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
|
/***********************************************/
/**
* @file normalsMultiplyAdd.cpp
*
* @brief n += alpha*N*x.
*
* @author Torsten Mayer-Guerr
* @date 2011-02-19
*/
/***********************************************/
// Latex documentation
#define DOCSTRING docstring
static const char *docstring = R"(
This program modifies \configFile{inputfileNormalEquation}{normalEquation} in a way
that $\bar{\M x}$ is estimated instead of $\M x$.
\begin{equation}
\bar{\M x} := \M x + \alpha\, \M x_0,
\end{equation}
where $\M x_0$ is \configFile{inputfileParameter}{matrix} and $\alpha$ is \config{factor}.
This can be used to re-add reduced reference fields before a combined estimation
at normal equation level.
Therefore the right hand side of the normal equations is modified by
\begin{equation}
\bar{\M n} := \M n + \alpha\,\M N\M x_0,
\end{equation}
and the quadratic sum of observations by
\begin{equation}
\bar{\M l^T\M P\M l} := \M l^T\M P\M l + \alpha^2\,\M x_0^T\M N\M x_0 + 2\alpha\,\M x_0^T\M n
\end{equation}
As the normal matrix itself is not modified, rewriting of the matrix can be disabled by setting
\config{writeNormalMatrix} to false.
)";
/***********************************************/
#include "programs/program.h"
#include "parallel/matrixDistributed.h"
#include "files/fileMatrix.h"
#include "files/fileNormalEquation.h"
/***** CLASS ***********************************/
/** @brief n += alpha*N*x.
* @ingroup programsGroup */
class NormalsMultiplyAdd
{
public:
void run(Config &config, Parallel::CommunicatorPtr comm);
};
GROOPS_REGISTER_PROGRAM(NormalsMultiplyAdd, PARALLEL, "n += alpha*N*x", NormalEquation)
/***********************************************/
void NormalsMultiplyAdd::run(Config &config, Parallel::CommunicatorPtr comm)
{
try
{
FileName outName, normalsName, xName;
Double factor;
Bool writeMatrix;
renameDeprecatedConfig(config, "outputfileNormalequation", "outputfileNormalEquation", date2time(2020, 6, 3));
renameDeprecatedConfig(config, "inputfileNormalequation", "inputfileNormalEquation", date2time(2020, 6, 3));
readConfig(config, "outputfileNormalEquation", outName, Config::MUSTSET, "", "");
readConfig(config, "inputfileNormalEquation", normalsName, Config::MUSTSET, "", "");
readConfig(config, "inputfileParameter", xName, Config::MUSTSET, "", "x");
readConfig(config, "factor", factor, Config::DEFAULT, "1.0", "alpha");
readConfig(config, "writeNormalMatrix", writeMatrix, Config::DEFAULT, "1", "write full coefficient matrix, right hand sides and info files");
if(isCreateSchema(config)) return;
// ==================================
logStatus<<"init normal equations"<<Log::endl;
MatrixDistributed normal;
Matrix n;
NormalEquationInfo info;
readFileNormalEquation(normalsName, info, normal, n, comm);
logInfo<<" number of parameters: "<<normal.parameterCount()<<Log::endl;
logInfo<<" number of right hand sides: "<<info.lPl.rows()<<Log::endl;
Matrix x;
if(Parallel::isMaster(comm))
{
readFileMatrix(xName, x);
if((x.rows() != n.rows()) || (x.columns() != n.columns()))
{
std::stringstream ss;
ss<<"Dimension error N("<<n.rows()<<" x "<<n.rows()<<")*x("<<x.rows()<<" x "<<x.columns()<<") = n("<<n.rows()<<" x "<<n.columns()<<")";
throw(Exception(ss.str()));
}
}
// ==================================
logStatus<<"multiply normal matrix"<<Log::endl;
Parallel::broadCast(x, 0, comm);
x *= factor;
Matrix Nx(x.rows(), x.columns());
for(UInt i=0; i<normal.blockCount(); i++)
{
if(normal.isMyRank(i,i))
matMult(1., normal.N(i,i), x.row(normal.blockIndex(i), normal.blockSize(i)), Nx.row(normal.blockIndex(i), normal.blockSize(i)));
for(UInt k=i+1; k<normal.blockCount(); k++)
if(normal.isMyRank(i,k))
{
matMult(1., normal.N(i,k), x.row(normal.blockIndex(k), normal.blockSize(k)), Nx.row(normal.blockIndex(i), normal.blockSize(i)));
matMult(1., normal.N(i,k).trans(), x.row(normal.blockIndex(i), normal.blockSize(i)), Nx.row(normal.blockIndex(k), normal.blockSize(k)));
}
}
Parallel::reduceSum(Nx, 0, comm);
if(Parallel::isMaster(comm))
{
for(UInt i=0; i<info.lPl.rows(); i++)
info.lPl(i) += 2.*inner(x.column(i), n.column(i)) + inner(x.column(i), Nx.column(i));
n += Nx;
}
// ==================================
// write normal equations
// ----------------------
logStatus<<"write normal equations to <"<<outName<<">"<<Log::endl;
if((normalsName==outName) || !writeMatrix)
{
logStatus<<"(only the information file and the right hand sides)"<<Log::endl;
if(Parallel::isMaster(comm))
writeFileNormalEquation(outName, info, n);
}
else
writeFileNormalEquation(outName, info, normal, n);
}
catch(std::exception &e)
{
GROOPS_RETHROW(e)
}
}
/***********************************************/
|
