Skip to content

Commit b4c8578

Browse files
memmettmemmett
committed
Merge pull request #63 from torbjoernk/feature/cleanup-scalar
Very nice.
2 parents 00462bd + 9bdf645 commit b4c8578

File tree

5 files changed

+152
-121
lines changed

5 files changed

+152
-121
lines changed

‎examples/advection_diffusion/CMakeLists.txt‎

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -30,7 +30,7 @@ set(all_advec_examples
3030
foreach(example ${all_advec_examples})
3131
add_executable(${example} ${CMAKE_CURRENT_SOURCE_DIR}/${example}.cpp)
3232
if(NOT FFTW_FOUND)
33-
add_dependencies(${example} fftw3)
33+
add_dependencies(${example} fftw3)
3434
endif()
3535
target_link_libraries(${example}
3636
${FFTW_LIBRARIES}
@@ -55,7 +55,7 @@ endforeach(example)
5555
if(${pfasst_WITH_MPI})
5656
include_directories(${MPI_CXX_INCLUDE_PATH})
5757
foreach(example ${advec_mpi_examples})
58-
if(MPI_COMPILE_FLAGS)
58+
if(MPI_COMPILE_FLAGS)
5959
set_target_properties(${example} PROPERTIES COMPILE_FLAGS "${MPI_COMPILE_FLAGS}")
6060
endif()
6161
if(MPI_LINK_FLAGS)
@@ -77,9 +77,9 @@ if(${pfasst_WITH_MPI})
7777
endif()
7878

7979
add_custom_target(run_example_advec_all
80-
DEPENDS ${all_advec_example_runs}
81-
COMMENT "Running all Advection-Diffusion examples" VERBATIM
80+
DEPENDS ${all_advec_example_runs}
81+
COMMENT "Running all Advection-Diffusion examples" VERBATIM
8282
)
8383
message(STATUS " - run_example_advec_all")
8484

85-
set(all_example_runs ${all_advec_example_runs} PARENT_SCOPE)
85+
set(all_example_runs ${all_example_runs} ${all_advec_example_runs} PARENT_SCOPE)

‎examples/scalar/CMakeLists.txt‎

Lines changed: 23 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,6 @@
1+
set(all_example_runs ${all_example_runs})
2+
set(all_scalar_example_runs)
3+
14
message(STATUS " scalar")
25
include_directories(
36
${pfasst_INCLUDES}
@@ -14,3 +17,23 @@ foreach(example ${scalar_examples})
1417
PROPERTIES RUNTIME_OUTPUT_DIRECTORY ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/examples/scalar
1518
)
1619
endforeach(example)
20+
21+
message(STATUS " creating make targets to run examples")
22+
foreach(example ${scalar_examples})
23+
add_custom_target(run_example_scalar_${example}
24+
COMMAND ${example}
25+
DEPENDS ${example}
26+
WORKING_DIRECTORY ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/examples/scalar
27+
COMMENT "Running Scalar example: ${example}" VERBATIM
28+
)
29+
list(APPEND all_scalar_example_runs run_example_scalar_${example})
30+
message(STATUS " - run_example_scalar_${example}")
31+
endforeach(example)
32+
33+
add_custom_target(run_example_scalar_all
34+
DEPENDS ${all_scalar_example_runs}
35+
COMMENT "Running all Scalar examples" VERBATIM
36+
)
37+
message(STATUS " - run_example_scalar_all")
38+
39+
set(all_example_runs ${all_example_runs} ${all_scalar_example_runs} PARENT_SCOPE)

‎examples/scalar/scalar_sdc.cpp‎

Lines changed: 9 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -18,28 +18,27 @@
1818
*/
1919

2020
double run_scalar_sdc(const size_t nsteps, const double dt, const size_t nnodes,
21-
const size_t niters, const complex<double> lambda)
21+
const size_t niters, const complex<double> lambda)
2222
{
23-
2423
pfasst::SDC<> sdc;
2524

26-
const complex<double> y0 = complex<double>(1.0, 0.0);
25+
const complex<double> y0 = complex<double>(1.0, 0.0);
2726
auto nodes = pfasst::compute_nodes(nnodes, "gauss-lobatto");
2827
auto factory = make_shared<pfasst::encap::VectorFactory<complex<double>>>(1);
29-
auto sweeper = make_shared<ScalarSweeper<>>(lambda,y0);
30-
28+
auto sweeper = make_shared<ScalarSweeper<>>(lambda, y0);
29+
3130
sweeper->set_nodes(nodes);
3231
sweeper->set_factory(factory);
33-
32+
3433
sdc.add_level(sweeper);
3534
sdc.set_duration(0.0, dt*nsteps, dt, niters);
3635
sdc.setup();
37-
36+
3837
auto q0 = sweeper->get_state(0);
3938
sweeper->exact(q0, 0.0);
40-
39+
4140
sdc.run();
42-
41+
4342
return sweeper->get_errors();
4443
}
4544

@@ -51,7 +50,7 @@ int main(int /*argc*/, char** /*argv*/)
5150
const size_t nnodes = 4;
5251
const size_t niters = 6;
5352
const complex<double> lambda = complex<double>(-1.0, 1.0);
54-
53+
5554
run_scalar_sdc(nsteps, dt, nnodes, niters, lambda);
5655
}
5756
#endif

‎examples/scalar/scalar_sweeper.hpp‎

Lines changed: 63 additions & 62 deletions
Original file line numberDiff line numberDiff line change
@@ -2,57 +2,63 @@
22
* Sweeper for scalar test equation
33
*/
44

5-
#ifndef _SCALAR_SWEEPER_HPP_
6-
#define _SCALAR_SWEEPER_HPP_
5+
#ifndef _EXAMPLES__SCALAR__SCALAR_SWEEPER_HPP_
6+
#define _EXAMPLES__SCALAR__SCALAR_SWEEPER_HPP_
77

88
#include <complex>
99
#include <vector>
10+
1011
#include <pfasst/encap/imex_sweeper.hpp>
1112
#include <pfasst/encap/vector.hpp>
1213

1314
using namespace std;
1415

1516
template<typename time = pfasst::time_precision>
16-
class ScalarSweeper : public pfasst::encap::IMEXSweeper<time>
17+
class ScalarSweeper
18+
: public pfasst::encap::IMEXSweeper<time>
1719
{
20+
private:
21+
typedef pfasst::encap::Encapsulation<time> encap_type;
22+
typedef pfasst::encap::VectorEncapsulation<complex<double>> complex_vector_type;
1823

19-
typedef pfasst::encap::Encapsulation<time> Encapsulation;
20-
typedef pfasst::encap::VectorEncapsulation<complex<double>> CVectorT;
24+
complex<double> lambda, u0;
25+
size_t n_f_expl_eval, n_f_impl_eval, n_impl_solve;
26+
const complex<double> i_complex = complex<double>(0, 1);
27+
double error;
2128

2229
public:
23-
24-
ScalarSweeper(complex<double> lambda, complex<double> y0)
25-
{
26-
this->_lambda = lambda;
27-
this->_y0 = y0;
28-
this->_nf1eval = 0;
29-
this->_nf2eval = 0;
30-
this->_nf2comp = 0;
31-
}
30+
ScalarSweeper(complex<double> lambda, complex<double> u0)
31+
: lambda(lambda)
32+
, u0(u0)
33+
, n_f_expl_eval(0)
34+
, n_f_impl_eval(0)
35+
, n_impl_solve(0)
36+
, error(0.0)
37+
{}
3238

3339
virtual ~ScalarSweeper()
3440
{
35-
cout << "Final error: " << scientific << this->_error << endl;
36-
cout << "Number of calls to f1eval: " << this->_nf1eval << endl;
37-
cout << "Number of calls to f2eval: " << this->_nf2eval << endl;
38-
cout << "Number of calls to f2comp: " << this->_nf2comp << endl;
41+
cout << "Final error: " << scientific << this->error << endl;
42+
cout << "Number of explicit evaluations: " << this->n_f_expl_eval << endl;
43+
cout << "Number of implicit evaluations: " << this->n_f_impl_eval << endl;
44+
cout << "Number of implicit solves: " << this->n_impl_solve << endl;
3945
}
4046

4147
void echo_error(time t)
4248
{
43-
auto& qend = pfasst::encap::as_vector<complex<double>,time>(this->get_end_state());
49+
auto& qend = pfasst::encap::as_vector<complex<double>, time>(this->get_end_state());
4450

45-
CVectorT qex(qend.size());
51+
complex_vector_type qex(qend.size());
4652

4753
this->exact(qex, t);
4854
double max_err = abs(qend[0] - qex[0]) / abs(qex[0]);
4955
cout << "err: " << scientific << max_err << endl;
50-
this->_error = max_err;
56+
this->error = max_err;
5157
}
52-
58+
5359
double get_errors()
5460
{
55-
return this->_error;
61+
return this->error;
5662
}
5763

5864
void predict(bool initial) override
@@ -71,64 +77,59 @@ class ScalarSweeper : public pfasst::encap::IMEXSweeper<time>
7177
this->echo_error(t + dt);
7278
}
7379

74-
void exact(CVectorT& q, time t)
80+
void exact(complex_vector_type& q, time t)
7581
{
76-
q[0] = _y0 * exp(_lambda * t);
82+
q[0] = this->u0 * exp(this->lambda * t);
7783
}
7884

79-
void exact(shared_ptr<Encapsulation> q_encap, time t)
85+
void exact(shared_ptr<encap_type> q_encap, time t)
8086
{
81-
auto& q = pfasst::encap::as_vector<complex<double>,time>(q_encap);
82-
exact(q, t);
87+
auto& q = pfasst::encap::as_vector<complex<double>, time>(q_encap);
88+
this->exact(q, t);
8389
}
8490

85-
void f_expl_eval(shared_ptr<Encapsulation> f_encap,
86-
shared_ptr<Encapsulation> q_encap, time t) override
91+
void f_expl_eval(shared_ptr<encap_type> f_encap,
92+
shared_ptr<encap_type> q_encap, time t) override
8793
{
8894
UNUSED(t);
89-
auto& f = pfasst::encap::as_vector<complex<double>,time>(f_encap);
90-
auto& q = pfasst::encap::as_vector<complex<double>,time>(q_encap);
95+
auto& f = pfasst::encap::as_vector<complex<double>, time>(f_encap);
96+
auto& q = pfasst::encap::as_vector<complex<double>, time>(q_encap);
9197

92-
// f1 = multiply with imaginary part of lambda
93-
f[0] = i_complex * imag(this->_lambda) * q[0];
94-
this->_nf1eval++;
98+
// f_expl = multiply with imaginary part of lambda
99+
f[0] = this->i_complex * imag(this->lambda) * q[0];
100+
101+
this->n_f_expl_eval++;
95102
}
96103

97-
void f_impl_eval(shared_ptr<Encapsulation> f_encap,
98-
shared_ptr<Encapsulation> q_encap, time t) override
104+
void f_impl_eval(shared_ptr<encap_type> f_encap,
105+
shared_ptr<encap_type> q_encap, time t) override
99106
{
100107
UNUSED(t);
101-
auto& f = pfasst::encap::as_vector<complex<double>,time>(f_encap);
102-
auto& q = pfasst::encap::as_vector<complex<double>,time>(q_encap);
108+
auto& f = pfasst::encap::as_vector<complex<double>, time>(f_encap);
109+
auto& q = pfasst::encap::as_vector<complex<double>, time>(q_encap);
110+
111+
// f_impl = multiply with real part of lambda
112+
f[0] = real(this->lambda) * q[0];
103113

104-
// f2 = multiply with real part of lambda
105-
f[0] = real(this->_lambda) * q[0];
106-
this->_nf2eval++;
114+
this->n_f_impl_eval++;
107115
}
108116

109-
void impl_solve(shared_ptr<Encapsulation> f_encap,
110-
shared_ptr<Encapsulation> q_encap, time t, time dt,
111-
shared_ptr<Encapsulation> rhs_encap) override
117+
void impl_solve(shared_ptr<encap_type> f_encap,
118+
shared_ptr<encap_type> q_encap, time t, time dt,
119+
shared_ptr<encap_type> rhs_encap) override
112120
{
113121
UNUSED(t);
114-
auto& f = pfasst::encap::as_vector<complex<double>,time>(f_encap);
115-
auto& q = pfasst::encap::as_vector<complex<double>,time>(q_encap);
116-
auto& rhs = pfasst::encap::as_vector<complex<double>,time>(rhs_encap);
122+
auto& f = pfasst::encap::as_vector<complex<double>, time>(f_encap);
123+
auto& q = pfasst::encap::as_vector<complex<double>, time>(q_encap);
124+
auto& rhs = pfasst::encap::as_vector<complex<double>, time>(rhs_encap);
117125

118-
// invert f2=multiply with inverse of real part of lambda
119-
double inv = 1 / (1 - double(dt) * real(this->_lambda));
126+
// invert f_impl = multiply with inverse of real part of lambda
127+
double inv = 1.0 / (1.0 - double(dt) * real(this->lambda));
120128
q[0] = inv * rhs[0];
121-
f[0] = real(this->_lambda) * q[0];
122-
this->_nf2comp++;
123-
}
124-
125-
126-
private:
127-
128-
complex<double> _lambda, _y0;
129-
int _nf1eval, _nf2eval, _nf2comp;
130-
const complex<double> i_complex = complex<double>(0, 1);
131-
double _error;
129+
f[0] = real(this->lambda) * q[0];
132130

131+
this->n_impl_solve++;
132+
}
133133
};
134-
#endif
134+
135+
#endif // _EXAMPLES__SCALAR__SCALAR_SWEEPER_HPP_

0 commit comments

Comments
 (0)