1: \begin{abstract} {We consider strongly convex distributed optimization problems in  decentralized connected network. EFIX, the proposed method, fits into the framework of quadratic penalty methods. Using the standard reformulation of distributed optimization problems to constrained problems in larger space, a sequence of quadratic penalty problems with increasing penalty parameters is defined. For quadratic objective function the sequence consists of  quadratic penalty subproblems. For strongly convex generic functions the objective function is approximated with a quadratic model and hence the sequence of resulting penalty subproblems is again a sequence of quadratic problems. Each of the quadratic penalty problems is solved by a fixed point linear solver. The proposed method converges to the exact solution of the distributed problem in the case of quadratic objective function. We also prove the worst case complexity of order $\cal{O}(\epsilon^{-1})$. The convergence result for generic strongly convex function corresponds to the result in centralized optimization. Numerical results indicate that the method is highly competitive with the state-of-the-art first order exact methods, requires smaller computational and communication effort  and robust to the choice of inner parameters. The presented results are obtained with Jacobi Over-Relaxation method as the linear solver.  }

2: 

3: {\bf Key words:} Fixed point methods, quadratic penalty method, distributed optimization., strongly convex problems\\

4: %{\bf AMS subject classification.} 90C15, 62L20, 60H40 

5: \end{abstract}

6: