On the conditions for the finite termination of ADMM and its applications to SOS polynomials feasibility problems

We study finite termination properties of the alternating direction method of multipliers (ADMM) method applied to semidefinite programs (SDPs) generated from sums of squares (SOS) feasibility problems. Expressing a polynomial as SOS of lower degree by formulating the problem as SDPs is a key problem in many fields, and ADMM is frequently used to efficiently solve the SDPs whose size grows very rapidly with the degree and number of variables of the polynomial. We present conditions for the ADMM method to converges to an optimal solution in finite iterations and prove its finite termination under the conditions. In addition, for the problem of representing a univariate trigonometric polynomial as an SOS, we also provide similar conditions for the finite termination of the ADMM at an optimal solution. Numerical results demonstrate the finite termination if the conditions are satisfied and the size of the strictly feasible region is not too small. The size is determined by solving an SDP whose optimal value indicates how much the variable matrix of the original SDP can be diagonally increased, without violating the constraints of the original SDP. The finite termination discussed in this paper is a distinctive property of ADMM, and cannot be observed when implementing the interior-point methods.

[1]  Pablo A. Parrilo,et al.  Computing sum of squares decompositions with rational coefficients , 2008 .

[2]  Katta G. Murty,et al.  Some NP-complete problems in quadratic and nonlinear programming , 1987, Math. Program..

[3]  Pablo A. Parrilo,et al.  Semidefinite programming relaxations for semialgebraic problems , 2003, Math. Program..

[4]  Grigoriy Blekherman.  There are significantly more nonegative polynomials than sums of squares , 2003, math/0309130.

[5]  Kim-Chuan Toh,et al.  Solving semidefinite-quadratic-linear programs using SDPT3 , 2003, Math. Program..

[6]  B. Dumitrescu,et al.  Trigonometric Polynomials Positive on Frequency Domains and Applications to 2-D FIR Filter Design , 2006, IEEE Transactions on Signal Processing.

[7]  Jiawang Nie,et al.  An exact Jacobian SDP relaxation for polynomial optimization , 2010, Math. Program..

[8]  Franz Rendl,et al.  Regularization Methods for Semidefinite Programming , 2009, SIAM J. Optim..

[9]  M. Fortin,et al.  Augmented Lagrangian methods : applications to the numerical solution of boundary-value problems , 1983 .

[10]  Kim-Chuan Toh,et al.  SDPNAL$$+$$+: a majorized semismooth Newton-CG augmented Lagrangian method for semidefinite programming with nonnegative constraints , 2014, Math. Program. Comput..

[11]  Pablo A. Parrilo,et al.  Sampling Algebraic Varieties for Sum of Squares Programs , 2015, SIAM J. Optim..

[12]  Jos F. Sturm,et al.  A Matlab toolbox for optimization over symmetric cones , 1999 .

[13]  Li Wang,et al.  Semidefinite Relaxations for Best Rank-1 Tensor Approximations , 2013, SIAM J. Matrix Anal. Appl..

[14]  Makoto Yamashita,et al.  Latest Developments in the SDPA Family for Solving Large-Scale SDPs , 2012 .

[15]  Wotao Yin,et al.  Alternating direction augmented Lagrangian methods for semidefinite programming , 2010, Math. Program. Comput..

[16]  Franz Rendl,et al.  A Boundary Point Method to Solve Semidefinite Programs , 2006, Computing.

[17]  Masakazu Kojima,et al.  Implementation and evaluation of SDPA 6.0 (Semidefinite Programming Algorithm 6.0) , 2003, Optim. Methods Softw..

[18]  Masakazu Muramatsu,et al.  Sums of Squares and Semidefinite Programming Relaxations for Polynomial Optimization Problems with Structured Sparsity , 2004 .

[19]  R. Glowinski,et al.  Numerical Methods for Nonlinear Variational Problems , 1985 .

[20]  Kim-Chuan Toh,et al.  A Convergent 3-Block SemiProximal Alternating Direction Method of Multipliers for Conic Programming with 4-Type Constraints , 2014, SIAM J. Optim..

[21]  Kazuo Murota,et al.  Exploiting Sparsity in Semidefinite Programming via Matrix Completion I: General Framework , 2000, SIAM J. Optim..

[22]  Lieven Vandenberghe,et al.  Multidimensional FIR Filter Design Via Trigonometric Sum-of-Squares Optimization , 2007, IEEE Journal of Selected Topics in Signal Processing.

[23]  Masakazu Kojima,et al.  Exploiting sparsity in linear and nonlinear matrix inequalities via positive semidefinite matrix completion , 2011, Math. Program..

[24]  P. Parrilo,et al.  From coefficients to samples: a new approach to SOS optimization , 2004, 2004 43rd IEEE Conference on Decision and Control (CDC) (IEEE Cat. No.04CH37601).