AdjointContract

Perform $B := \sum_i A_i^H$, where the summation is performed over the local data of each member of the process team that was redundantly assigned entries of $A$ but is not redundantly assigned entries of $B$. Thus, in the general case where each column and row of $A$ is respectively distributed over the process sets $U_0 \times U_1$ and $V_0 \times V_1$, while each column and row of $B$ is respectively distributed over $U_0$ and $V_0$, then the result is of the form

$$B = \sum_{i \in U_1 \times V_1} A_i^H.$$

C++ API

void AdjointContract(const ElementalMatrix<T> &A, ElementalMatrix<T> &B)

void AdjointContract(const BlockMatrix<T> &A, BlockMatrix<T> &B)

C API

TODO

Python API

TODO