prod() [3/6]

template<Rank summed, Rank frank, Rank srank>

std::enable_if_t< frank+srank - 2 *summed !=0, Tensor< frank+srank - 2 *summed > > syten::prod ( Tensor< frank > const &  a, Tensor< srank > const &  b, std::map< int, int > const &  m_a, std::map< int, int > const &  m_b, bool const  conj = false, EliminateZeros const  ezeros = EliminateZeros::No, SourceLocation  location = SourceLocation::current()  )

inline

Legacy variant 2 of the Tensor-Tensor-to-Tensor product.

Parameters

a	first tensor
b	second tensor
m_a	tensor target specification for first tensor
m_b	tensor target specification for second tensor
conj	if true, complex-conjugate the second tensor
ezeros	if Yes, eliminate zeros.
location	Callsite identifier.

Mappings m_a, m_b are translated into c_a/c_b by using the key of the map as 1-indexed location of the value in c_?. That is, a map 2 → 3, 1 → -2, 3 → 1 is equivalent to a product specification {-2, 3, 1}.

References syten::CudaDenseTensorImpl::conj(), and SYTEN_ASSERT.

Here is the call graph for this function: