@asadchev The following code returns an incorrect tensor
auto ein_rbl = TA::expressions::einsum(rbij("r,b,i,j"), rijl1("r,i,j,l"), "r,b,l");
value -= ein_rbl("r,b,l").dot(rbl1("r,b,l"));
The contractions are T("r,b,i,j") * G("r,i,j,l") = W("r,b,l")
I have verified this problem 2 ways. 1. I constructed the term W("r,b,l") using the vector_of_arrays.h code and I have computed value using the vector_of_arrays based method.
I have verified that the expressions are equivalent and I also checked to see if its related to block size and don't believe it is (I tried multiple systems and single and multiple blocks and see inconsistencies across the board)
@asadchev The following code returns an incorrect tensor
The contractions are
T("r,b,i,j") * G("r,i,j,l") = W("r,b,l")I have verified this problem 2 ways. 1. I constructed the term
W("r,b,l")using thevector_of_arrays.hcode and I have computed value using thevector_of_arraysbased method.I have verified that the expressions are equivalent and I also checked to see if its related to block size and don't believe it is (I tried multiple systems and single and multiple blocks and see inconsistencies across the board)