1: \begin{abstract}

2: Quadratic programs arise in robotics, communications, smart grids,

3: and many other applications. As these problems grow in size, finding

4: solutions becomes more computationally demanding, and new algorithms

5: are needed to efficiently solve them at massive scales. Targeting large-scale problems,

6: we develop a multi-agent quadratic programming framework in which

7: each agent updates only a small number of the total decision variables

8: in a problem. Agents communicate their updated values to each other,

9: though we do not impose any restrictions on the timing with which

10: they do so, nor on the delays in these transmissions. Furthermore,

11: we allow weak parametric coupling among agents, in the sense that

12: they are free to independently choose their stepsizes, subject to

13: mild restrictions. We further provide the means

14: for agents to independently regularize the problems they solve, thereby

15: improving convergence properties

16: while preserving agents' independence in selecting parameters and ensuring a global bound on regularization error is satisfied. Larger regularizations accelerate convergence but increase error in the solution obtained, and we quantify the tradeoff between convergence rates and quality of solutions. Simulation results are presented to illustrate these developments.

17: \end{abstract}

18: