1: \begin{abstract}

2: Recently, researchers showed that dirty paper coding (DPC) is the optimal transmission strategy for

3: multiple-input multiple-output broadcast channels (MIMO-BC). In this paper, we study how to

4: determine the maximum weighted sum of DPC rates through solving the maximum weighted sum rate

5: problem of the dual MIMO multiple access channel (MIMO-MAC) with a sum power constraint. We first

6: simplify the maximum weighted sum rate problem such that enumerating all possible decoding orders

7: in the dual MIMO-MAC is unnecessary. We then design an efficient algorithm based on conjugate

8: gradient projection (CGP) to solve the maximum weighted sum rate problem. Our proposed CGP method

9: utilizes the powerful concept of Hessian conjugacy. We also develop a rigorous algorithm to solve

10: the projection problem. We show that CGP enjoys provable convergence, nice scalability, and great

11: efficiency for large MIMO-BC systems.

12: \end{abstract}

13: