×

The construction of high order convergent look-ahead finite difference formulas for Zhang neural networks. (English) Zbl 1425.65207

Summary: Zhang Neural Networks rely on convergent 1-step ahead finite difference formulas of which very few are known. Those which are known have been constructed in ad-hoc ways and suffer from low truncation error orders. This paper develops a constructive method to find convergent look-ahead finite difference schemes of higher truncation error orders. The method consists of seeding the free variables of a linear system comprised of Taylor expansion coefficients followed by a minimization algorithm for the maximal magnitude root of the formula’s characteristic polynomial. This helps us find new convergent 1-step ahead finite difference formulas of any truncation error order. Once a polynomial has been found with roots inside the complex unit circle and no repeated roots on it, the associated look-ahead ZNN discretization formula is convergent and can be used for solving any discretized ZNN based model. Our method recreates and validates the few known convergent formulas, all of which have truncation error orders at most 4. It also creates new convergent 1-step ahead difference formulas with truncation error orders 5 through 8.

MSC:

65Q10 Numerical methods for difference equations
65L12 Finite difference and finite volume methods for ordinary differential equations
92B20 Neural networks for/in biological studies, artificial life and related topics

Software:

fminsearch
PDFBibTeX XMLCite
Full Text: DOI arXiv

References:

[1] Bunse-Gerstner, A.; Byers, R.; Mehrmann, V.; Nichols, N. K., Numerical computation of an analytic singular value decomposition of a matrix valued function, Numer. Math., 60, 1-39 (1991) · Zbl 0743.65035 · doi:10.1007/BF01385712
[2] Engeln-Müllges, G.; Uhlig, F., Numerical Algorithms with C (1996), Springer: Springer, New York · Zbl 0857.65003
[3] Lagarias, J. C.; Reeds, J. A.; Wright, M. H.; Wright, P. E., Convergence properties of the Nelder-Mead simplex method in low dimensions, SIAM J. Optim., 9, 112-147 (1998) · Zbl 1005.90056 · doi:10.1137/S1052623496303470
[4] Li, J.; Mao, M.; Uhlig, F.; Zhang, Y., Z-type neural-dynamics for time-varying nonlinear optimization under a linear equality constraint with robot application, J. Comput. Appl. Math., 327, 155-166 (2018) · Zbl 1378.90077 · doi:10.1016/j.cam.2017.06.017
[5] Li, J., Mao, M., Uhlig, F., and Zhang, Y., A 5-instant finite difference formula to find discrete time-varying generalized matrix inverses and scalar reciprocals, 17 p., submitted. · Zbl 1434.65022
[6] Nelder, J. A.; Mead, R., A simplex method for function minimization, Comput. J., 7, 308-313 (1965) · Zbl 0229.65053 · doi:10.1093/comjnl/7.4.308
[7] Qiu, B., Zhang, Y., Guo, J., Yang, Z., and Li, X., New five-step DTZD algorithm for future nonlinear minimization with quartic steady-state error pattern, Numer. Algor. (2018), in print. Available at . · Zbl 1435.65096
[8] Qiu, B., Zhang, Y., and Yang, Z., New discrete-time ZNN models for least-squares solutions of dynamic linear equation systems with time-varying rank-deficient coefficients, IEEE Trans. Neural Netw. Learn. Syst., (2019), 12 p., to appear.
[9] Uhlig, F., Matlab code runconv1step(runs,jend,k,s) is available at .
[10] Uhlig, F., Zhang neural networks for fast and accurate computations of the field of values, submitted, 13 p. Preprint available as FOVviaZNNa.pdf at .
[11] Wang, X.-Z.; Wei, Y.; Stanimirovic, P. S., Complex neural network models for time-varying Drazin inverse, Neural Comput., 28, 2790-2824 (2016) · Zbl 1474.68157 · doi:10.1162/NECO_a_00866
[12] Wang, X.-Z.; Stanimirovic, P. S.; Wei, Y., Complex ZFs for computing time-varying complex outer inverses, Neurocomputing, 275, 983-1001 (2018) · doi:10.1016/j.neucom.2017.09.034
[13] Xiao, L.; Liao, B., A convergence-accelerated Zhang neural network and its solution application to Lyapunov equation, Neurocomputing, 193, 213-218 (2016) · doi:10.1016/j.neucom.2016.02.021
[14] Xiao, L.; Zhang, Z.; Zhang, Z.; Li, W.; Li, S., Design, verification and robotic application of a novel recurrent neural network for computing dynamic Sylvester equation, Neural Netw., 105, 185-196 (2018) · Zbl 1470.68044 · doi:10.1016/j.neunet.2018.05.008
[15] Xiao, L.; Liao, B.; Li, S.; Chen, K., Nonlinear recurrent neural networks for finite-time solution of general time-varying linear matrix equations, Neural Netw., 98, 102-113 (2018) · Zbl 1441.93265 · doi:10.1016/j.neunet.2017.11.011
[16] Xiao, L., Li, S., Lin, F.-J., Tan, Z., and Khan, A.H., Zeroing neural dynamics for control design: comprehensive analysis on stability, robustness, and convergence speed, IEEE Trans. Ind. Inform., (2019), in print, DOI:.
[17] Zhang, Y.; Guo, D., Zhang Functions and Various Models (2015), Springer · Zbl 1339.65002
[18] Zhang, Y.; Yi, C., Zhang Neural Networks and Neural-Dynamic Method (2011), Nova Science Publishers Inc.: Nova Science Publishers Inc., New York
[19] Zhang, Y.; Jiang, D.; Wang, J., A recurrent neural network for solving Sylvester equation with time-varying coefficients, IEEE Trans. Neural Netw., 13, 5, 1053-1063 (2002) · doi:10.1109/TNN.2002.1031938
[20] Zhang, Y.; Ke, Z.; Xu, P.; Yi, C., Time-varying square roots finding via Zhang dynamics versus gradient dynamics and the former’s link and new explanation to Newton-Raphson iteration, Inf. Process. Lett., 110, 24, 1103-1109 (2010) · Zbl 1380.65086 · doi:10.1016/j.ipl.2010.09.013
[21] Zhang, Y.; Yi, C.; Guo, D.; Zheng, J., Comparison on Zhang neural dynamics and gradient-based neural dynamics for online solution of nonlinear time-varying equation, Neural Comput. Appl., 20, 1, 1-7 (2011) · doi:10.1007/s00521-010-0452-y
[22] Zhang, Y.; Li, F.; Yang, Y.; Li, Z., Different Zhang functions leading to different Zhang-dynamics models illustrated via time-varying reciprocal solving, Appl. Math. Model., 36, 9, 4502-4511 (2012) · Zbl 1252.65222 · doi:10.1016/j.apm.2011.11.081
[23] Zhang, Y.; Yang, M.; Li, J.; He, L.; Wu, S., ZFD formula 4IgSFD_Y applied to future minimization, Phys. Lett. A, 381, 19, 1677-1681 (2017) · Zbl 1378.82042 · doi:10.1016/j.physleta.2017.03.025
[24] Zhang, Y., Yang, M., Li, C., Uhlig, F., and Hu, H., New continuous ZD model for computation of time-varying eigenvalues and corresponding eigenvectors, 17 p., submitted.
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.