1
<?xml version="1.0" encoding="ISO-8859-1" standalone="no"?>
2
<!DOCTYPE MAN SYSTEM "../../manrev.dtd">
4
<LANGUAGE>eng</LANGUAGE>
6
<TYPE>Scilab Function</TYPE>
7
<DATE>April 1993</DATE>
8
<SHORT_DESCRIPTION name="svd"> singular value decomposition</SHORT_DESCRIPTION>
10
<CALLING_SEQUENCE_ITEM>s=svd(X) </CALLING_SEQUENCE_ITEM>
11
<CALLING_SEQUENCE_ITEM>[U,S,V]=svd(X) </CALLING_SEQUENCE_ITEM>
12
<CALLING_SEQUENCE_ITEM>[U,S,V]=svd(X,0) (obsolete) </CALLING_SEQUENCE_ITEM>
13
<CALLING_SEQUENCE_ITEM>[U,S,V]=svd(X,"e") </CALLING_SEQUENCE_ITEM>
14
<CALLING_SEQUENCE_ITEM>[U,S,V,rk]=svd(X [,tol]) </CALLING_SEQUENCE_ITEM>
19
<PARAM_NAME>X</PARAM_NAME>
21
<SP>: a real or complex matrix</SP>
25
<PARAM_NAME>s</PARAM_NAME>
27
<SP>: real vector (singular values)</SP>
31
<PARAM_NAME>S</PARAM_NAME>
33
<SP>: real diagonal matrix (singular values)</SP>
37
<PARAM_NAME>U,V</PARAM_NAME>
39
<SP>: orthogonal or unitary square matrices (singular vectors).</SP>
43
<PARAM_NAME>tol</PARAM_NAME>
45
<SP>: real number</SP>
51
<P><VERB>[U,S,V] = svd(X)</VERB> produces a diagonal matrix
52
<VERB>S</VERB> , of the same dimension as <VERB>X</VERB> and with
53
nonnegative diagonal elements in decreasing order, and unitary
54
matrices <VERB>U</VERB> and <VERB>V</VERB> so that <VERB>X =
56
<P><VERB>[U,S,V] = svd(X,0)</VERB> produces the "economy
57
size" decomposition. If <VERB>X</VERB> is m-by-n with m >
58
n, then only the first n columns of <VERB>U</VERB> are computed
59
and <VERB>S</VERB> is n-by-n.</P>
60
<P><VERB>s = svd(X)</VERB> by itself, returns a vector <VERB>s</VERB>
61
containing the singular values.</P>
62
<P><VERB>[U,S,V,rk]=svd(X,tol)</VERB> gives in addition <VERB>rk</VERB>, the numerical rank of <VERB>X</VERB> i.e. the number of
63
singular values larger than <VERB>tol</VERB>.</P>
65
The default value of <VERB>tol</VERB> is the same as in <VERB>rank</VERB>.</P>
96
svd decompositions are based on the Lapack routines DGESVD for
97
real matrices and ZGESVD for the complex case.</P>