1: \begin{abstract}

2: 	Low rank matrix recovery problems appear widely in

3: 	statistics, combinatorics, and imaging.

4: 	One celebrated method for solving these problems is

5: 	to formulate and solve a semidefinite program (SDP).

6: 	It is often known that the exact solution to the SDP

7: 	with perfect data recovers the solution to the original

8: 	low rank matrix recovery problem.

9: 	It is more challenging to show that an approximate solution to the SDP

10: 	formulated with noisy problem data acceptably solves the original problem;

11: 	arguments are usually ad hoc for each problem setting, and can be complex.

12: 	

13: 	In this note,

14: 	we identify a set of conditions that we call \emph{\newcontent{simplicity}} that

15: 	limit the error due to noisy problem data or incomplete convergence.

16: 	In this sense, \newcontent{simple} SDPs are robust: \newcontent{simple} SDPs can be (approximately)

17: 	solved efficiently at scale;

18: 	and the resulting approximate solutions, even with noisy data, can be trusted.

19: 	Moreover, we show that \newcontent{simplicity}  holds generically,

20: 	and also for many structured low rank matrix recovery problems,

21: 	including the stochastic block model, $\mathbb{Z}_2$ synchronization, and matrix completion.

22: 	Formally, we call an SDP \newcontent{simple} if it has a surjective constraint map,

23: 	admits a unique primal and dual solution pair, and satisfies strong duality and strict complementarity.

24: 	

25: 	However, \newcontent{simplicity}  is not a panacea:

26: 	we show the Burer-Monteiro formulation of the SDP

27: 	may have spurious second-order critical points,

28: 	even for a \newcontent{simple} SDP with a rank 1 solution.

29: \end{abstract}