Skip to content
GitLab
Explore
Sign in
Register
Primary navigation
Search or go to…
Project
O
ogs-feliks
Manage
Activity
Members
Labels
Plan
Issues
Issue boards
Milestones
Wiki
Requirements
Code
Merge requests
Repository
Branches
Commits
Tags
Repository graph
Compare revisions
Locked files
Build
Pipelines
Jobs
Pipeline schedules
Test cases
Artifacts
Deploy
Releases
Package Registry
Container Registry
Model registry
Operate
Environments
Terraform modules
Monitor
Incidents
Service Desk
Analyze
Value stream analytics
Contributor analytics
CI/CD analytics
Repository analytics
Code review analytics
Issue analytics
Insights
Model experiments
Help
Help
Support
GitLab documentation
Compare GitLab plans
Community forum
Contribute to GitLab
Provide feedback
Keyboard shortcuts
?
Snippets
Groups
Projects
Show more breadcrumbs
Feliks Kiszkurno
ogs-feliks
Commits
16945500
Commit
16945500
authored
11 years ago
by
Norihiro Watanabe
Committed by
Dmitri Naumov
10 years ago
Browse files
Options
Downloads
Patches
Plain Diff
add tests for NewtonRaphson
parent
d78bdaf6
No related branches found
Branches containing commit
No related tags found
Tags containing commit
No related merge requests found
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
Tests/MathLib/TestNonlinearNewton.cpp
+295
-0
295 additions, 0 deletions
Tests/MathLib/TestNonlinearNewton.cpp
with
295 additions
and
0 deletions
Tests/MathLib/TestNonlinearNewton.cpp
0 → 100644
+
295
−
0
View file @
16945500
/**
* Copyright (c) 2012, OpenGeoSys Community (http://www.opengeosys.com)
* Distributed under a Modified BSD License.
* See accompanying file LICENSE.txt or
* http://www.opengeosys.com/LICENSE.txt
*
*
* Created on 2012-08-03 by Norihiro Watanabe
*/
#include
<gtest/gtest.h>
#include
"MathLib/LinAlg/Dense/DenseMatrix.h"
#include
"MathLib/LinAlg/Dense/DenseVector.h"
#include
"MathLib/LinAlg/Solvers/GaussAlgorithm.h"
#include
"MathLib/Nonlinear/NewtonRaphson.h"
#include
"Tests/TestTools.h"
namespace
{
typedef
MathLib
::
DenseMatrix
<
double
>
MatrixType
;
typedef
MathLib
::
DenseVector
<
double
>
VectorType
;
typedef
MathLib
::
GaussAlgorithm
<
MatrixType
,
VectorType
>
DenseSolverType
;
template
<
class
F_JACOBIAN
>
class
ScalarDx
{
public:
ScalarDx
(
F_JACOBIAN
&
f_J
)
:
_f_Jacobain
(
f_J
)
{}
// dx = - r/J
void
operator
()(
const
double
&
x
,
const
double
&
r
,
double
&
dx
)
{
double
j
;
_f_Jacobain
(
x
,
j
);
dx
=
-
r
/
j
;
}
private
:
F_JACOBIAN
&
_f_Jacobain
;
};
template
<
class
F_JACOBIAN
>
class
VectorDx
{
public:
VectorDx
(
F_JACOBIAN
&
f_J
,
MatrixType
&
matJ
)
:
_f_Jacobain
(
f_J
),
_matJ
(
matJ
)
{}
// dx = - r/J
void
operator
()(
const
VectorType
&
x
,
const
VectorType
&
r
,
VectorType
&
dx
)
{
_f_Jacobain
(
x
,
_matJ
);
DenseSolverType
solver
(
_matJ
);
VectorType
rhs
(
r
);
rhs
*=
-
1.
;
solver
.
solve
(
rhs
,
dx
);
}
private
:
F_JACOBIAN
&
_f_Jacobain
;
MatrixType
&
_matJ
;
};
//##############################################################################
// Example problem 1 (one variable)
// f(x) = x*x -4 = 0
// x = 2,-2
//##############################################################################
namespace
Example1
{
class
Residual
{
public:
void
operator
()(
const
double
&
x
,
double
&
r
)
{
r
=
x
*
x
-
4.
;
}
};
class
Jacobian
{
public:
void
operator
()(
const
double
&
x
,
double
&
j
)
{
j
=
2
*
x
;
}
};
}
// Example1
//##############################################################################
// Example problem 2 (two variables)
// 3x-y=-2
// 2x^2-y=0
// (x,y) = (-1/2, 1/2) and (2, 8)
//##############################################################################
namespace
Example2
{
class
Residual
{
public:
void
operator
()(
const
VectorType
&
x
,
VectorType
&
r
)
{
r
[
0
]
=
3
*
x
[
0
]
-
x
[
1
]
+
2.
;
r
[
1
]
=
2
*
x
[
0
]
*
x
[
0
]
-
x
[
1
];
}
};
class
Jacobian
{
public:
void
operator
()(
const
VectorType
&
x
,
MatrixType
&
j
)
{
j
(
0
,
0
)
=
3.
;
j
(
0
,
1
)
=
-
1.0
;
j
(
1
,
0
)
=
4.
*
x
[
0
];
j
(
1
,
1
)
=
-
1.0
;
}
};
}
// Example2
//##############################################################################
// Example problem 3 (two variables)
// 3x-y=-2
// 2x^2-y=0
// (x,y) = (-1/2, 1/2) and (2, 8)
//##############################################################################
namespace
Example3
{
class
Residual
{
public:
void
operator
()(
const
VectorType
&
x
,
VectorType
&
r
)
{
double
P
=
1.
;
double
R
=
10.
;
double
s
=
sqrt
(
2.
);
r
[
1
-
1
]
=
(
9
*
P
*
x
[
1
-
1
])
/
4
+
(
9
*
x
[
2
-
1
]
*
x
[
3
-
1
])
/
(
8
*
s
)
+
(
P
*
R
*
x
[
7
-
1
])
/
s
;
r
[
2
-
1
]
=
(
81
*
P
*
x
[
2
-
1
])
/
4
+
(
9
*
x
[
1
-
1
]
*
x
[
3
-
1
])
/
(
8
*
s
)
+
(
P
*
R
*
x
[
8
-
1
])
/
s
;
r
[
3
-
1
]
=
(
-
9
*
x
[
1
-
1
]
*
x
[
2
-
1
])
/
(
4
*
s
)
+
9
*
P
*
x
[
3
-
1
]
+
s
*
P
*
R
*
x
[
9
-
1
];
r
[
4
-
1
]
=
36
*
P
*
x
[
4
-
1
]
+
s
*
P
*
R
*
x
[
10
-
1
];
r
[
5
-
1
]
=
-
2
*
x
[
5
-
1
]
+
(
x
[
2
-
1
]
*
x
[
7
-
1
])
/
(
2
*
s
)
+
(
x
[
1
-
1
]
*
x
[
8
-
1
])
/
(
2
*
s
)
-
(
x
[
4
-
1
]
*
x
[
9
-
1
])
/
s
+
s
*
x
[
4
-
1
]
*
x
[
9
-
1
]
-
(
x
[
3
-
1
]
*
x
[
10
-
1
])
/
s
+
s
*
x
[
3
-
1
]
*
x
[
10
-
1
];
r
[
6
-
1
]
=
-
8
*
x
[
6
-
1
]
-
(
x
[
1
-
1
]
*
x
[
7
-
1
])
/
s
-
s
*
x
[
3
-
1
]
*
x
[
9
-
1
];
r
[
7
-
1
]
=
-
(
x
[
1
-
1
]
/
s
)
-
(
x
[
2
-
1
]
*
x
[
5
-
1
])
/
(
2
*
s
)
+
(
x
[
1
-
1
]
*
x
[
6
-
1
])
/
s
-
(
3
*
x
[
7
-
1
])
/
2.0
+
(
3
*
x
[
3
-
1
]
*
x
[
8
-
1
])
/
(
4
*
s
)
+
(
3
*
x
[
2
-
1
]
*
x
[
9
-
1
])
/
(
4
*
s
);
r
[
8
-
1
]
=
-
(
x
[
2
-
1
]
/
s
)
-
(
x
[
1
-
1
]
*
x
[
5
-
1
])
/
(
2
*
s
)
-
(
3
*
x
[
3
-
1
]
*
x
[
7
-
1
])
/
(
4
*
s
)
-
(
9
*
x
[
8
-
1
])
/
2.0
-
(
3
*
x
[
1
-
1
]
*
x
[
9
-
1
])
/
(
4
*
s
);
r
[
9
-
1
]
=
-
(
s
*
x
[
3
-
1
])
-
(
x
[
4
-
1
]
*
x
[
5
-
1
])
/
s
+
s
*
x
[
3
-
1
]
*
x
[
6
-
1
]
-
(
3
*
x
[
2
-
1
]
*
x
[
7
-
1
])
/
(
4
*
s
)
+
(
3
*
x
[
1
-
1
]
*
x
[
8
-
1
])
/
(
4
*
s
)
-
3
*
x
[
9
-
1
];
r
[
10
-
1
]
=
-
(
s
*
x
[
4
-
1
])
-
(
x
[
3
-
1
]
*
x
[
5
-
1
])
/
s
-
6
*
x
[
10
-
1
];
}
};
class
Jacobian
{
public:
void
operator
()(
const
VectorType
&
x
,
MatrixType
&
j
)
{
double
P
=
1.
;
double
R
=
10.
;
double
s
=
sqrt
(
2.
);
j
=
.0
;
j
(
1
-
1
,
1
-
1
)
=
(
9
*
P
)
/
4.0
;
j
(
7
-
1
,
1
-
1
)
=
-
(
1
/
s
)
+
x
[
6
-
1
]
/
s
;
j
(
1
-
1
,
2
-
1
)
=
(
9
*
x
[
3
-
1
])
/
(
8
*
s
);
j
(
7
-
1
,
2
-
1
)
=
-
x
[
5
-
1
]
/
(
2
*
s
)
+
(
3
*
x
[
9
-
1
])
/
(
4
*
s
);
j
(
1
-
1
,
3
-
1
)
=
(
9
*
x
[
2
-
1
])
/
(
8
*
s
);
j
(
7
-
1
,
3
-
1
)
=
(
3
*
x
[
8
-
1
])
/
(
4
*
s
);
j
(
1
-
1
,
7
-
1
)
=
(
P
*
R
)
/
s
;
j
(
7
-
1
,
5
-
1
)
=
-
x
[
2
-
1
]
/
(
2
*
s
);
j
(
2
-
1
,
1
-
1
)
=
(
9
*
x
[
3
-
1
])
/
(
8
*
s
);
j
(
7
-
1
,
6
-
1
)
=
x
[
1
-
1
]
/
s
;
j
(
2
-
1
,
2
-
1
)
=
(
81
*
P
)
/
4.0
;
j
(
7
-
1
,
7
-
1
)
=
-
1.5
;
j
(
2
-
1
,
3
-
1
)
=
(
9
*
x
[
1
-
1
])
/
(
8
*
s
);
j
(
7
-
1
,
8
-
1
)
=
(
3
*
x
[
3
-
1
])
/
(
4
*
s
);
j
(
2
-
1
,
8
-
1
)
=
(
P
*
R
)
/
s
;
j
(
7
-
1
,
9
-
1
)
=
(
3
*
x
[
2
-
1
])
/
(
4
*
s
);
j
(
3
-
1
,
1
-
1
)
=
(
-
9
*
x
[
2
-
1
])
/
(
4
*
s
);
j
(
8
-
1
,
1
-
1
)
=
-
x
[
5
-
1
]
/
(
2
*
s
)
-
(
3
*
x
[
9
-
1
])
/
(
4
*
s
);
j
(
3
-
1
,
2
-
1
)
=
(
-
9
*
x
[
1
-
1
])
/
(
4
*
s
);
j
(
8
-
1
,
2
-
1
)
=
-
(
1
/
s
);
j
(
3
-
1
,
3
-
1
)
=
9
*
P
;
j
(
8
-
1
,
3
-
1
)
=
(
-
3
*
x
[
7
-
1
])
/
(
4
*
s
);
j
(
3
-
1
,
9
-
1
)
=
s
*
P
*
R
;
j
(
8
-
1
,
5
-
1
)
=
-
x
[
1
-
1
]
/
(
2
*
s
);
j
(
4
-
1
,
4
-
1
)
=
36
*
P
;
j
(
8
-
1
,
7
-
1
)
=
(
-
3
*
x
[
3
-
1
])
/
(
4
*
s
);
j
(
4
-
1
,
10
-
1
)
=
s
*
P
*
R
;
j
(
8
-
1
,
8
-
1
)
=
-
4.5
;
j
(
5
-
1
,
1
-
1
)
=
x
[
8
-
1
]
/
(
2
*
s
);
j
(
8
-
1
,
9
-
1
)
=
(
-
3
*
x
[
1
-
1
])
/
(
4
*
s
);
j
(
5
-
1
,
2
-
1
)
=
x
[
7
-
1
]
/
(
2
*
s
);
j
(
9
-
1
,
1
-
1
)
=
(
3
*
x
[
8
-
1
])
/
(
4
*
s
);
j
(
5
-
1
,
3
-
1
)
=
-
(
x
[
10
-
1
]
/
s
)
+
s
*
x
[
10
-
1
];
j
(
9
-
1
,
2
-
1
)
=
(
-
3
*
x
[
7
-
1
])
/
(
4
*
s
);
j
(
5
-
1
,
4
-
1
)
=
-
(
x
[
9
-
1
]
/
s
)
+
s
*
x
[
9
-
1
];
j
(
9
-
1
,
3
-
1
)
=
-
s
+
s
*
x
[
6
-
1
];
j
(
5
-
1
,
5
-
1
)
=
-
2.0
;
j
(
9
-
1
,
4
-
1
)
=
-
(
x
[
5
-
1
]
/
s
);
j
(
5
-
1
,
7
-
1
)
=
x
[
2
-
1
]
/
(
2
*
s
);
j
(
9
-
1
,
5
-
1
)
=
-
(
x
[
4
-
1
]
/
s
);
j
(
5
-
1
,
8
-
1
)
=
x
[
1
-
1
]
/
(
2
*
s
);
j
(
9
-
1
,
6
-
1
)
=
s
*
x
[
3
-
1
];
j
(
5
-
1
,
9
-
1
)
=
-
(
x
[
4
-
1
]
/
s
)
+
s
*
x
[
4
-
1
];
j
(
9
-
1
,
7
-
1
)
=
(
-
3
*
x
[
2
-
1
])
/
(
4
*
s
);
j
(
5
-
1
,
10
-
1
)
=
-
(
x
[
3
-
1
]
/
s
)
+
s
*
x
[
3
-
1
];
j
(
9
-
1
,
8
-
1
)
=
(
3
*
x
[
1
-
1
])
/
(
4
*
s
);
j
(
6
-
1
,
1
-
1
)
=
-
(
x
[
7
-
1
]
/
s
);
j
(
9
-
1
,
9
-
1
)
=
-
3.0
;
j
(
6
-
1
,
3
-
1
)
=
-
(
s
*
x
[
9
-
1
]);
j
(
10
-
1
,
3
-
1
)
=
-
(
x
[
5
-
1
]
/
s
);
j
(
6
-
1
,
6
-
1
)
=
-
8.0
;
j
(
10
-
1
,
4
-
1
)
=
-
s
;
j
(
6
-
1
,
7
-
1
)
=
-
(
x
[
1
-
1
]
/
s
);
j
(
10
-
1
,
5
-
1
)
=
-
(
x
[
3
-
1
]
/
s
);
j
(
6
-
1
,
9
-
1
)
=
-
(
s
*
x
[
3
-
1
]);
j
(
10
-
1
,
10
-
1
)
=
-
6.0
;
}
};
}
// Example 3
}
//namespace
//##############################################################################
// Tests
//##############################################################################
TEST
(
MathLib
,
NonlinearNR_double
)
{
Example1
::
Residual
f_r
;
Example1
::
Jacobian
f_j
;
ScalarDx
<
Example1
::
Jacobian
>
f_dx
(
f_j
);
double
x0
=
6.0
;
double
x
=
.0
;
MathLib
::
Nonlinear
::
NewtonRaphson
nr
;
nr
.
solve
(
f_r
,
f_dx
,
x0
,
x
);
ASSERT_NEAR
(
2.0
,
x
,
1e-5
);
}
TEST
(
MathLib
,
NonlinearNR_dense
)
{
Example2
::
Residual
f_r
;
Example2
::
Jacobian
f_j
;
MatrixType
matJ
(
2
,
2
);
VectorDx
<
Example2
::
Jacobian
>
f_dx
(
f_j
,
matJ
);
VectorType
x0
(
2
),
x
(
2
);
x0
=
6.0
;
x
=
.0
;
MathLib
::
Nonlinear
::
NewtonRaphson
nr
;
nr
.
solve
(
f_r
,
f_dx
,
x0
,
x
);
double
my_expect
[]
=
{
2.
,
8.
};
ASSERT_ARRAY_NEAR
(
my_expect
,
x
,
2
,
1e-5
);
}
TEST
(
MathLib
,
NonlinearNR_dense2
)
{
Example3
::
Residual
f_r
;
Example3
::
Jacobian
f_j
;
const
std
::
size_t
n
=
10
;
MatrixType
matJ
(
n
,
n
,
.0
);
VectorDx
<
Example3
::
Jacobian
>
f_dx
(
f_j
,
matJ
);
VectorType
x0
(
n
),
x
(
n
);
x0
=
1.
;
x
=
0.
;
MathLib
::
Nonlinear
::
NewtonRaphson
nr
;
nr
.
solve
(
f_r
,
f_dx
,
x0
,
x
);
double
my_expect
[]
=
{
3.39935
,
3.70074e-018
,
-
1.42576e-017
,
1.4903e-021
,
4.35602e-018
,
0.325
,
-
1.08167
,
-
5.61495e-018
,
7.58394e-018
,
-
3.79368e-021
};
ASSERT_ARRAY_NEAR
(
my_expect
,
x
,
n
,
1e-5
);
}
#if 0
TEST(MathLib, NonlinearNR_sparse)
{
typedef std::valarray<double> MyVector;
typedef MathLib::CRSMatrix<double, signed> MyMatrix;
typedef NRCheckConvergence<MyVector,NRErrorAbsResMNormOrRelDxMNorm > MyConverge;
typedef NewtonFunctionDXVector<NL3_NR_D1, MathLib::LisLinearEquation> MyDxFunction;
NL3_NR f;
NL3_NR_D1 df(f.getLinearSolver());
MyVector x0(6.0, 2);
MyVector x(0.0, 2);
MyVector r(2), dx(2);
//MyMatrix* j = f.getLinearSolver()->getA();
MyDxFunction f_dx(df, *f.getLinearSolver());
NewtonRaphsonMethod nr;
nr.solve<NL3_NR,MyDxFunction,MyVector,MyConverge, NRIterationStepInitializerDummy>(f, f_dx, x0, x, r, dx);
double my_expect[] = {2., 8.};
ASSERT_DOUBLE_ARRAY_EQ(my_expect, x, 2, 1e-5);
}
#endif
This diff is collapsed.
Click to expand it.
Preview
0%
Loading
Try again
or
attach a new file
.
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Save comment
Cancel
Please
register
or
sign in
to comment