1: \begin{abstract}

2: This paper proposes and analyzes a dampened proximal alternating direction

3: method of multipliers (DP.ADMM) for solving linearly-constrained nonconvex

4: optimization problems where the smooth part of the objective function

5: is nonseparable. Each iteration of DP.ADMM consists of: (ii) a sequence

6: of partial proximal augmented Lagrangian (AL) updates, (ii) an under-relaxed

7: Lagrange multiplier update, and (iii) a novel test to check whether

8: the penalty parameter of the AL function should be updated. Under

9: a basic Slater point condition and some requirements on the dampening

10: factor and under-relaxation parameter, it is shown that DP.ADMM obtains

11: an approximate first-order stationary point of the constrained problem in ${\cal O}(\varepsilon^{-3})$

12: iterations for a given numerical tolerance $\varepsilon>0$. One of

13: the main novelties of the paper is that convergence of the method

14: is obtained without requiring any rank assumptions on the constraint

15: matrices. 

16: \end{abstract}

17: