\`x^2+y_1+z_12^34\`
Advanced Search
Article Contents
Article Contents

Quadratic optimization with two ball constraints

  • * Corresponding author: Maziar Salahi

    * Corresponding author: Maziar Salahi

The authors are grateful to the reviewers for their comments and suggestions and University of Guilan for partially supporting this research

Abstract Full Text(HTML) Figure(1) / Table(7) Related Papers Cited by
  • In this paper, the minimization of a general quadratic function subject to two ball constraints, called two ball trust-region subproblem (TBTRS), is studied. It is shown that the global optimal solution can be found by solving two extended trust-region subproblems. Strong duality conditions for two special cases are discussed. Finally, a comparison of results of the new algorithm with the other two recently proposed algorithms and CVX software are presented for several classes of randomly generated test problems.

    Mathematics Subject Classification: Primary: 90C20, 90C26.

    Citation:

    \begin{equation} \\ \end{equation}
  • 加载中
  • Figure 1.   $ + $: global solution of $ Pc_1 $ and $ Pc_2 $, $ \Box $: LNGMs of $ Pc_1 $ and $ Pc_2 $

    Table 1.  Notations in the tables

    Notation Description
    n Dimension of problem
    Den Density of A
    CPU Run time
    $ F_{TB} $ Objective value of the new algorithm
    $ F_{AD} $ Objective value of ADMM algorithm [2]
    $ F_{SD} $ Objective value of SDP relaxation
    $ F_{SA} $ Objective value of Sakaue et. al's algorithm [25]
    $ x_{TB}^* $ Optimal solution of the new algorithm
    $ x_{SA}^* $ Optimal solution of Sakaue et. al's algorithm [25]
     | Show Table
    DownLoad: CSV

    Table 2.  Comparison of the new algorithm with ADMM algorithm [2] (first class)

    n Den CPU(TBTRS) CPU(ADMM) $ F_{TB}-F_{AD} $
    50 1 0.23 0.36 -9.31e-11
    100 1 0.28 0.47 -1.39 e-10
    200 1 0.55 0.79 2.33e-11
    300 1 0.80 1.75 1.86 e-10
    400 1 1.58 2.63 4.66e-10
    500 1 2.72 3.99 8.73e-10
    700 0.1 1.49 5.174 4.54e-10
    1000 0.1 3.27 20.64 3.49e-10
    2000 0.01 2.48 7.64 2.79e-10
    5000 0.001 2.74 6.59 -1.16e-10
    10000 0.0001 4.62 7.29 -1.46e-11
     | Show Table
    DownLoad: CSV

    Table 3.  Comparison of the new algorithm with Sakaue et. al's algorithm [25] (first class)

    n CPU(TBTRS) CPU(Sakaue et. al's algorithm [25]) $ F_{TB}-F_{SA} $ $ ||x^*_{TB}-x^*_{SA}|| $
    5 0.10 0.07 -2.27e-14 3.24e-14
    10 0.12 0.46 -1.71e-11 1.45e-13
    15 0.23 5.94 -3.87e-12 1.78e-14
    20 0.18 33.12 -1.82e-09 8.06e-08
    25 0.23 130.96 -2.96e-11 9.50e-14
    30 0.22 428.21 -6.58e-09 1.26e-10
     | Show Table
    DownLoad: CSV

    Table 4.  Comparison of the new algorithm with ADMM algorithm [2] (second class)

    n Den CPU(TBTRS) CPU(ADMM) $ F_{TB}-F_{AD} $
    50 1 0.091 2.46 -3.69e-08
    100 1 0.12 3.0022 -1.63e-07
    200 1 0.15 4.70 -3.81e-07
    300 1 0.24 6.71 -4.61e-07
    500 1 0.68 13.5 -6.15e-07
    700 1 1.17 21.43 -2.05e-06
    1000 1 2.82 47.35 -1.69e-06
    2000 0.1 2.39 39.16 -9.22e-07
    5000 0.01 2.96 55.92 -1.61e-06
    10000 0.001 3.14 132.33 -4.85e-06
     | Show Table
    DownLoad: CSV

    Table 5.  Comparison of the new algorithm with Sakaue et. al's algorithm [25] (second class)

    n CPU(TBTRS) CPU(Sakaue et. al's algorithm [25]) $ F_{TB}-F_{SA} $ $ ||x^*_{TB}-x^*_{SA}|| $
    5 0.06 0.051 1.42e-11 1.22e-06
    10 0.06 0.42 2.58e-10 3.69e-06
    15 0.12 6.18 -1.72e-10 2.89e-06
    20 0.08 35.03 2.84e-09 1.07e-05
    25 0.13 139.93 2.96e-09 1.02e-05
    30 0.14 449.08 4.73e-08 3.45e-05
     | Show Table
    DownLoad: CSV

    Table 6.  Comparison of the new algorithm with SDP relaxation (third class)

    n Den CPU(TBTRS) CPU(ADMM) CPU(CVX) $ F_{TB}-F_{AD} $ $ F_{TB}-F_{SD} $
    50 1 0.09 2.04 0.97 2.30e-09 -1.58e-05
    100 1 0.18 2.01 2.93 1.75e-08 -7.33e-05
    200 1 0.16 4.05 5.41 3.82e-07 -9.45e-04
    300 1 0.24 6.16 14.39 5.71e-07 -2.23e-4
    400 1 0.41 10.78 38.06 1.35e-06 -3.78e-4
    500 1 0.67 11.87 63.26 4.67e-07 -4.67e-4
    700 0.1 0.32 8.39 132.96 1.43e-07 -2.75e-4
    800 0.1 0.39 9.02 206.75 2.42e-07 -2.904e-4
    900 0.1 0.44 9.67 268.47 8.91e-07 -4.88e-4
    1000 0.1 0.52 11.21 427.06 3.29e-07 -5.71e-4
    2000 0.1 2.69 39.09 $ - $ 5.28e-05 $ - $
    3000 0.1 5.27 82.52 $ - $ 3.85e-05 $ - $
    5000 0.01 2.99 57.99 $ - $ 6.16e-05 $ - $
    10000 0.001 2.34 121.13 $ - $ 7.12e-05 $ - $
     | Show Table
    DownLoad: CSV

    Table 7.  Comparison of the new algorithm with Sakaue et. al's algorithm [25] (third class)

    n CPU(TBTRS) CPU(Sakaue et. al's algorithm [25]) $ F_{AD}-F_{SA} $ $ ||x^*_{AD}-x^*_{SA}|| $
    5 0.08 0.059 1.32e-13 9.59e-08
    10 0.06 0.36 1.86e-09 1.08e-05
    15 0.06 5.69 5.12e-09 1.66e-05
    20 0.07 32.62 1.25e-09 6.09e-06
    25 0.14 130.46 7.24e-10 3.75e-06
    30 0.14 419.58 8.38e-09 1.72e-05
     | Show Table
    DownLoad: CSV
  • [1] S. AdachiS. IwataY. Nakatsukasa and A. Takeda, Solving the trust-region subproblem by a generalized eigenvalue problem, SIAM Journal on Optimization, 27 (2017), 269-291.  doi: 10.1137/16M1058200.
    [2] S. Ansary Karbasy and M. Salahi, A hybrid algorithm for the two-trust-region subproblem, Computational and Applied Mathematics, https://doi.org/10.1007/s40314-019-0864-y doi: 10.1007/s40314-019-0864-y.
    [3] H. T. Banks and K. Kunisch, Estimation Techniques for Distributed Parameter Systems, Springer Science and Business Media, 2012. doi: 10.1007/978-1-4612-3700-6.
    [4] A. Beck and Y. C. Eldar, Strong duality in nonconvex quadratic optimization with two quadratic constraints, SIAM Journal on Optimization, 17 (2006), 844-860.  doi: 10.1137/050644471.
    [5] Im. Bomze and Ml. Overton, Narrowing the difficulty gap for the Celis-Dennis-Tapia problem, Mathematical Programming, 151 (2015), 459-476.  doi: 10.1007/s10107-014-0836-3.
    [6] S. Burer and K. M. Anstreicher, Second-order-cone constraints for extended trust-region subproblems, SIAM Journal on Optimization, 23 (2013), 432-451.  doi: 10.1137/110826862.
    [7] S. Burer and B. Yang, The trust-region subproblem with non-intersecting linear constraints, Mathematical Programming, 149 (2015), 253-264.  doi: 10.1007/s10107-014-0749-1.
    [8] M. R. Celis, J. E. Dennis and R. A. Tapia, A trust-region strategy for nonlinear equality constrained optimization, Numerical Optimization, (1984), 71–82.
    [9] A. R. Conn, N. I. Gould and P. L. Toint, Trust-Region Methods, SIAM Philadelphia, Vol: 1, 2000. doi: 10.1137/1.9780898719857.
    [10] J. E. Dennis and R. B. Schnabel, Numerical Methods for Unconstrained Optimization and Nonlinear Equations, SIAM publisher, Vol: 16, 1996. doi: 10.1137/1.9781611971200.
    [11] S. Fallahi, M. Salahi and S. Ansary Karbasy, On SOCP/SDP formulation of the extended trust-region subproblem, to appear in Iranian Journal of Operations Research, 2019. doi: 10.1007/s40314-019-0864-y.
    [12] M. Grant and S. Boyd, CVX: Matlab software for disciplined convex programming, version 2.1, March 2017. Available from: http://cvxr.com/cvx.
    [13] S. GugercinA. C. Antoulas and H. P. Zhang, An approach to identification for robust control, IEEE Transactions on Automatic Control, 48 (2003), 1109-1115.  doi: 10.1109/TAC.2003.812821.
    [14] M. Heinkenschloss, Mesh independence for nonlinear least squares problems with norm constraints, SlAM Journal on Optimization, 3 (1993), 81-117. doi: 10.1137/0803005.
    [15] M. Heinkenschloss, On the solution of a two ball trust-region subproblem, Mathematical Programming, 64 (1994), 249-276.  doi: 10.1007/BF01582576.
    [16] V. Jeyakumar and G. Y. Li, Trust-region problems with linear inequality constraints: Exact SDP relaxation, global optimality and robust optimization, Mathematical Programming, 147 (2014), 171-206.  doi: 10.1007/s10107-013-0716-2.
    [17] B. KaltenbacherF. Rendl and E. Resmerita, Computing quasisolutions of nonlinear inverse problems via efficient minimization of trust region problems, Journal of Inverse and Ill-Posed Problems, 24 (2016), 435-447.  doi: 10.1515/jiip-2015-0087.
    [18] C. Kravaris and J. H. Seinfeld, Identification of parameters in distributed parameter systems by regularization, SIAM Journal on Control and Optimization, 23 (1985), 217-241.  doi: 10.1137/0323017.
    [19] G. Kristensson, Inverse problems for acoustic waves using the penalised likelihood method, Inverse Problems, 2 (1986), 461.
    [20] J. M. Martínez, Local minimizers of quadratic functions on Euclidean balls and spheres, SIAM Journal on Optimization, 4 (1994), 159-176.  doi: 10.1137/0804009.
    [21] Y. Nesterov, H. Wolkowicz and Y. Ye, Semidefinite Programming Relaxations of Nonconvex Quadratic Optimization, Handbook of Semidefinite Programming, Springer, Boston, (2000), 361–419. doi: 10.1007/978-1-4615-4381-7_13.
    [22] S. Omatu and  J. H. SeinfeldDistributed Parameter Systems: Theory and Applications,, Clarendon Press, 1989. 
    [23] F. O'Sullivan and G. Wahba, A cross validated Bayesian retrieval algorithm for nonlinear remote sensing experiments, Journal of Computational Physics, 59 (1985), 441-455. 
    [24] J.-M. Peng and Y. Yuan, Optimality conditions for the minimization of a quadratic with two quadratic constraints, SIAM Journal on Optimization, 7 (1997), 579-594.  doi: 10.1137/S1052623494261520.
    [25] S. SakaueY. NakatsukasaA. Takeda and S. Iwata, Solving generalized CDT problems via two-parameter eigenvalues, SIAM Journal on Optimization, 26 (2016), 1669-1694.  doi: 10.1137/15100624X.
    [26] M. Salahi and A. Taati, A fast eigenvalue approach for solving the trust-region subproblem with an additional linear inequality, Computational and Applied Mathematics, 37 (2018), 329-347.  doi: 10.1007/s40314-016-0347-3.
    [27] M. SalahiA. Taati and H. Wolkowicz, Local nonglobal minima for solving large scale extended trust-region subproblems, Computational Optimization and Applications, 66 (2016), 223-244.  doi: 10.1007/s10589-016-9867-4.
    [28] M. Salah and S. Fallahi, Trust-region subproblem with an additional linear inequality constraint, Optimization Letters, 10 (2016), 821-832.  doi: 10.1007/s11590-015-0957-5.
    [29] J. F. Sturm and S. Zhang, On cones of nonnegative quadratic functions, Mathematics of Operations Research, 28 (2003), 246-267.  doi: 10.1287/moor.28.2.246.14485.
    [30] C. R. Vogel, A constrained least squares regularization method for nonlinear iii-posed problems, SIAM Journal on Control and Optimization, 28 (1990), 34-49.  doi: 10.1137/0328002.
    [31] A. Zhang and S. Hayashi, Celis-Dennis-Tapia based approach to quadratic fractional programming problems with two quadratic constraints, Numerical Algebra, Control and Optimization, 1 (2011), 83-98.  doi: 10.3934/naco.2011.1.83.
  • 加载中

Figures(1)

Tables(7)

SHARE

Article Metrics

HTML views(1498) PDF downloads(475) Cited by(0)

Access History

Other Articles By Authors

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return