Made AutoDiffJacobian more intuitive to use and updated for C++11

Changes:
* Removed unnecessary types from the Functor by inferring from its types
* Removed inputs() function reference, replaced with .rows()
* Updated the forward constructor to use variadic templates
* Added optional parameters to the Fuctor for passing parameters,
  control signals, etc
* Has been tested with fixed size and dynamic matricies

Ammendment by chtz: overload operator() for compatibility with not fully conforming compilers
This commit is contained in:
Emil Fresk
2016-09-16 14:03:55 +02:00
parent 4adeababf9
commit 6edd2e2851
2 changed files with 126 additions and 15 deletions

View File

@@ -105,6 +105,89 @@ struct TestFunc1
}
};
#if EIGEN_HAS_VARIADIC_TEMPLATES
/* Test functor for the C++11 features. */
template <typename Scalar>
struct integratorFunctor
{
typedef Matrix<Scalar, 2, 1> InputType;
typedef Matrix<Scalar, 2, 1> ValueType;
/*
* Implementation starts here.
*/
integratorFunctor(const Scalar gain) : _gain(gain) {}
integratorFunctor(const integratorFunctor& f) : _gain(f._gain) {}
const Scalar _gain;
template <typename T1, typename T2>
void operator() (const T1 &input, T2 *output, const Scalar dt) const
{
T2 &o = *output;
/* Integrator to test the AD. */
o[0] = input[0] + input[1] * dt * _gain;
o[1] = input[1] * _gain;
}
/* Only needed for the test */
template <typename T1, typename T2, typename T3>
void operator() (const T1 &input, T2 *output, T3 *jacobian, const Scalar dt) const
{
T2 &o = *output;
/* Integrator to test the AD. */
o[0] = input[0] + input[1] * dt * _gain;
o[1] = input[1] * _gain;
if (jacobian)
{
T3 &j = *jacobian;
j(0, 0) = 1;
j(0, 1) = dt * _gain;
j(1, 0) = 0;
j(1, 1) = _gain;
}
}
};
template<typename Func> void forward_jacobian_cpp11(const Func& f)
{
typedef typename Func::ValueType::Scalar Scalar;
typedef typename Func::ValueType ValueType;
typedef typename Func::InputType InputType;
typedef typename AutoDiffJacobian<Func>::JacobianType JacobianType;
InputType x = InputType::Random(InputType::RowsAtCompileTime);
ValueType y, yref;
JacobianType j, jref;
const Scalar dt = internal::random<double>();
jref.setZero();
yref.setZero();
f(x, &yref, &jref, dt);
//std::cerr << "y, yref, jref: " << "\n";
//std::cerr << y.transpose() << "\n\n";
//std::cerr << yref << "\n\n";
//std::cerr << jref << "\n\n";
AutoDiffJacobian<Func> autoj(f);
autoj(x, &y, &j, dt);
//std::cerr << "y j (via autodiff): " << "\n";
//std::cerr << y.transpose() << "\n\n";
//std::cerr << j << "\n\n";
VERIFY_IS_APPROX(y, yref);
VERIFY_IS_APPROX(j, jref);
}
#endif
template<typename Func> void forward_jacobian(const Func& f)
{
typename Func::InputType x = Func::InputType::Random(f.inputs());
@@ -128,7 +211,6 @@ template<typename Func> void forward_jacobian(const Func& f)
VERIFY_IS_APPROX(j, jref);
}
// TODO also check actual derivatives!
template <int>
void test_autodiff_scalar()
@@ -141,6 +223,7 @@ void test_autodiff_scalar()
VERIFY_IS_APPROX(res.value(), foo(p.x(),p.y()));
}
// TODO also check actual derivatives!
template <int>
void test_autodiff_vector()
@@ -151,7 +234,7 @@ void test_autodiff_vector()
VectorAD ap = p.cast<AD>();
ap.x().derivatives() = Vector2f::UnitX();
ap.y().derivatives() = Vector2f::UnitY();
AD res = foo<VectorAD>(ap);
VERIFY_IS_APPROX(res.value(), foo(p));
}
@@ -164,6 +247,9 @@ void test_autodiff_jacobian()
CALL_SUBTEST(( forward_jacobian(TestFunc1<double,3,2>()) ));
CALL_SUBTEST(( forward_jacobian(TestFunc1<double,3,3>()) ));
CALL_SUBTEST(( forward_jacobian(TestFunc1<double>(3,3)) ));
#if EIGEN_HAS_VARIADIC_TEMPLATES
CALL_SUBTEST(( forward_jacobian_cpp11(integratorFunctor<double>(10)) ));
#endif
}