As I was learning about the following theorem and its proof from the book Nonlinear Systems by H. K. Khalil, I encountered a difficulty in grasping some parts of the proof.

Theorem: Consider the scalar autonomous differential equation

\begin{equation} \dot{y}=-\alpha(y),\ y(t_0)=y_0,\tag{1} \end{equation}

where $\alpha$ is a locally Lipschitz class $\kappa$ function defined on $[0,a)$. For all $0\leq{y_0}<a$, this equation has a unique solution $y(t)$ defined for all $t\geq{t_0}$. Moreover,

\begin{equation} y(t)=\sigma(y_0,t-t_0),\tag{2} \end{equation}

where $\sigma$ is a class $\kappa\ell$ function defined on $[0,a)\times[0,\infty)$.

The proof goes as follows.

Since $\alpha(.)$ is locally Lipschitz, the equation (1) has a unique solution $\forall\ {y_0}\geq{0}$. Because $\dot{y}(t)<0$ whenever $y(t)>0$, the solution has the property that $y(t)\leq{y_0}$ forall $t\geq{t_0}$. By integration we have,

\begin{equation} -\int_{y_0}^{y} \dfrac{dx}{\alpha(x)}= \int_{t_0}^{t} d\tau. \end{equation}

Let b be any positive number less than $a$ and define $\eta(y)=-\int_{b}^{y}\dfrac{dx}{\alpha(x)}$. The function $\eta(y)$ is strictly decreasing differentiable function on $(0,a)$. Moreover, $\lim_{y\to{0}}\eta(y)=\infty$. This limit follows from two facts.

First, the solution of the differential equation $y(t)\to{0}$ as $t\to\infty$, since $\dot{y}(t)<0$ whenever $y(t)>0$.

Second, the limit $y(t)\to{0}$ can happen only asymptotically as $t\to\infty$; it cannot happen in finite time due to the uniqueness of the solution.

Here I do not quite understand the second fact (in italics) how the uniqueness of solution ensures that $y(t)$ goes to $0$ asymptotically as $t\to\infty$.

Any hints on this are greatly appreciated.


That's not what it's saying. It's saying $y(t) \to 0$ can't happen in finite time, i.e. there can't be a solution $Y(t)$ of the differential equation with $Y(t_0) = y_0$ and $Y(t_1) = 0$ for some $t_1 > t_0$.

Suppose that did happen. Note that $y(t) = 0$ is also a solution of the differential equation, because part of the definition of class $\kappa$ is $\alpha(0)=0$. So this would contradict the Existence and Uniqueness Theorem, as there would be two different solutions $Y$ and $0$ having the same value at $t_1$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.