1: \begin{abstract}

2: To estimate accurate voltage phasors from inaccurate voltage magnitude

3: and complex power measurements, the standard approach is to iteratively

4: refine a good initial guess using the Gauss-Newton method. But the

5: nonconvexity of the estimation makes the Gauss-Newton method sensitive

6: to its initial guess, so human intervention is needed to detect convergence

7: to plausible but ultimately spurious estimates. This paper makes a

8: novel connection between the angle estimation subproblem and phase

9: synchronization to yield two key benefits: (1) an exceptionally high

10: quality initial guess over the angles, known as a \emph{spectral initialization};

11: (2) a correctness guarantee for the estimated angles, known as a \emph{global

12: optimality certificate}. These are formulated as sparse eigenvalue-eigenvector

13: problems, which we efficiently compute in time comparable to a few

14: Gauss-Newton iterations. Our experiments on the complete set of Polish, PEGASE, and RTE

15: models show, where voltage magnitudes are already

16: reasonably accurate, that spectral initialization provides an almost-perfect

17: single-shot estimation of $n$ angles from $2n$ moderately noisy

18: bus power measurements (i.e. $n$ pairs of PQ measurements), whose

19: correctness becomes guaranteed after a single Gauss--Newton iteration. 

20: For less accurate voltage magnitudes, the performance of the method degrades gracefully; even with moderate voltage magnitude errors, the estimated voltage angles remain surprisingly accurate.

21: \end{abstract}

22: