Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!lsuc!sickkids!gw
From: gw@sickkids.UUCP (CFI/Graham Wilson )
Newsgroups: comp.sources.wanted,sci.math,sci.math.stat
Subject: Need Smooth Spline & Axes Generation Code
Message-ID: <65@sickkids.UUCP>
Date: Thu, 30-Jul-87 12:01:20 EDT
Article-I.D.: sickkids.65
Posted: Thu Jul 30 12:01:20 1987
Date-Received: Sat, 1-Aug-87 08:57:46 EDT
Reply-To: gw@sickkids.UUCP (CFI/Graham Wilson (977-5450) [RI7921Z])
Organization: Hospital for Sick Children, Toronto
Lines: 31
Xref: mnetor comp.sources.wanted:1771 sci.math:1702 sci.math.stat:165

Wanted:   i) Code for Smooth Splines
         ii) Code for Axes Generation

Hello.

I am looking for algorithms and/or code for producing smooth cubic splines.
The problem is defined as follows:

Given a series of (non-uniformly spaced) data points <Xi,Yi> (i=0...n) and
error values <dYi> (i = 0...n) along with a smoothing factor S >= 0, I need
an algorithm to produce a smooth spline g() such that for each i=0...n, it
holds that
			|g(Xi) - Yi| <= S*dYi

(i.e., S*dYi is the maximum that the curve can deviate from Yi - hence if
S == 0 or dYi == 0 for i=0...n, then a natural cubic spline is generated).
Smooth splines are used for weeding out outlier points.  I need them to
produce sigmodial curves (no max or min, max one point of inflection).
I have some old fortran code which seems to work for uniformly-spaced data,
but produces roller-coasters for non-uniformly spaces data.

I am also looking for algorithms and/or code to generated axes where
the axes are either linear/linear, linear/log, log/linear, or log/log.

Great thanks in advance.

Graham Wilson              |      ...allegra   ...decvax
CyberFluor Inc             |            \            \  
179 John St., Suite 400    | ...linus!utcrsi!utzoo!sickkids!cfi!graham
Toronto, Ontario, M5T 4B1  |            /            /
(416) 977-5450             |      ...watmath   ...ihnp4