Principal whitened gradient-projection algorithm for distribution control

  • 102 Accesses

  • 1 Citations


In this paper, we use an information geometric algorithm to solve the distribution control problem. The new designed algorithm is called principal whitened gradient-projection algorithm. In the principal natural gradient step, we use principal whitened gradient algorithm to obtain an optimal trajectory of the weight vector on the B-spline manifold from the viewpoint of information geometry. In the projection step, we project the selected points on B onto M. The coordinates of the projections on M give the trajectory of the control input u.

This is a preview of subscription content, log in to check access.

Access options

Buy single article

Instant unlimited access to the full article PDF.

US$ 39.95

Price includes VAT for USA


  1. 1

    Donson C, Wang H. Itertative approximation of statistical distribution and relation to information geometry. J Stat Infer Stoch Proc, 2001, 147: 307–318

  2. 2

    Gao Z, Wang H, Chai T. A robust fault detection filtering for stochastic distribution systems via descriptor estimator and parametric gain design. IET P-Contr Theor Ap, 2007, 1: 1286–1293

  3. 3

    Wang H. Detect unexpected changes of particle size distribution in paper-making white water system. In: Proceedings of IFAC Fault Detection and Diagnosis Lyon. New York: Marcel Dekker Inc, 1998. 78–84

  4. 4

    Wang H. Minimum entropy control of non-Gaussian dynamic stochastic systems. IEEE Trans Automat Contr, 2002, 47: 398–403

  5. 5

    Zhang Z, Sun H, Zhong F. Natural gradient-projection algorithm for distribution control. Optim Contr Appl Met, 2009, 30: 495–504

  6. 6

    Wang H, Baki H, Kabore P. Control of bounded dynamic stochastic distributions using squre root models: an applicability study in papermaking systems. Trans I Meas Control, 2001, 23: 51–68

  7. 7

    Wang H. Model reference adaptive control of the output stochastic distributions for unknown linear stochastic systems. Int J Syst Sci, 1999, 30: 707–715

  8. 8

    Wang H. Bounded Dynamic Stochastic Systems: Modeling and Control. London: Springer-Verlag, 2000

  9. 9

    Amari S. Differential Geometrical Methods in Statistics. New York/Tokyo: Springer-Verlag, 1985. 15–24

  10. 10

    Amari S. Natural gradient works efficiently in learning. Neural Comput, 1998, 10: 251–276

  11. 11

    Yang Z, Laaksonen J. Principal whitened gradient for information geometry. Neural Networks, 2008, 21: 232–240

  12. 12

    Amari S. Information geometry of the EM and em algorithm for neural networks. Neural Networks, 1995, 8: 1379–1408

  13. 13

    Amari S, Kurata K, Nagaoka H. Information geometry of Boltzmann machines. IEEE Trans Neural Networ, 1992, 3: 260–271

  14. 14

    Ohara A, Suda N, Amari S. Dualistic differential geometry of positive definite matrices and its applications to related problems. Linear Algebra Appl, 1996, 247: 31–53

Download references

Author information

Correspondence to ShiCheng Zhang.

Electronic supplementary material

Supplementary material, approximately 16.7 MB.

Supplementary material, approximately 16.7 MB.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Zhang, S., Sun, H. & Li, C. Principal whitened gradient-projection algorithm for distribution control. Sci. China Inf. Sci. 56, 1–8 (2013).

Download citation


  • principal whitened gradient
  • distribution control
  • Kullback divergence