Lyapunov function: Difference between revisions

Revision as of 19:39, 14 December 2020

Latest revision as of 15:03, 25 May 2025

(34 intermediate revisions by 24 users not shown)

Line 1:

{{shortShort description|Concept in the analysis of dynamical systems}}{{No footnotes|date=October 2023}}

In the theory of [[ordinary differential equations]] (ODEs), '''Lyapunov functions''' are scalar functions that may be used to prove the stability of an [[equilibrium point|equilibrium]] of an ODE. Named after the [[Russia]]n [[mathematician]] [[Aleksandr Lyapunov|Aleksandr Mikhailovich Lyapunov]], Lyapunov functions (also called the Lyapunov’s second method for stability) are important to [[stability theory]] of [[dynamical system]]s and [[control theory]]. A similar concept appears in the theory of general state space [[Markov chain]]s, usually under the name Foster–Lyapunov functions.

In the theory of [[ordinary differential equations]] (ODEs), '''Lyapunov functions''', named after [[Aleksandr Lyapunov]], are scalar functions that may be used to prove the stability of an [[equilibrium point|equilibrium]] of an ODE. Named after the [[Russia]]n [[mathematician]] [[Aleksandr Lyapunov|Aleksandr Mikhailovich Lyapunov]], Lyapunov functions (also called the Lyapunov’s second method for stability) are important to [[stability theory]] of [[dynamical system]]s and [[control theory]]. A similar concept appears in the theory of general state -space [[Markov chain]]s, usually under the name Foster–Lyapunov functions.

For certain classes of ODEs, the existence of Lyapunov functions is a necessary and sufficient condition for stability. Whereas there is no general technique for constructing Lyapunov functions for ODEs, in many specific cases the construction of Lyapunov functions is known. For instance, [[quadratic function|quadratic]] functions suffice for systems with one state; the solution of a particular [[linear matrix inequality]] provides Lyapunov functions for linear systems; and [[Conservation law (physics)|conservation law]]s can often be used to construct Lyapunov functions for [[physical system]]s.

For certain classes of ODEs, the existence of Lyapunov functions is a necessary and sufficient condition for stability. Whereas there is no general technique for constructing Lyapunov functions for ODEs, in many specific cases the construction of Lyapunov functions is known. For instance, [[quadratic function|quadratic]] functions suffice for systems with one state;, the solution of a particular [[linear matrix inequality]] provides Lyapunov functions for linear systems;, and [[Conservation law (physics)|conservation law]]s can often be used to construct Lyapunov functions for [[physical system]]s.

== Definition ==

A Lyapunov function for an autonomous [[dynamical system]]

:<math>\begin{cases} g : \R ^n \to \R ^n &
\\ \dot{y} = g(y) \end{cases}</math>

with an equilibrium point at <math>y=0</math> is a [[scalar function]] <math>V:\R^n\to\R</math> that is continuous, has continuous first derivatives, is strictly positive for <math>y\neq 0</math>, and for which the time derivative <math>-\dot{V} = \nabla{V}\cdot g</math> is also strictlynon positive (these conditions are required on some region containing the origin). The (stronger) condition that <math>-\nabla{V}\cdot g</math> is strictly positive for <math>y\neq 0</math> is sometimes stated as <math>-\nabla{V}\cdot g</math> is "''locally positive definite"'', or <math>\nabla{V}\cdot g</math> is "''locally negative definite"''.

===Further discussion of the terms arising in the definition===

Lyapunov functions arise in the study of equilibrium points of dynamical systems. In <math>\R^n,</math> an arbitrary autonomous [[dynamical system]] can be written as

:<math>\dot{y} = g(y)</math>

Line 18 ⟶ 20:

for some smooth <math>g:\R^n \to \R^n.</math>

An equilibrium point is a point <math>y^*</math> such that <math>g\left(y^*\right) = 0.</math> Given an equilibrium point, <math>y^*,</math> there always exists a coordinate transformation <math>x = y - y^*,</math> such that:

:<math>\begin{cases} \dot{x} = \dot{y} = g(y) = g\left(x + y^*\right) = f(x) \\ f(0) = 0 \end{cases}</math>

Thus, in studying equilibrium points, it is sufficient to assume the equilibrium point occurs at <math>0</math>.

By the chain rule, for any function, <math>H:\R^n \to \R,</math> the time derivative of the function evaluated along a solution of the dynamical system is

:<math> \dot{H} = \frac{d}{dt} H(x(t)) = \frac{\partial H}{\partial x}\cdot \frac{dx}{dt} = \nabla H \cdot \dot{x} = \nabla H\cdot f(x).</math>

Line 35 ⟶ 37:

{{main article|Lyapunov stability}}

Let <math>x^* = 0</math> be an equilibrium point of the autonomous system

:<math>\dot{x} = f(x).</math>

and use the notation <math>\dot{V}(x)</math> to denote the time derivative of the Lyapunov-candidate-function <math>V</math>:

:<math>\dot{V}(x) = \frac{d}{dt} V(x(t)) = \frac{\partial V}{\partial x}\cdot \frac{dx}{dt} = \nabla V \cdot \dot{x} = \nabla V\cdot f(x).</math>

===Locally asymptotically stable equilibrium===

If the equilibrium point is isolated, the Lyapunov-candidate-function <math>V</math> is locally positive definite, and the time derivative of the Lyapunov-candidate-function is locally negative definite:

:<math>\dot{V}(x) < 0 \quad \forall x \in \mathcal{B}(0)\setminus\{0\},</math>

for some neighborhood <math>\mathcal{B}(0)</math> of origin, then the equilibrium is proven to be locally asymptotically stable.

===Stable equilibrium===

If <math>V</math> is a Lyapunov function, then the equilibrium is [[Stability theory|Lyapunov stable]].

The converse is also true, and was proved by [[José Luis Massera|J. L. Massera]].

===Globally asymptotically stable equilibrium===

If the Lyapunov-candidate-function <math>V</math> is globally positive definite, [[Radially unbounded function|radially unbounded]], the equilibrium isolated and the time derivative of the Lyapunov-candidate-function is globally negative definite:

:<math>\dot{V}(x) < 0 \quad \forall x \in \R ^n\setminus\{0\},</math>

then the equilibrium is proven to be [[Stability theory|globally asymptotically stable]].

The Lyapunov-candidate function <math>V(x)</math> is radially unbounded if

:<math>\| x \| \to \infty \Rightarrow V(x) \to \infty. </math>

(This is also referred to as norm-coercivity.)

The converse is also true,<ref name=Massera1949>{{Citation

| author = Massera, José Luis

| year = 1949

| title = On Liapounoff's conditions of stability

| journal = Annals of Mathematics |series=Second Series

| volume = 50

| issue = 3

| pages = 705–721

| doi = 10.2307/1969558

| mr = 0035354

| jstor = 1969558

The converse is also true,}}</ref> and was proved by [[José Luis Massera|J.]] L.(see also [[Massera's lemma]]).

==Example==

Consider the following differential equation with solution <math>x</math> on <math>\R </math>:

:<math>\dot x = -x.</math>

Considering that <math>x^2</math> is always positive around the origin it is a natural candidate to be a Lyapunov function to help us study <math>x</math>. So let <math>V(x)=x^2</math> on <math>\R </math>. Then,

So let <math>V(x)=x^2</math> on <math>\R </math>. Then,

:<math>\dot V(x) = V'(x) \dot x = 2x\cdot (-x) = -2x^2< 0.</math>

Line 86 ⟶ 97:

* {{cite book |first=Joseph |last=La Salle |first2=Solomon |last2=Lefschetz |title=Stability by Liapunov's Direct Method: With Applications |___location=New York |publisher=Academic Press |year=1961 }}

*{{PlanetMath attribution|id=4386|title=Lyapunov function}}

==External links==

* [https://web.archive.org/web/20110926230621/http://www.exampleproblems.com/wiki/index.php/ODELF1 Example] of determining the stability of the equilibrium solution of a system of ODEs with a Lyapunov function

{{Authority control}}

[[Category:Stability theory]]