Abstract
This chapter describes the approximate solution of infinite-dimensional optimization problems by the “Extended Ritz Method” (ERIM). The ERIM consists in substituting the admissible functions with fixed-structure parametrized (FSP) functions containing vectors of “free” parameters. The larger the dimensions, the more accurate the approximations of the optimal solutions of the original functional optimization problems. This requires solving easier nonlinear programming problems. In the area of function approximation, we review the definition of approximating sequences of sets, which enjoy the property of density in the sets of functions one wants to approximate. Then, we provide the definition of polynomially complex approximating sequences of sets, which are able to approximate functions provided with suitable regularity properties by using, for a desired arbitrary accuracy, a number of “free” parameters increasing at most polynomially when the number of function arguments grows. In the less studied area of approximate solution of infinite-dimensional optimization problems, the optimizing sequences and the polynomially complex optimizing sequences of FSP functions are defined. Results are presented that allow to conclude that, if appropriate hypotheses occur, polynomially complex approximating sequences of sets give rise to polynomially complex optimizing sequences of FSP functions, possibly mitigating the curse of dimensionality.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
Four sequences of approximating FSP functions are defined and discussed. Such sequences are connected to each other by properties that the reader may find somewhat difficult and tedious. To better understand the relationships among the four families, the reader may refer to Tables 2.1 and 2.2, and to Fig. 2.10.
- 2.
The reader should now understand why Remark on Notation 2.1 is useful to simplify the notation; see Point 3 in the remark.
- 3.
Note that one should specify the space \({\mathscr {G}}^d\) in the definition, since the same set \({\mathcal {M}}^d\) might be considered as subset of various normed linear spaces. We refer the reader to Chap. 3 for a more detailed treatment.
- 4.
In Remark 2.2, we stressed the meaning of the term “absolute constant,” i.e., a constant that does not depend on any other quantity involved in the context at issue. As regards \({\tau }\), \({\kappa }_1^-, {\kappa }_1^+,\kappa _2^-, {\kappa }_2^+\), we have to point out that such “constants” are not absolute. Indeed, they may depend on d.
- 5.
The functions \(L_{{\tau }}\) and \(U_{{\tau }}\) take on the form of the “comparison functions” of class \({{\mathcal {K}}}_{\infty }\) typically used in Lyapunov stability analysis of nonlinear dynamic systems (see, e.g., [30]).
- 6.
It is worth mentioning also the work [31], where suboptimal feedback control laws are searched for dynamic systems under LQ assumptions via the Ritz method. However, in [31] the point of view is different: the authors aim at deriving, for an optimal control problem whose solution is known in closed form, an approximate control law taking on a simple structure.
References
Alessandri A, Cervellera C, Sanguineti M (2007) Functional optimal estimation problems and their solution by nonlinear approximation schemes. J Optim Theory Appl 134:445–466
Alessandri A, Gnecco G, Sanguineti M (2010) Minimizing sequences for a family of functional optimal estimation problems. J Optim Theory Appl 147:243–262
Alessandri A, Gaggero M, Zoppoli R (2012) Feedback optimal control of distributed parameter systems by using finite-dimensional approximation schemes. IEEE Trans Neural Netw Learn Syst 23:984–996
Alt W (1984) On the approximation of infinite optimization problems with an application to optimal control problems. Appl Math Optim 12:15–27
Barron AR (1992) Neural net approximation. In: Narendra KS (ed) Proceedings of the 7th Yale workshop on adaptive and learning systems. Yale University Press, pp 69–72
Barron AR (1993) Universal approximation bounds for superpositions of a sigmoidal function. IEEE Trans Inf Theory 39:930–945
Breiman L (1993) Hinging hyperplanes for regression, classification, and function approximation. IEEE Trans Inf Theory 39:993–1013
Dal Maso G (1993) Introduction to gamma-convergence. Springer
Darken C, Donahue M, Gurvits L, Sontag E (1993) Rate of approximation results motivated by robust neural network learning. In: Proceedings of the sixth annual ACM conference on computational learning theory. ACM, pp 303–309
Donahue M, Gurvits L, Darken C, Sontag E (1997) Rates of convex approximation in non-Hilbert spaces. Constr Approx 13:187–220
Gaggero M, Gnecco G, Sanguineti M (2014) Suboptimal policies for stochastic \(N\)-stage optimization problems: accuracy analysis and a case study from optimal consumption. In: El Ouardighi F, Kogan K (eds) Models and methods in economics and management. Springer, pp 27–50
Gelfand IM, Fomin SV (1963) Calculus of variations. Prentice Hall
Girosi F, Poggio T (1990) Networks and the best approximation property. Biol Cybern 63:169–176
Girosi F, Anzellotti G (1993) Rates of convergence for Radial Basis Functions and neural networks. In: Mammone RJ (ed) Artificial neural networks for speech and vision. Chapman & Hall, pp 97–113
Giulini S, Sanguineti M (2009) Approximation schemes for functional optimization problems. J Optim Theory Appl 140:33–54
Gnecco G, Sanguineti M (2009) Accuracy of suboptimal solutions to kernel principal component analysis. Comput Optim Appl 42:265–287
Gnecco G, Sanguineti M (2010) Error bounds for suboptimal solutions to kernel principal component analysis. Optim Lett 4:197–210
Gnecco G, Sanguineti M (2010) Estimates of variation with respect to a set and applications to optimization problems. J Optim Theory Appl 145:53–75
Gnecco G, Sanguineti M (2010) Regularization techniques and suboptimal solutions to optimization problems in learning from data. Neural Comput 22:793–829
Gnecco G, Sanguineti M (2010) Suboptimal solutions to dynamic optimization problems via approximations of the policy functions. J Optim Theory Appl 146:764–794
Gnecco G, Sanguineti M (2011) Team optimization problems with Lipschitz continuous strategies. Optim Lett 5:333–346
Gnecco G, Sanguineti M (2012) New insights into Witsenhausen’s counterexample. Optim Lett 6:1425–1446
Gnecco G, Sanguineti M, Gaggero M (2012) Suboptimal solutions to team optimization problems with stochastic information structure. SIAM J Optim 22:212–243
Gurvits L, Koiran P (1997) Approximation and learning of convex superpositions. J Comput Syst Sci 55:161–170
Jones LK (1992) A simple lemma on greedy approximation in Hilbert space and convergence rates for projection pursuit regression and neural network training. Ann Stat 20:608–613
Kainen P, Kůrková V, Sanguineti M (2003) Minimization of error functionals over variable-basis functions. SIAM J Optim 14:732–742
Kainen PC, Kůrková V (2005) Rates of minimization of error functionals over Boolean variable-basis functions. J Math Model Algorithms 4:355–368
Kainen PC, Kůrková V (2009) Complexity of Gaussian radial basis networks approximating smooth functions. J Complex 25:63–74
Kainen PC, Kůrková V (2012) Dependence of computational models on input dimension: tractability of approximation and optimization tasks. IEEE Trans Inf Theory 58:1203–1214
Khalil HK (1996) Nonlinear systems. Prentice-Hall
Kleinman DL, Athans M (1968) The design of suboptimal linear time-varying systems. IEEE Trans Autom Control 13:150–159
Knuth DE (1976) Big omicron and big omega and big theta. SIGACT News
Kůrková V (2008) Minimization of error functionals over perceptron networks. Neural Comput 20:252–270
Kůrková V, Sanguineti M (2005) Error estimates for approximate optimization by the Extended Ritz Method. SIAM J Optim 15:461–487
Kůrková V, Savický P, Hlavácková K (1998) Representations and rates of approximation of real-valued Boolean functions by neural networks. Neural Netw 11:651–659
Makovoz Y (1996) Random approximants and neural networks. J Approx Theory 85:98–109
Mhaskar HN (2004) On the tractability of multivariate integration and approximation by neural networks. J Complex 20:561–590
Parisini T, Zoppoli R (2009) Connections between approximating sequences and optimizing sequences in the Extended Ritz Method. Technical Report 2009.1, DIPTEM, Università di Genova
Traub JF, Werschulz AG (1999) Complexity and information. Cambridge University Press
Wasilkowski GW, Woźniakowski H (2001) Complexity of weighted approximation over \(\Re ^d\). J Complex 17:722–740
Woźniakowski H (1994) Tractability and strong tractability of linear multivariate problems. J Complex 10:96–128
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Zoppoli, R., Sanguineti, M., Gnecco, G., Parisini, T. (2020). From Functional Optimization to Nonlinear Programming by the Extended Ritz Method. In: Neural Approximations for Optimal Control and Decision. Communications and Control Engineering. Springer, Cham. https://doi.org/10.1007/978-3-030-29693-3_2
Download citation
DOI: https://doi.org/10.1007/978-3-030-29693-3_2
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-29691-9
Online ISBN: 978-3-030-29693-3
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)