Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 1801.03234

Optimal linear responses for Markov chains and stochastically perturbed dynamical systems

Fadi Antown·Davor Dragičević·Gary Froyland
View original source ↗Read source document ↗Export metadata ↓

Abstract

The linear response of a dynamical system refers to changes to properties of the system when small external perturbations are applied. We consider the little-studied question of selecting an optimal perturbation so as to (i) maximise the linear response of the equilibrium distribution of the system, (ii) maximise the linear response of the expectation of a specified observable, and (iii) maximise the linear response of the rate of convergence of the system to the equilibrium distribution. We also consider the inhomogeneous or time-dependent situation where the governing dynamics is not stationary and one wishes to select a sequence of small perturbations so as to maximise the overall linear response at some terminal time. We develop the theory for finite-state Markov chains, provide explicit solutions for some illustrative examples, and numerically apply our theory to stochastically perturbed dynamical systems, where the Markov chain is replaced by a matrix representation of an approximate annealed transfer operator for the random dynamical system.

Explore related subjects

math.DS

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Fadi Antown, Davor Dragičević, Gary Froyland. 2018-01-10. Optimal linear responses for Markov chains and stochastically perturbed dynamical systems. https://doi.org/10.1007/s10955-018-1985-1

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Monotonicity, global symplectification and the stability of Dry Ten Martini Problem

We prove that, for every irrational frequency and every analytic Type I potential, each supercritical spectral energy satisfying the gap-labelling condition is an endpoint of an open spectral gap. This establishes the conjecture of Ge--Jitomirskaya--You \cite{GJY,You} in the supercritical regime. Consequently, the ``all gaps open'' property of the supercritical almost Mathieu operator persists under sufficiently small analytic perturbations. The main ingredient is a global symplectification of the center bundle that preserves quantitative monotonicity. This allows us to study gap opening through the center dynamics of the dual long-range operator, which has no natural Schrödinger form. We first establish the result for trigonometric polynomial potentials and then pass to general analytic potentials by controlling the dependence on the truncation dimension. The proof combines a discrete Hellmann--Feynman identity, dimension-free Aubry duality in weighted analytic norms, and a quantitative cone argument based on pre-monotonicity. These estimates ensure that the gaps survive in the analytic limit. Our results establish analytic stability of the Dry Ten Martini Problem in the supercritical regime and give a partial answer to a question of M. Shamis on the persistence of periodic spectral gaps.

math.DS↗
arXiv2026

Asymmetry of a class of Mellin transforms via bounded solutions

We introduce a family of parametrized non-homogeneous linear complex differential equations on $[1,\infty)$, depending on a complex parameter $s$ in the critical strip. We identify sufficient conditions on the non-homogeneous term that induce a structural asymmetry between the solutions corresponding to the parameters $s$ and $1-s$. More precisely, if both solutions with initial value $1$ are bounded on $[1,\infty)$, then necessarily $\Re(s)=\tfrac12$. The initial condition associated with the unique bounded solution corresponding to a parameter $s$ represents a zero of the Mellin transform associated with the non-homogeneous term at the point $s$.

math.DS↗
arXiv2026

Self-similar Delone sets and Pisot numbers

We consider Delone point patterns with self-similarity. Under mild conditions, the similarity factor is a Pisot number if and only if the pattern is uniformly discrete. The classical case is a Meyer set $Λ$ with $Λ\supset θΛ$ for some $θ>1,$ for which $θ$ must be a Pisot number or a Salem number. When $Λ$ contains several similar copies of itself, the case of a Salem number drops out for $θ<2.$ On the other hand, strictly self-similar patterns with a Pisot factor must be Meyer sets. Various examples are given.

math.DS↗