• Home
  • Mathematics
  • Stability of the Turnpike Phenomenon in Discrete-Time Optimal Control Problems

Stability of the Turnpike Phenomenon in Discrete-Time Optimal Control Problems

By Alexander J. Zaslavski

The constitution of approximate suggestions of self sustaining discrete-time optimum keep watch over difficulties and person turnpike effects for optimum keep an eye on difficulties with out convexity (concavity) assumptions are tested during this ebook. specifically, the booklet makes a speciality of the houses of approximate suggestions that are self sustaining of the size of the period, for all sufficiently huge durations; those effects observe to the so-called turnpike estate of the optimum keep watch over difficulties. by means of encompassing the so-called turnpike estate the approximate strategies of the problemsare decided essentially via the target functionality and are essentially self sufficient of the alternative of period and endpoint stipulations, other than in areas as regards to the endpoints. This bookalso explores the turnpike phenomenon for 2 huge sessions of self sustaining optimum regulate difficulties. it's illustrated that the turnpike phenomenon is strong for an optimum keep an eye on challenge if the corresponding limitless horizon optimum keep an eye on challenge possesses an asymptotic turnpike estate. If an optimum keep an eye on challenge belonging to the 1st type possesses the turnpike estate, then the turnpike is a singleton (unit set). the steadiness of the turnpike estate below small perturbations of an target functionality and of a constraint map is tested. For the second one type of difficulties the place the turnpike phenomenon isn't really unavoidably a singleton the steadiness of the turnpike estate less than small perturbations of an target functionality is demonstrated. Containing suggestions of adverse difficulties in optimum controland offering new techniques, ideas and techniques this publication is of curiosity formathematiciansworking in optimum keep watch over and the calculus of variations.It may also be beneficial in instruction classes for graduate students."

Show description

Quick preview of Stability of the Turnpike Phenomenon in Discrete-Time Optimal Control Problems PDF

Show sample text content

See, for instance, [45, fifty five, fifty six] and the references pointed out therein). For those sessions of difficulties a turnpike isn't unavoidably a singleton yet could in its place be an nonstationary trajectory (in the discrete time nonautonomous case) or a completely non-stop functionality at the period [0, ∞) (in the continual time nonautonomous case) or a compact subset of the distance X (in the self sustaining case). For periods of difficulties thought of in [45, 56], utilizing the Baire class technique, it was once proven that the turnpike estate holds for a widespread (typical) challenge.

Tk+1 } and t = Sk + i + 1. It follows from (2. 26), (2. 18), (2. 15), and (2. 1) that {xt }∞ t=0 is an (Ω)-program. relatives (2. 25), (2. 26), (2. 18), and (2. 15) suggest that for every integer okay ≥ 1 |v(xSk , xSk +1 ) − v(x, ¯ x)| ¯ ≤ 2 · 2−k . (2. 27) by means of (2. 25), (2. 26), (2. 24), (2. 21), and the alternative of δj , j = 1, 2, . . . (see (2. 15)–(2. 18)) for any integer ok ≥ 2 ⎛ ⎞ Sk −1 okay ⎝ v(xt , xt+1 ) − Sk v(x, ¯ x) ¯ = j =1 t=0 k−1 (j ) + (j +1) [v xTj , x0 Tj −1 [v xt , xt+1 − v(x, ¯ x)] ¯ ⎠ (j ) (j ) t=0 okay j =1 k−1 (2 · 2−j + δj ) − 2 − v(x, ¯ x)] ¯ ≥− j =1 2−j .

Three. three Proofs of Theorems three. 2 and three. three . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . three. four evidence of Theorem three. four . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . forty seven forty seven 50 fifty seven sixty one four optimum keep an eye on issues of Nonsingleton Turnpikes . . . . . . . . . . . . four. 1 Discrete-Time optimum regulate platforms . . . . . . . . . . . . . . . . . . . . . . . four. 2 The Turnpike estate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . four. three Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . four. four Auxiliary effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . four. five evidence of Theorem four. five . .

T − 1 gratifying ut − v ≤ γL , t = zero, . . . , T − 1 −1 )-program {yt }Tt=0 pleasant and every ({Ωt }Tt=0 ¯ ρ(yT , x) ¯ ≤γ ρ(y0 , x), and T −1 −1 −1 ut (yt , yt+1 ) ≥ σ ({ut }Tt=0 , {Ωt }Tt=0 , zero, T , y0 , yT ) − γ t=0 ¯ ≤ holds for all t = zero, . . . , T . the inequality ρ(yt , x) with regards to Theorem 2. three set l2 = 1. by means of Lemmas 2. 17 and a pair of. 18 (with = γ and M0 = 1), there exist γ˜ ∈ (0, γ ) and a typical quantity L > l1 + l2 such that the subsequent homes carry: 2. five Proofs of Theorems 2. 2 and a pair of.

This regulate process is defined by way of a nonempty closed set Ω ⊂ X×X which determines a category of admissible trajectories (programs) and by way of a bounded higher semicontinuous aim functionality v : X × X → R 1 which determines an optimality criterion. We convey the soundness of the turnpike phenomenon less than small perturbations of the target functionality v and the set Ω. 2. 1 Preliminaries and balance effects allow (X, ρ) be a compact metric house. for every x ∈ X and every nonempty set C ⊂ X set ρ(x, C) = inf{ρ(x, y) : y ∈ C}.

Download PDF sample

Rated 4.50 of 5 – based on 44 votes