1: \begin{abstract}

2: \textcolor{black}{In this letter, the achievable rate maximization

3: problem is considered for intelligent reflecting surface (IRS) assisted

4: multiple-input multiple-output (MIMO) systems in an underlay spectrum

5: sharing scenario, subject to interference power constraints at the

6: primary users. The formulated non-convex optimization problem is challenging

7: to solve due to its non-convexity as well as coupling design variables

8: in the constraints. Different from existing works that are mostly

9: based on alternating optimization (AO), we propose a penalty }dual

10: decomposition based gradient projection (PDDGP) algorithm to solve

11: this problem. We also provide a convergence proof and a complexity

12: analysis for the proposed algorithm. We benchmark the proposed algorithm

13: against two known solutions, namely a minimum mean-square error based

14: AO algorithm and an inner approximation method with block coordinate

15: descent. Specifically, the complexity of the proposed algorithm \textcolor{black}{grows

16: }\textit{\textcolor{black}{linearly}}\textcolor{black}{{} with respect

17: to the number of reflecting elements at the IRS,} while that of the

18: two benchmark methods\textcolor{black}{{} grows with the third power

19: of the number of IRS elements.} Moreover, numerical results show that

20: the proposed PDDGP algorithm yields considerably higher achievable

21: rate than the benchmark solutions.

22: \end{abstract}

23: