home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!caen!zaphod.mps.ohio-state.edu!not-for-mail
- From: edgar@function.mps.ohio-state.edu (Gerald Edgar)
- Newsgroups: sci.math.symbolic
- Subject: Re: New class of functions found, impact of CAS on math
- Date: 23 Jul 1992 12:32:35 -0400
- Organization: The Ohio State University, Dept. of Math.
- Lines: 75
- Distribution: usa
- Message-ID: <14mmv3INN4gq@function.mps.ohio-state.edu>
- References: <2350@nic.cerf.net> <1992Jul23.152053.17774@news.eng.convex.com>
- NNTP-Posting-Host: function.mps.ohio-state.edu
-
- >In article <2350@nic.cerf.net> jcbhrb@nic.cerf.net (Jacob Hirbawi) writes:
-
- >>Very interesting article! I'm not sure I beleive his claim: "I've shown that
- >>problems of divergence can be banished from math applications"; but the
- >>example about the "companion" function to exp(x) is worth a closer look.
- >>You have to admit that you don't expect this function to converge quickly
- >>(if at all):
- >>
- >> B(x) = 1 + x + 2!x^2 + 3!x^3 + 4!x^4 + ... + n!x^n + ...
- >>
-
- In article <1992Jul23.152053.17774@news.eng.convex.com> jchrist@convex.com (Jeff Christenson) writes:
- >Well, I'm only interested in this because it showed up in EE Times, and
- >I'm only in sci.math.symbolic because that's where someone mentioned the
- >problem; but could someone tell me what I'm missing about this.
- >
- >The equation is accompanied by a graph that I assume is showing B(x) for
- >various values of x. On the right side of the vertical axis (positive values
- >of x, right?), the line rises until about where x=1, then dips towards about
- >-0.8 around x=90; then sorta heads back towards 0, but does not cross it.
- >(nice technical description, huh?).
- >
- >What I don't understand is why B(x) is ever negative when x is positive.
- >All of the terms are added, and n! seems like it would always be positive
- >for n>0, and x^n seems like it would always be positive for x>0 and n>0.
- >
- >Do n! and x^n do wierd things as n approaches infinity? Or is that the whole
- >reason for the symbolic algebra and terminal summation?
- >
-
- The series diverges for all nonzero x. So the sense in which the
- series represents the function is only in terms of "summation methods"
- for divergent series. This function, used by Euler, can be
- written as
-
- exp(-1/x)*Ei(1/x)/x,
-
- where Ei is the exponential integral. It has a removable discontinuity
- at zero, and its Taylor series is, indeed, the one given above.
-
- An ASCII version of the graph looks like this: [drawn by Maple]
- Near x=0 it does turn around and arrive at y=1.
-
- | A
- | A
- + 0.4
- | A
- | A
- | A
- | AA
- + 0.2A
- | A
- | A
- | A x
- +-0----*-----+------------+------------+------------+------------+------------+
- 0 AA 5 10 15 20 25 30
- | AAA AAAAAAAAAAAAAAAAAAAAAAAAAA
- | AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
- + -0.2
- |
- |
- |
- |
- + -0.4
- |
- |
-
- The minumum seems to be at about x=8.
-
-
- --
- Gerald A. Edgar Internet: edgar@mps.ohio-state.edu
- Department of Mathematics Bitnet: EDGAR@OHSTPY
- The Ohio State University telephone: 614-292-0395 (Office)
- Columbus, OH 43210 -292-4975 (Math. Dept.) -292-1479 (Dept. Fax)
-