From: Paul
Date: Fri, 05 Jan 2001 10:46:30 +0000
Subject: Re: [TSSP] Virtual Secondary Database
Terry,
I cobbled together a little curve fitter - gradient descent with
a dash of monte carlo, and left it running on a spare PC overnight.
So far its come up with
y = m * x + c
where
m = 0.73261
c = 0.6040
and
x = (k1 + hd)^k2 *
(k3 + b)^k4 *
(k5 + td)^k6 *
(k7 + awg)^k8 *
(k9 + h)^k10 *
(k11 + sr)^k12 *
(k13 + tb)^k14
in which
k1 = 0.5836
k2 = -0.4048
k3 = 0.0423
k4 = 0.1519
k5 = -0.9487
k6 = 0.1143
k7 = 3.1508
k8 = 0.0096
k9 = 0.0721
k10 = 0.0039
k11 = 0.6325
k12 = -0.0070
k13 = 0.0793
k14 = 0.0259
and the variables hd,b,td,awg,h,sr,tb are the seven configuration
space axes, with the modification:
if (td is zero) then set td = 1.0 and set tb = 0.0
The computed value y is approximately equal to Lfac
where Lfac = les/ldc, to within an RMS deviation of 1.4% over
1200 vsd records.
Cheers,
--
Paul Nicholson,
Manchester, UK.
--
Maintainer Paul Nicholson, paul@abelian.demon.co.uk.