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
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
|
// GetDP - Copyright (C) 1997-2018 P. Dular and C. Geuzaine, University of Liege
//
// See the LICENSE.txt file for license information. Please report all
// issues on https://gitlab.onelab.info/getdp/getdp/issues
//
// Contributor(s):
// Michael Asam
#include <stdio.h>
#include <limits>
#include <math.h>
#include "ProData.h"
#include "DofData.h"
#include "SolvingOperations.h"
#include "Message.h"
void Cal_SolutionErrorRatio(gVector *dx, gVector *x,
double reltol, double abstol,
int NormType, double *ErrorRatio)
{
int xLength;
double AbsVal_x, AbsVal_dx, ImagVal_x, ImagVal_dx;
double *ErrorRatioVec;
bool Is_NaN_or_Inf;
LinAlg_GetVectorSize(dx, &xLength);
ErrorRatioVec = new double[xLength];
*ErrorRatio = 0.;
Is_NaN_or_Inf = false;
for (int i = 0; i < xLength; i++) {
if (gSCALAR_SIZE == 1)
{
LinAlg_GetAbsDoubleInVector(&AbsVal_x, x, i) ;
LinAlg_GetAbsDoubleInVector(&AbsVal_dx, dx, i) ;
}
if (gSCALAR_SIZE == 2)
{
LinAlg_GetComplexInVector(&AbsVal_x, &ImagVal_x, x, i, -1);
LinAlg_GetComplexInVector(&AbsVal_dx, &ImagVal_dx, dx, i, -1);
AbsVal_x = sqrt( AbsVal_x*AbsVal_x + ImagVal_x*ImagVal_x);
AbsVal_dx = sqrt( AbsVal_dx*AbsVal_dx + ImagVal_dx*ImagVal_dx);
}
ErrorRatioVec[i] = AbsVal_dx / (abstol + reltol * AbsVal_x);
if ( ErrorRatioVec[i] != ErrorRatioVec[i] || // Solution is NaN
ErrorRatioVec[i] == -std::numeric_limits<double>::infinity() || // Solution is -Inf
ErrorRatioVec[i] == std::numeric_limits<double>::infinity() ) // Solution is Inf
Is_NaN_or_Inf = true;
}
if ( Is_NaN_or_Inf ) {
//Message::Error("No valid solution found (NaN or Inf)!"); // kj+++
Message::Warning("No valid solution found (NaN or Inf)!");
*ErrorRatio = std::numeric_limits<double>::quiet_NaN();
}
else{
// Calculating the norm of the error ratio vector
switch (NormType){
case LINFNORM:
for (int i = 0; i < xLength; i++) {
if (ErrorRatioVec[i] > *ErrorRatio)
*ErrorRatio = ErrorRatioVec[i];
}
break;
case L1NORM:
for (int i = 0; i < xLength; i++) {
*ErrorRatio += ErrorRatioVec[i];
}
break;
case MEANL1NORM:
for (int i = 0; i < xLength; i++) {
*ErrorRatio += ErrorRatioVec[i];
}
*ErrorRatio /= xLength;
break;
case L2NORM:
for (int i = 0; i < xLength; i++) {
*ErrorRatio += ErrorRatioVec[i] * ErrorRatioVec[i];
}
*ErrorRatio = sqrt(*ErrorRatio);
break;
case MEANL2NORM:
for (int i = 0; i < xLength; i++) {
*ErrorRatio += ErrorRatioVec[i] * ErrorRatioVec[i];
}
*ErrorRatio = sqrt(*ErrorRatio / xLength);
break;
default:
Message::Error("Wrong error norm in Cal_SolutionErrorRatio");
break;
}
}
delete [] ErrorRatioVec;
}
/* ------------------------------------------------------------------------ */
/* C a l _ S o l u t i o n E r r o r */
/* ------------------------------------------------------------------------ */
void Cal_SolutionError(gVector *dx, gVector *x, int diff, double *MeanError)
{
// This is not a very good implementation: it should be replaced with
// Cal_SolutionErrorRatio above
int i, n;
double valx, valdx, valxi = 0., valdxi = 0.,errsqr = 0., xmoy = 0., dxmoy = 0.;
double tol, nvalx, nvaldx ;
//bool Is_NaN_or_Inf=0; // kj+++
LinAlg_GetVectorSize(dx, &n);
if (gSCALAR_SIZE == 1)
for (i=0 ; i<n ; i++) {
LinAlg_GetAbsDoubleInVector(&valx, x, i) ;
LinAlg_GetAbsDoubleInVector(&valdx, dx, i) ;
xmoy += valx ;
if(diff) dxmoy += (valdx-valx) ;
else dxmoy += valdx ;
}
if (gSCALAR_SIZE == 2)
for (i=0 ; i<n ; i++) {
LinAlg_GetComplexInVector(&valx, &valxi, x, i, -1);
LinAlg_GetComplexInVector(&valdx, &valdxi, dx, i, -1);
xmoy += sqrt(valx*valx+valxi*valxi) ;
if(diff) dxmoy += sqrt((valdx-valx)*(valdx-valx)+(valdxi-valxi)*(valdxi-valxi)) ;
else dxmoy += sqrt(valdx*valdx + valdxi*valdxi) ;
}
xmoy /= (double)n ;
dxmoy /= (double)n ;
if (xmoy > 1.e-30) {
tol = xmoy*1.e-10 ;
//tol = 1e200; // kj+++
//tol=(tol<1e-3)?1e-3:tol; // kj+++
if (gSCALAR_SIZE == 1)
for (i=0 ; i<n ; i++){
LinAlg_GetAbsDoubleInVector(&valx, x, i) ;
LinAlg_GetAbsDoubleInVector(&valdx, dx, i) ;
if(diff){
if (valx > tol) errsqr += fabs(valdx-valx)/valx ;
else errsqr += fabs(valdx-valx) ;
}
else{
if (valx > tol) errsqr += valdx/valx ;
else errsqr += valdx ;
}
}
if (gSCALAR_SIZE == 2)
for (i=0 ; i<n ; i++) {
LinAlg_GetComplexInVector(&valx, &valxi, x, i, -1);
LinAlg_GetComplexInVector(&valdx, &valdxi, dx, i, -1);
nvalx = sqrt(valx*valx+valxi*valxi) ;
nvaldx = sqrt(valdx*valdx+valdxi*valdxi) ;
if(diff){
if (nvalx > tol)
errsqr += sqrt((valdx-valx)*(valdx-valx)+(valdxi-valxi)*(valdxi-valxi))/nvalx ;
else
errsqr += sqrt((valdx-valx)*(valdx-valx)+(valdxi-valxi)*(valdxi-valxi));
}
else{
if (nvalx > tol) errsqr += nvaldx/nvalx ;
else errsqr += nvaldx ;
}
}
*MeanError = errsqr/(double)n ;
}
else{
if (dxmoy > 1.e-30)
*MeanError = 1. ;
else
*MeanError = 0. ;
}
/* // kj+++
if ( *MeanError != *MeanError || // Solution is NaN
*MeanError == -std::numeric_limits<double>::infinity() || // Solution is -Inf
*MeanError == std::numeric_limits<double>::infinity() ) // Solution is Inf
Is_NaN_or_Inf = true;
if ( Is_NaN_or_Inf ) {
Message::Error("No valid solution found (NaN or Inf)!");
*MeanError = std::numeric_limits<double>::quiet_NaN();
}
*/
}
|
