EST) Message-ID: <200001031528.KAA12024@saul.cis.upenn.edu> NNTP-Posting-Host: main.gmane.org X-Trace: ger.gmane.org 1241017748 30709 80.91.229.2 (29 Apr 2009 15:09:08 GMT) X-Complaints-To: usenet@ger.gmane.org NNTP-Posting-Date: Wed, 29 Apr 2009 15:09:08 +0000 (UTC) To: categories@mta.ca Original-X-From: rrosebru@mta.ca Mon Jan 3 17:24:00 2000 -0400 Original-Received: (from Majordom@localhost) by mailserv.mta.ca (8.9.3/8.9.3) id QAA32429 for categories-list; Mon, 3 Jan 2000 16:18:19 -0400 (AST) X-Authentication-Warning: mailserv.mta.ca: Majordom set sender to cat-dist@mta.ca using -f Original-Sender: cat-dist@mta.ca Precedence: bulk Original-Lines: 36 Xref: news.gmane.org gmane.science.mathematics.categories:1320 Archived-At: Given a center-preserving function between intervals, f, is there a natural way to obtain from it one that's midpoint-preserving? Part of preserving midpoints is preserving halving: f(hx) = hf(x). We can rewrite that to convert from a commutativity condition to a fixed- point condition: f = \x.2f(hx) (the fixed-pont of an operation that seems to be innate to every graphing calculator). The process needn't converge, indeed, one will often need to shrink the domain of the function down to a smaller concentric interval and it can converge to fixed-points not to our liking (such as \x.x*sin(2pi*log|x|/log2).) But the closed interval is the final coalgebra not just for the ordered-wedge functor X v X but for for the ordered-wedge of any positive number of copies of X. (There's a general theorem that says that a final coalgebra for X v X is automatically a final coalgebra for X v X v X v X, indeed, for any such ordered-wedge where the number of copies of X is a power of two. I haven't been able to find the proof for a general theorem that specializes to X v X v X.) Using that the closed interval is the final coalgebra for X v X v X we can define the thirding map t:I --> I in a manner similar to (and simpler than) the definition of the halving map. (Yes, the OED lists "third" as a verb) So we are also looking for solutions to f = \x.3f(tx). We obtain a commutative monoid of operators of the form \fx.nf(x/n). The process of applying a one-generator free monoid of operators easily generalizes to an arbitrary commutative monoid of operators. Starting with a center-preserving f between intervals the process needn't converge even if we restrict the domain to a smaller concentric interval. But if it does converge to the germ of a monotonic duality-preserving function then it converges to a germ of a midpoint-preserving function from the original source interval to the original target interval. And if these intervals should coincide, it converges to what in my last posting I defined as an element of the reals.