1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
|
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html>
<!-- Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016 The GSL Team.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License, Version 1.3 or
any later version published by the Free Software Foundation; with the
Invariant Sections being "GNU General Public License" and "Free Software
Needs Free Documentation", the Front-Cover text being "A GNU Manual",
and with the Back-Cover Text being (a) (see below). A copy of the
license is included in the section entitled "GNU Free Documentation
License".
(a) The Back-Cover Text is: "You have the freedom to copy and modify this
GNU Manual." -->
<!-- Created by GNU Texinfo 5.1, http://www.gnu.org/software/texinfo/ -->
<head>
<title>GNU Scientific Library – Reference Manual: Fitting Overview</title>
<meta name="description" content="GNU Scientific Library – Reference Manual: Fitting Overview">
<meta name="keywords" content="GNU Scientific Library – Reference Manual: Fitting Overview">
<meta name="resource-type" content="document">
<meta name="distribution" content="global">
<meta name="Generator" content="makeinfo">
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<link href="index.html#Top" rel="start" title="Top">
<link href="Function-Index.html#Function-Index" rel="index" title="Function Index">
<link href="Least_002dSquares-Fitting.html#Least_002dSquares-Fitting" rel="up" title="Least-Squares Fitting">
<link href="Linear-regression.html#Linear-regression" rel="next" title="Linear regression">
<link href="Least_002dSquares-Fitting.html#Least_002dSquares-Fitting" rel="previous" title="Least-Squares Fitting">
<style type="text/css">
<!--
a.summary-letter {text-decoration: none}
blockquote.smallquotation {font-size: smaller}
div.display {margin-left: 3.2em}
div.example {margin-left: 3.2em}
div.indentedblock {margin-left: 3.2em}
div.lisp {margin-left: 3.2em}
div.smalldisplay {margin-left: 3.2em}
div.smallexample {margin-left: 3.2em}
div.smallindentedblock {margin-left: 3.2em; font-size: smaller}
div.smalllisp {margin-left: 3.2em}
kbd {font-style:oblique}
pre.display {font-family: inherit}
pre.format {font-family: inherit}
pre.menu-comment {font-family: serif}
pre.menu-preformatted {font-family: serif}
pre.smalldisplay {font-family: inherit; font-size: smaller}
pre.smallexample {font-size: smaller}
pre.smallformat {font-family: inherit; font-size: smaller}
pre.smalllisp {font-size: smaller}
span.nocodebreak {white-space:nowrap}
span.nolinebreak {white-space:nowrap}
span.roman {font-family:serif; font-weight:normal}
span.sansserif {font-family:sans-serif; font-weight:normal}
ul.no-bullet {list-style: none}
-->
</style>
</head>
<body lang="en" bgcolor="#FFFFFF" text="#000000" link="#0000FF" vlink="#800080" alink="#FF0000">
<a name="Fitting-Overview"></a>
<div class="header">
<p>
Next: <a href="Linear-regression.html#Linear-regression" accesskey="n" rel="next">Linear regression</a>, Up: <a href="Least_002dSquares-Fitting.html#Least_002dSquares-Fitting" accesskey="u" rel="up">Least-Squares Fitting</a> [<a href="Function-Index.html#Function-Index" title="Index" rel="index">Index</a>]</p>
</div>
<hr>
<a name="Overview-4"></a>
<h3 class="section">38.1 Overview</h3>
<p>Least-squares fits are found by minimizing <em>\chi^2</em>
(chi-squared), the weighted sum of squared residuals over <em>n</em>
experimental datapoints <em>(x_i, y_i)</em> for the model <em>Y(c,x)</em>,
</p>
<div class="example">
<pre class="example">\chi^2 = \sum_i w_i (y_i - Y(c, x_i))^2
</pre></div>
<p>The <em>p</em> parameters of the model are <em>c = {c_0, c_1, …}</em>. The
weight factors <em>w_i</em> are given by <em>w_i = 1/\sigma_i^2</em>,
where <em>\sigma_i</em> is the experimental error on the data-point
<em>y_i</em>. The errors are assumed to be
Gaussian and uncorrelated.
For unweighted data the chi-squared sum is computed without any weight factors.
</p>
<p>The fitting routines return the best-fit parameters <em>c</em> and their
<em>p \times p</em> covariance matrix. The covariance matrix measures the
statistical errors on the best-fit parameters resulting from the
errors on the data, <em>\sigma_i</em>, and is defined
<a name="index-covariance-matrix_002c-linear-fits"></a>
as <em>C_{ab} = <\delta c_a \delta c_b></em> where <em>< ></em> denotes an average over the Gaussian error distributions of the underlying datapoints.
</p>
<p>The covariance matrix is calculated by error propagation from the data
errors <em>\sigma_i</em>. The change in a fitted parameter <em>\delta
c_a</em> caused by a small change in the data <em>\delta y_i</em> is given
by
</p>
<div class="example">
<pre class="example">\delta c_a = \sum_i (dc_a/dy_i) \delta y_i
</pre></div>
<p>allowing the covariance matrix to be written in terms of the errors on the data,
</p>
<div class="example">
<pre class="example">C_{ab} = \sum_{i,j} (dc_a/dy_i) (dc_b/dy_j) <\delta y_i \delta y_j>
</pre></div>
<p>For uncorrelated data the fluctuations of the underlying datapoints satisfy
<em><\delta y_i \delta y_j> = \sigma_i^2 \delta_{ij}</em>, giving a
corresponding parameter covariance matrix of
</p>
<div class="example">
<pre class="example">C_{ab} = \sum_i (1/w_i) (dc_a/dy_i) (dc_b/dy_i)
</pre></div>
<p>When computing the covariance matrix for unweighted data, i.e. data with unknown errors,
the weight factors <em>w_i</em> in this sum are replaced by the single estimate <em>w =
1/\sigma^2</em>, where <em>\sigma^2</em> is the computed variance of the
residuals about the best-fit model, <em>\sigma^2 = \sum (y_i - Y(c,x_i))^2 / (n-p)</em>.
This is referred to as the <em>variance-covariance matrix</em>.
<a name="index-variance_002dcovariance-matrix_002c-linear-fits"></a>
</p>
<p>The standard deviations of the best-fit parameters are given by the
square root of the corresponding diagonal elements of
the covariance matrix, <em>\sigma_{c_a} = \sqrt{C_{aa}}</em>.
The correlation coefficient of the fit parameters <em>c_a</em> and <em>c_b</em>
is given by <em>\rho_{ab} = C_{ab} / \sqrt{C_{aa} C_{bb}}</em>.
</p>
<hr>
<div class="header">
<p>
Next: <a href="Linear-regression.html#Linear-regression" accesskey="n" rel="next">Linear regression</a>, Up: <a href="Least_002dSquares-Fitting.html#Least_002dSquares-Fitting" accesskey="u" rel="up">Least-Squares Fitting</a> [<a href="Function-Index.html#Function-Index" title="Index" rel="index">Index</a>]</p>
</div>
</body>
</html>
|
