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
|
<HTML>
<HEAD>
<TITLE>TSP (libtsp/SP) - SPcovar</TITLE>
</HEAD>
<BODY BGCOLOR="#FFFACD">
<H2>SPcovar</H2>
<HR>
<H4>Routine</H4>
<DL>
<DT>
void SPcovar (x, Nx, Cov, Np)
</DL>
<H4>Purpose</H4>
<DL>
<DT>
Calculate a correlation (covariance) matrix for a data sequence
</DL>
<H4>Description</H4>
This routine fills in the elements of a correlation matrix given an array of
data. The elements of the matrix are calculated as
<PRE>
N+Np-1
C(i,j) = SUM x(n-i) x(n-j) , 0 <= i,j <= Np
n=Np
</PRE>
where Np is the number of correlation lags and N is the number of terms in
each correlation calculation. To compute the correlation terms, the
following recursion is used,
<P>
<PRE>
C(i,j) = C(i-1,j-1) + x(Np-i) x(Np-j) - x(N+Np-i) x(N+Np-j) ,
</PRE>
<P>
starting from C(i,0)=C(0,i). Note that finite precision effects in
calculating the recursion can result in a non-positive definite matrix.
<H4>Parameters</H4>
<DL>
<DT>
-> const float x[]
<DD>
Input data (Nx elements). The first Np values normally represent data
from a previous frame of data.
<DT>
-> int Nx
<DD>
Number of data elements. Nx=N+Np, where N is the number of terms in
each correlation sum and Np is the number of correlation lags to be
calculated. Note that Nx must be larger than Np.
<DT>
<- float *Cov[]
<DD>
Cov is an array of pointers to the rows of a Np+1 by Np+1 matrix. On
return, the (symmetric) matrix contains the correlation values.
<DT>
-> int Np
<DD>
Maximum correlation lag
</DL>
<H4>Author / revision</H4>
P. Kabal Copyright (C) 1997
/ Revision 1.7 1997/10/10
<P>
<HR>
Main Index <A HREF="../libtsp.html">libtsp</A>
</BODY>
</HTML>
|
