prodD() [4/5]

template<Rank summed, Rank frank, Rank srank, typename Scalar , std::enable_if_t<(int(frank)+int(srank) - 2 *int(summed) > 0), int > = 0>

GenericDenseTensor< frank+srank - 2 *summed, Scalar > syten::IdentityDenseTensorImpl::prodD ( IdentityDenseTensor< frank, Scalar > const &  a, CudaDenseTensor< srank, Scalar > const &  b, std::array< int, frank > const &  c_a, std::array< int, srank > const &  c_b, bool  conjugate = false, EliminateZeros const  ezeros = EliminateZeros::No, DenseProduct::TemporaryTransposeStorage< Scalar, frank, srank > *  = nullptr  )

inline

Overload for product of an identity dense tensor and a cuda dense tensor.

Againg rank-3 identities are taken care of by merge and split and others are casted into a DenseTensor and then loaded onto the GPU.

References syten::Cuda::copy(), if_constexpr, merge(), split(), and SYTEN_ASSERT_MSG.

Here is the call graph for this function: