diff --git a/src/analysis/PolynomialP.hpp b/src/analysis/PolynomialP.hpp
index 42f0c46663019d7f4e65220c8e25fbd6a89c94e6..76f6b46d63512f5bb32f2a6e0dc466a48e03e627 100644
--- a/src/analysis/PolynomialP.hpp
+++ b/src/analysis/PolynomialP.hpp
@@ -3,6 +3,7 @@
#include <algebra/TinyVector.hpp>
#include <analysis/CubeGaussQuadrature.hpp>
+#include <analysis/GaussLegendreQuadratureDescriptor.hpp>
#include <analysis/GaussQuadratureDescriptor.hpp>
#include <analysis/QuadratureManager.hpp>
#include <analysis/SquareGaussQuadrature.hpp>
@@ -616,7 +617,8 @@ integrate(const PolynomialP<N, Dimension> Q, const std::array<TinyVector<Dimensi
} else if constexpr (Dimension == 2) {
static_assert(P <= 4, "In 2D number of positions should be lesser or equal to 4");
if constexpr (P == 2) {
- const QuadratureFormula<1>& lN = QuadratureManager::instance().getLineFormula(GaussQuadratureDescriptor(N));
+ const QuadratureFormula<1>& lN =
+ QuadratureManager::instance().getLineFormula(GaussLegendreQuadratureDescriptor(N + 1));
const LineTransformation<2> t{positions[0], positions[1]};
double value = 0.;
for (size_t i = 0; i < lN.numberOfPoints(); ++i) {
@@ -635,9 +637,10 @@ integrate(const PolynomialP<N, Dimension> Q, const std::array<TinyVector<Dimensi
}
integral = t.jacobianDeterminant() * value;
} else {
- const QuadratureFormula<2>& lN = QuadratureManager::instance().getSquareFormula(GaussQuadratureDescriptor(N));
- auto point_list = lN.pointList();
- auto weight_list = lN.weightList();
+ const QuadratureFormula<2>& lN =
+ QuadratureManager::instance().getSquareFormula(GaussLegendreQuadratureDescriptor(N + 1));
+ auto point_list = lN.pointList();
+ auto weight_list = lN.weightList();
SquareTransformation<2> s{positions[0], positions[1], positions[2], positions[3]};
auto value = weight_list[0] * s.jacobianDeterminant(point_list[0]) * Q(s(point_list[0]));
for (size_t i = 1; i < weight_list.size(); ++i) {
diff --git a/src/analysis/TaylorPolynomial.hpp b/src/analysis/TaylorPolynomial.hpp
index 1406a5989a4393e90933f4e59ff232f793311082..19ceec686a5f18728506f3daba9e27daa888fbcb 100644
--- a/src/analysis/TaylorPolynomial.hpp
+++ b/src/analysis/TaylorPolynomial.hpp
@@ -3,6 +3,7 @@
#include <algebra/TinyVector.hpp>
#include <analysis/CubeGaussQuadrature.hpp>
+#include <analysis/GaussLegendreQuadratureDescriptor.hpp>
#include <analysis/GaussQuadratureDescriptor.hpp>
#include <analysis/QuadratureManager.hpp>
#include <analysis/SquareGaussQuadrature.hpp>
@@ -169,7 +170,7 @@ class TaylorPolynomial
PUGS_INLINE
constexpr double
- operator[](const TinyVector<Dimension, size_t> relative_pos) const
+ operator[](const TinyVector<Dimension, size_t>& relative_pos) const
{
size_t total_degree = 0;
for (size_t i = 0; i < Dimension; ++i) {
@@ -179,8 +180,7 @@ class TaylorPolynomial
}
Assert((total_degree <= N), "The sum of the degrees of the coefficient you are looking for is greater than the "
"degree of the polynomial");
- TinyVector<size_coef> absolute_coefs = this->coefficients();
- size_t absolute_position = 0;
+ size_t absolute_position = 0;
if constexpr (Dimension == 1) {
absolute_position = relative_pos[0];
} else if constexpr (Dimension == 2) {
@@ -195,12 +195,12 @@ class TaylorPolynomial
total_sub_degree * (total_sub_degree + 1) / 2 + relative_pos[2];
}
- return absolute_coefs[absolute_position];
+ return m_coefficients[absolute_position];
}
PUGS_INLINE
- constexpr double
- operator[](const TinyVector<Dimension, size_t> relative_pos)
+ constexpr double&
+ operator[](const TinyVector<Dimension, size_t>& relative_pos)
{
size_t total_degree = 0;
for (size_t i = 0; i < Dimension; ++i) {
@@ -210,8 +210,7 @@ class TaylorPolynomial
}
Assert((total_degree <= N), "The sum of the degrees of the coefficient you are looking for is greater than the "
"degree of the polynomial");
- TinyVector<size_coef> absolute_coefs = this->coefficients();
- size_t absolute_position = 0;
+ size_t absolute_position = 0;
if constexpr (Dimension == 1) {
absolute_position = relative_pos[0];
} else if constexpr (Dimension == 2) {
@@ -226,7 +225,7 @@ class TaylorPolynomial
total_sub_degree * (total_sub_degree + 1) / 2 + relative_pos[2];
}
- return absolute_coefs[absolute_position];
+ return m_coefficients[absolute_position];
}
PUGS_INLINE
@@ -525,12 +524,10 @@ class TaylorPolynomial
<< "(y - " << P.x0()[1] << ")^" << j;
} else if (j == 1) {
os << "(x - " << P.x0()[0] << ")"
- << "^i"
- << "(y - " << P.x0()[1] << ")";
+ << "^" << i << "(y - " << P.x0()[1] << ")";
} else {
os << "(x - " << P.x0()[0] << ")"
- << "^i"
- << "(y - " << P.x0()[1] << ")^" << j;
+ << "^" << i << "(y - " << P.x0()[1] << ")^" << j;
}
}
all_coef_zero = false;
@@ -663,9 +660,10 @@ integrate(const TaylorPolynomial<N, Dimension> Q, const std::array<TinyVector<Di
}
integral = t.jacobianDeterminant() * value;
} else {
- const QuadratureFormula<2>& lN = QuadratureManager::instance().getSquareFormula(GaussQuadratureDescriptor(N));
- auto point_list = lN.pointList();
- auto weight_list = lN.weightList();
+ const QuadratureFormula<2>& lN =
+ QuadratureManager::instance().getSquareFormula(GaussLegendreQuadratureDescriptor(N + 1));
+ auto point_list = lN.pointList();
+ auto weight_list = lN.weightList();
SquareTransformation<2> s{positions[0], positions[1], positions[2], positions[3]};
auto value = weight_list[0] * s.jacobianDeterminant(point_list[0]) * Q(s(point_list[0]));
for (size_t i = 1; i < weight_list.size(); ++i) {
diff --git a/tests/test_PolynomialP.cpp b/tests/test_PolynomialP.cpp
index 7b46193c38434c2400a34fa0248f143461a2adac..bd6d4a82d57f16447ffdb70bba302eee1e9267e5 100644
--- a/tests/test_PolynomialP.cpp
+++ b/tests/test_PolynomialP.cpp
@@ -1,5 +1,6 @@
#include <Kokkos_Core.hpp>
#include <algebra/TinyMatrix.hpp>
+#include <analysis/GaussLegendreQuadratureDescriptor.hpp>
#include <analysis/GaussQuadratureDescriptor.hpp>
#include <analysis/PolynomialP.hpp>
#include <analysis/QuadratureManager.hpp>
@@ -167,10 +168,12 @@ TEST_CASE("PolynomialP", "[analysis]")
SECTION("integrate")
{
TinyVector<6> coef(1, -2, 10, 7, 2, 9);
+ TinyVector<10> coef2(1, -2, 10, 7, 2, 9, 0, 0, 0, 1);
std::array<TinyVector<2>, 4> positions;
std::array<TinyVector<2>, 3> positions2;
std::array<TinyVector<2>, 2> positions4;
std::array<TinyVector<2>, 4> positions3;
+ std::array<TinyVector<2>, 4> positions5;
positions[0] = TinyVector<2>{0, 0};
positions[1] = TinyVector<2>{0, 0.5};
positions[2] = TinyVector<2>{0.3, 0.7};
@@ -184,21 +187,40 @@ TEST_CASE("PolynomialP", "[analysis]")
positions3[1] = TinyVector<2>{1, 0};
positions3[2] = TinyVector<2>{1, 1};
positions3[3] = TinyVector<2>{0, 1};
+ positions5[0] = TinyVector<2>{0, 0};
+ positions5[1] = TinyVector<2>{1, 0};
+ positions5[2] = TinyVector<2>{1, 3};
+ positions5[3] = TinyVector<2>{1, 1};
PolynomialP<2, 2> P(coef);
+ PolynomialP<3, 2> Q(coef2);
auto p1 = [](const TinyVector<2>& X) {
const double x = X[0];
const double y = X[1];
return 1 - 2. * x + 10 * y + 7 * x * x + 2 * x * y + 9 * y * y;
};
- const QuadratureFormula<2>& l3 = QuadratureManager::instance().getSquareFormula(GaussQuadratureDescriptor(3));
- auto point_list = l3.pointList();
- auto weight_list = l3.weightList();
+ auto q1 = [](const TinyVector<2>& X) {
+ const double x = X[0];
+ const double y = X[1];
+ return 1 - 2. * x + 10 * y + 7 * x * x + 2 * x * y + 9 * y * y + y * y * y;
+ };
+ const QuadratureFormula<2>& l3 =
+ QuadratureManager::instance().getSquareFormula(GaussLegendreQuadratureDescriptor(4));
+ auto point_list = l3.pointList();
+ auto weight_list = l3.weightList();
SquareTransformation<2> s{positions[0], positions[1], positions[2], positions[3]};
- auto value = weight_list[0] * s.jacobianDeterminant(point_list[0]) * p1(s(point_list[0]));
+ auto value = weight_list[0] * s.jacobianDeterminant(point_list[0]) * q1(s(point_list[0]));
for (size_t i = 1; i < weight_list.size(); ++i) {
- value += weight_list[i] * s.jacobianDeterminant(point_list[i]) * p1(s(point_list[i]));
+ value += weight_list[i] * s.jacobianDeterminant(point_list[i]) * q1(s(point_list[i]));
}
+ // const QuadratureFormula<2>& l3bis =
+ // QuadratureManager::instance().getSquareFormula(GaussLegendreQuadratureDescriptor(4));
+ // auto point_listbis = l3bis.pointList();
+ // auto weight_listbis = l3bis.weightList();
+ // auto valuebis = weight_listbis[0] * s.jacobianDeterminant(point_listbis[0]) * p1(s(point_listbis[0]));
+ // for (size_t i = 1; i < weight_listbis.size(); ++i) {
+ // valuebis += weight_listbis[i] * s.jacobianDeterminant(point_listbis[i]) * p1(s(point_listbis[i]));
+ // }
const QuadratureFormula<2>& t3 = QuadratureManager::instance().getTriangleFormula(GaussQuadratureDescriptor(3));
auto point_list2 = t3.pointList();
auto weight_list2 = t3.weightList();
@@ -208,7 +230,7 @@ TEST_CASE("PolynomialP", "[analysis]")
value2 += weight_list2[i] * p1(t(point_list2[i]));
}
value2 *= t.jacobianDeterminant();
- const QuadratureFormula<1>& l1 = QuadratureManager::instance().getLineFormula(GaussQuadratureDescriptor(2));
+ const QuadratureFormula<1>& l1 = QuadratureManager::instance().getLineFormula(GaussLegendreQuadratureDescriptor(2));
const LineTransformation<2> u{positions4[0], positions4[1]};
double value4 = 0.;
for (size_t i = 0; i < l1.numberOfPoints(); ++i) {
@@ -220,10 +242,12 @@ TEST_CASE("PolynomialP", "[analysis]")
value3 += weight_list[i] * s3.jacobianDeterminant(point_list[i]) * p1(s3(point_list[i]));
}
- REQUIRE(value == Catch::Approx(integrate(P, positions)));
+ REQUIRE(value == Catch::Approx(integrate(Q, positions)));
REQUIRE(value2 == Catch::Approx(integrate(P, positions2)));
REQUIRE(value3 == Catch::Approx(integrate(P, positions3)));
REQUIRE(value4 == Catch::Approx(integrate(P, positions4)));
+ // REQUIRE(valuebis == Catch::Approx(integrate(P, positions)));
+ // REQUIRE(valuebis == value);
}
// // SECTION("product")
diff --git a/tests/test_TaylorPolynomial.cpp b/tests/test_TaylorPolynomial.cpp
index b5871d246d4841c50e55f4618f93d731c7fe1ea4..4d9149232202adf49de64107fb3bfbe8e20eed22 100644
--- a/tests/test_TaylorPolynomial.cpp
+++ b/tests/test_TaylorPolynomial.cpp
@@ -160,6 +160,7 @@ TEST_CASE("TaylorPolynomial", "[analysis]")
SECTION("integrate")
{
TinyVector<6> coef(1, -2, 10, 7, 2, 9);
+ TinyVector<10> coef2(1, -2, 10, 7, 2, 9, 0, 0, 0, 1);
std::array<TinyVector<2>, 4> positions;
std::array<TinyVector<2>, 3> positions2;
std::array<TinyVector<2>, 4> positions3;
@@ -178,41 +179,60 @@ TEST_CASE("TaylorPolynomial", "[analysis]")
positions4[0] = TinyVector<2>{0, 0.5};
positions4[1] = TinyVector<2>{0.3, 0.7};
TaylorPolynomial<2, 2> P(coef, x0);
+ TaylorPolynomial<2, 2> Q(coef, zero);
+ TaylorPolynomial<3, 2> R(coef2, x0);
auto p1 = [](const TinyVector<2>& X) {
+ const double x = X[0];
+ const double y = X[1];
+ return 1 - 2. * (x - 1.) + 10 * (y + 1) + 7 * (x - 1.) * (x - 1) + 2 * (x - 1) * (y + 1) + 9 * (y + 1) * (y + 1) +
+ (y + 1) * (y + 1) * (y + 1);
+ };
+ auto p2 = [](const TinyVector<2>& X) {
const double x = X[0];
const double y = X[1];
return 1 - 2. * (x - 1.) + 10 * (y + 1) + 7 * (x - 1.) * (x - 1) + 2 * (x - 1) * (y + 1) + 9 * (y + 1) * (y + 1);
};
- const QuadratureFormula<2>& l3 = QuadratureManager::instance().getSquareFormula(GaussQuadratureDescriptor(3));
- auto point_list = l3.pointList();
- auto weight_list = l3.weightList();
+ auto q1 = [](const TinyVector<2>& X) {
+ const double x = X[0];
+ const double y = X[1];
+ return 1 - 2. * x + 10 * y + 7 * x * x + 2 * x * y + 9 * y * y;
+ };
+ const QuadratureFormula<2>& l3 =
+ QuadratureManager::instance().getSquareFormula(GaussLegendreQuadratureDescriptor(4));
+ auto point_list = l3.pointList();
+ auto weight_list = l3.weightList();
SquareTransformation<2> s{positions[0], positions[1], positions[2], positions[3]};
auto value = weight_list[0] * s.jacobianDeterminant(point_list[0]) * p1(s(point_list[0]));
for (size_t i = 1; i < weight_list.size(); ++i) {
value += weight_list[i] * s.jacobianDeterminant(point_list[i]) * p1(s(point_list[i]));
}
+ auto valuebis = weight_list[0] * s.jacobianDeterminant(point_list[0]) * q1(s(point_list[0]));
+ for (size_t i = 1; i < weight_list.size(); ++i) {
+ valuebis += weight_list[i] * s.jacobianDeterminant(point_list[i]) * q1(s(point_list[i]));
+ }
SquareTransformation<2> s3{positions3[0], positions3[1], positions3[2], positions3[3]};
- auto value3 = weight_list[0] * s3.jacobianDeterminant(point_list[0]) * p1(s3(point_list[0]));
+ auto value3 = weight_list[0] * s3.jacobianDeterminant(point_list[0]) * p2(s3(point_list[0]));
for (size_t i = 1; i < weight_list.size(); ++i) {
- value3 += weight_list[i] * s3.jacobianDeterminant(point_list[i]) * p1(s3(point_list[i]));
+ value3 += weight_list[i] * s3.jacobianDeterminant(point_list[i]) * p2(s3(point_list[i]));
}
const QuadratureFormula<2>& t3 = QuadratureManager::instance().getTriangleFormula(GaussQuadratureDescriptor(3));
auto point_list2 = t3.pointList();
auto weight_list2 = t3.weightList();
TriangleTransformation<2> t{positions2[0], positions2[1], positions2[2]};
- auto value2 = weight_list2[0] * p1(t(point_list2[0]));
+ auto value2 = weight_list2[0] * p2(t(point_list2[0]));
for (size_t i = 1; i < weight_list2.size(); ++i) {
- value2 += weight_list2[i] * p1(t(point_list2[i]));
+ value2 += weight_list2[i] * p2(t(point_list2[i]));
}
value2 *= t.jacobianDeterminant();
const QuadratureFormula<1>& l1 = QuadratureManager::instance().getLineFormula(GaussQuadratureDescriptor(2));
const LineTransformation<2> u{positions4[0], positions4[1]};
double value4 = 0.;
for (size_t i = 0; i < l1.numberOfPoints(); ++i) {
- value4 += l1.weight(i) * u.velocityNorm() * p1(u(l1.point(i)));
+ value4 += l1.weight(i) * u.velocityNorm() * p2(u(l1.point(i)));
}
- REQUIRE(value == Catch::Approx(integrate(P, positions)));
+ REQUIRE(value == Catch::Approx(integrate(R, positions)));
+ REQUIRE(valuebis == Catch::Approx(integrate(Q, positions)));
REQUIRE(value2 == Catch::Approx(integrate(P, positions2)));
REQUIRE(value3 == Catch::Approx(integrate(P, positions3)));
REQUIRE(value4 == Catch::Approx(integrate(P, positions4)));