View Single Post
  #5  
Old June 11th, 2009, 12:27 PM
bkarpuz's Avatar
bkarpuz bkarpuz is offline
Senior Member
 
Join Date: Sep 2008
Location: R³
Posts: 320
Country:
Thanks: 80
Thanked 65 Times in 59 Posts
bkarpuz will become famous soon enoughbkarpuz will become famous soon enough
Send a message via ICQ to bkarpuz Send a message via AIM to bkarpuz Send a message via MSN to bkarpuz Send a message via Yahoo to bkarpuz Send a message via Skype™ to bkarpuz
Thumbs up Proved.

Quote:
Originally Posted by bkarpuz View Post
Dear friends, I need some help with the existence and/or uniqueness of global solutions to first-order linear differential equations.
For instance, let x_{0},t_{0}\in\mathbb{R} and A,B\in C([t_{0},\infty),\mathbb{R}) and consider the following differential equation
\begin{cases}x^{\prime}(t)=A(t)x(t)+B(t),& t\geq t_{0}\\x(t_{0})=x_{0}.&\end{cases}
Which theorem ensures existence of global solutions to this initial value problem?
...
Consider the following initial value problem
\begin{cases}x^{\prime}(t)=f(t,x(\tau(t))),&t\in[b,c]\\x(t)=\varphi(t),&t\in[a,b],\end{cases}\quad(1)
where f:[b,c]\times\mathbb{R}\to\mathbb{R}, \varphi\in C([a,b],\mathbb{R}) and \tau\in C([b,c],[a,c]) satisfies \tau(t)\leq t for all t\in[b,c].
Set I(\varphi,\varepsilon):=\{x\in\mathbb{R}:x\in B(\varphi(t),\varepsilon)\ \text{for}\ t\in[a,b]\}, where B(x_{0},\varepsilon):=\{x\in\mathbb{R}:|x-x_{0}|\leq\varepsilon\}.

Theorem 1. Let f:[b,c]\times I(\varphi,\varepsilon)\to\mathbb{R} for some \varepsilon>0. Assume that there exist M>0 and L>0 such that |f(t,x)|\leq M for all (t,x)\in[b,c]\times I(\varphi,\varepsilon) and |f(t,u)-f(t,v)|\leq L|u-v| for all (t,u),(t,v)\in[b,c]\times I(\varphi,\varepsilon). Then (1) has a unique solution on [a,b+\delta], where \delta:=\min\{c-b,\varepsilon/M\}.

Proof. The proof can be given by following exactly the same arguments in the proof of Picard-Lidelof theorem, and thus omitted here. \rule{0.2cm}{0.2cm}

Now, consider the following initial value problem
\begin{cases}x^{\prime}(t)=p(t)x(\tau(t))+q(t),&t\in[b,c]\\x(t)=\varphi(t),&t\in[a,b],\end{cases}\quad(2)
where p,q\in C([b,c],\mathbb{R}), and the other arguments are same to that of (1).

Corollary 1. (2) admits a unique solution on [a,c].

Proof. Let f(t,u):=p(t)u+q(t) for (t,u)\in[b,c]\times\mathbb{R}, and M_{1},M_{2}>0 satisfy |p(t)|\leq M_{1} and [math]|q(t)|\leq M_{2}[math] for all t\in[b,c] (since p,q are continuous, we may always find such constants). The Lipschitz condition holds on [\xi_{0},\xi_{1}]\times\mathbb{R} with the Lipschitz constant M_{1}>0. Let b=\xi_{0}<\xi_{1}<\cdots<\xi_{k_{0}}=c satisfy \xi_{k}-\xi_{k-1}\leq1/(2M_{1}) for k=1,2,\ldots,k_{0}. For convenience in the notation define x_{0}:=\varphi and N_{0}:=\max\nolimits_{t\in[a,\xi_{0}]}\{|x_{0}(t)|\}. We may pick \varepsilon_{0}>0 such that \varepsilon_{0}/(M_{1}(\varepsilon_{0}+N_{0})+M_{2})\geq1/(2M_{1}), we see that |f(t,u)|\leq M_{1}(\varepsilon_{0}+N_{0})+M_{2} for all (t,u)\in[\xi_{0},\xi_{1}]\times B(0,\varepsilon_{0}+N_{0}). Applying Theorem 1, we see that
\begin{cases}x^{\prime}(t)=p(t)x(\tau(t))+q(t),&t\in[\xi_{0},\xi_{1}]\\x(t)=x_{0}(t),&t\in[a,\xi_{0}]\end{cases}
admits a unique solution x_{1} on [a,\xi_{1}] since \min\{\xi_{1}-\xi_{0},\varepsilon_{0}/(M_{1}(\varepsilon_{0}+N_{0})+M_{2})\}\geq\min\{\xi_{1}-\xi_{0},1/(2M_{1})\}. Next, let N_{1}:=\max\nolimits_{t\in[a,\xi_{1}]}\{|x_{1}(t)|\}. We may find \varepsilon_{1}>0 such that \varepsilon_{1}/(M_{1}(\varepsilon_{1}+N_{1})+M_{2})\geq1/(2M_{1}). And we have |f(t,u)|\leq M_{1}(\varepsilon_{1}+N_{1})+M_{2} for all (t,u)\in[\xi_{1},\xi_{2}]\times B(0,\varepsilon_{1}+N_{1}). Applying Theorem 1, we see that
\begin{cases}x^{\prime}(t)=p(t)x(\tau(t))+q(t),&t\in[\xi_{1},\xi_{2}]\\x(t)=x_{1}(t),&t\in[a,\xi_{1}]\end{cases}
admits a unique solution x_{2} on [a,\xi_{2}]. Repeating in this manner, we obtain the unique solution x_{k_{0}} of (2) on [a,c]. \rule{0.2cm}{0.2cm}

Remark 1
. If we need to obtain the unique global solution to
\begin{cases}x^{\prime}(t)=p(t)x(\tau(t))+q(t),&t\in[b,\infty)\\x(t)=\varphi(t),&t\in[a,b],\end{cases}\quad(3)
where in addition \lim\nolimits_{t\to\infty}\tau(t)=\infty is assumed to hold, we may pick an increasing divergent sequence \{\xi_{k}\}_{k\in\mathbb{N}}\subset[b,\infty) with the convenience \xi_{-1}:=a and \xi_{0}:=b such that \xi_{k-1}\leq\min\nolimits_{t\in[\xi_{k},\infty)}\{\tau(t)\} for all k\in\mathbb{N}, and apply Corollary 1 successively to obtain the unique solution x_{k} on each of the intervals [\xi_{k-1},\xi_{k}] for k\in\mathbb{N} by assuming the solution x_{k-1} obtained in the previous step as the initial function on the current interval with the convenience x_{0}:=\varphi. Then, letting x(t)=x_{k}(t) for t\in[\xi_{k-1},\xi_{k}] for k\in\mathbb{N}, we obtain the unique global solution to (3).

Appendix. It is clear that the function t/(at+b)\to1/a\geq1/(2a) as t\to\infty for any fixed a,b>0.

Remark. The proof is very simple in the case \inf\nolimits_{t\in[b,\infty)}\{t-\tau(t)\}>0.

proof by bkarpuz

Last edited by bkarpuz; September 19th, 2009 at 05:25 AM.
Reply With Quote