home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
The Datafile PD-CD 1 Issue 2
/
PDCD-1 - Issue 02.iso
/
_utilities
/
utilities
/
001
/
meschach
/
!Meschach
/
c
/
norm
< prev
next >
Wrap
Text File
|
1994-01-13
|
4KB
|
199 lines
/**************************************************************************
**
** Copyright (C) 1993 David E. Steward & Zbigniew Leyk, all rights reserved.
**
** Meschach Library
**
** This Meschach Library is provided "as is" without any express
** or implied warranty of any kind with respect to this software.
** In particular the authors shall not be liable for any direct,
** indirect, special, incidental or consequential damages arising
** in any way from use of the software.
**
** Everyone is granted permission to copy, modify and redistribute this
** Meschach Library, provided:
** 1. All copies contain this copyright notice.
** 2. All modified copies shall carry a notice stating who
** made the last modification and the date of such modification.
** 3. No charge is made for this software or works derived from it.
** This clause shall not be construed as constraining other software
** distributed on the same medium as this software, nor is a
** distribution fee considered a charge.
**
***************************************************************************/
/*
A collection of functions for computing norms: scaled and unscaled
*/
static char rcsid[] = "$Id: norm.c,v 1.6 1994/01/13 05:34:35 des Exp $";
#include <stdio.h>
#include <math.h>
#include "matrix.h"
/* _v_norm1 -- computes (scaled) 1-norms of vectors */
double _v_norm1(x,scale)
VEC *x, *scale;
{
int i, dim;
Real s, sum;
if ( x == (VEC *)NULL )
error(E_NULL,"_v_norm1");
dim = x->dim;
sum = 0.0;
if ( scale == (VEC *)NULL )
for ( i = 0; i < dim; i++ )
sum += fabs(x->ve[i]);
else if ( scale->dim < dim )
error(E_SIZES,"_v_norm1");
else
for ( i = 0; i < dim; i++ )
{ s = scale->ve[i];
sum += ( s== 0.0 ) ? fabs(x->ve[i]) : fabs(x->ve[i]/s);
}
return sum;
}
/* square -- returns x^2 */
double square(x)
double x;
{ return x*x; }
/* cube -- returns x^3 */
double cube(x)
double x;
{ return x*x*x; }
/* _v_norm2 -- computes (scaled) 2-norm (Euclidean norm) of vectors */
double _v_norm2(x,scale)
VEC *x, *scale;
{
int i, dim;
Real s, sum;
if ( x == (VEC *)NULL )
error(E_NULL,"_v_norm2");
dim = x->dim;
sum = 0.0;
if ( scale == (VEC *)NULL )
for ( i = 0; i < dim; i++ )
sum += square(x->ve[i]);
else if ( scale->dim < dim )
error(E_SIZES,"_v_norm2");
else
for ( i = 0; i < dim; i++ )
{ s = scale->ve[i];
sum += ( s== 0.0 ) ? square(x->ve[i]) :
square(x->ve[i]/s);
}
return sqrt(sum);
}
#define max(a,b) ((a) > (b) ? (a) : (b))
/* _v_norm_inf -- computes (scaled) infinity-norm (supremum norm) of vectors */
double _v_norm_inf(x,scale)
VEC *x, *scale;
{
int i, dim;
Real s, maxval, tmp;
if ( x == (VEC *)NULL )
error(E_NULL,"_v_norm_inf");
dim = x->dim;
maxval = 0.0;
if ( scale == (VEC *)NULL )
for ( i = 0; i < dim; i++ )
{ tmp = fabs(x->ve[i]);
maxval = max(maxval,tmp);
}
else if ( scale->dim < dim )
error(E_SIZES,"_v_norm_inf");
else
for ( i = 0; i < dim; i++ )
{ s = scale->ve[i];
tmp = ( s== 0.0 ) ? fabs(x->ve[i]) : fabs(x->ve[i]/s);
maxval = max(maxval,tmp);
}
return maxval;
}
/* m_norm1 -- compute matrix 1-norm -- unscaled */
double m_norm1(A)
MAT *A;
{
int i, j, m, n;
Real maxval, sum;
if ( A == (MAT *)NULL )
error(E_NULL,"m_norm1");
m = A->m; n = A->n;
maxval = 0.0;
for ( j = 0; j < n; j++ )
{
sum = 0.0;
for ( i = 0; i < m; i ++ )
sum += fabs(A->me[i][j]);
maxval = max(maxval,sum);
}
return maxval;
}
/* m_norm_inf -- compute matrix infinity-norm -- unscaled */
double m_norm_inf(A)
MAT *A;
{
int i, j, m, n;
Real maxval, sum;
if ( A == (MAT *)NULL )
error(E_NULL,"m_norm_inf");
m = A->m; n = A->n;
maxval = 0.0;
for ( i = 0; i < m; i++ )
{
sum = 0.0;
for ( j = 0; j < n; j ++ )
sum += fabs(A->me[i][j]);
maxval = max(maxval,sum);
}
return maxval;
}
/* m_norm_frob -- compute matrix frobenius-norm -- unscaled */
double m_norm_frob(A)
MAT *A;
{
int i, j, m, n;
Real sum;
if ( A == (MAT *)NULL )
error(E_NULL,"m_norm_frob");
m = A->m; n = A->n;
sum = 0.0;
for ( i = 0; i < m; i++ )
for ( j = 0; j < n; j ++ )
sum += square(A->me[i][j]);
return sqrt(sum);
}