Consider a differential separable equation $$y^{\prime}=f(y,t)$$ with initial solution $y(t_0)=y_0$. Suppose that $f(y_0,t_0)$ is not defined. Is there a theorem which can be used to prove the existence and the uniqueness of the solution of this Differential Equation?

The trouble is because $f(y_0,t_0)$ is not defined (much worse than discontinuous where we can still use Carathéodory's existence theorem)

For example a separable differential equation $y^{\prime}=\frac{1}{y-1}+2$ with initial solution $y(0)=1$.

  • $\begingroup$ Another example would be $y'=\frac{1}{2\sqrt y},\quad y(0)=0.$ It illustrates the existence much better=) $\endgroup$ Jul 12 '13 at 13:28
  • $\begingroup$ If $f(y_0,t_0)$ is not defined, how do you interpret the equality $y'(t)=f(y(t),t)$ in the point $t=t_0$? $\endgroup$
    – Tomás
    Jul 12 '13 at 13:29
  • $\begingroup$ @Tomás we don't need to. We pose the differential equation on the open set $\mathbb R\setminus \{t_0\}$ and initial condition in $t_0$. $\endgroup$ Jul 12 '13 at 13:30
  • $\begingroup$ @TZakrevskiy, ok I understood, thank you. $\endgroup$
    – Tomás
    Jul 12 '13 at 13:31
  • $\begingroup$ I don't think there exists a general result in that direction. Not sure at all. In your example, I believe there are two solutions. $\endgroup$ Sep 6 '13 at 23:23

I think what you are after is the concept of weak solution, that is precisely used to give a sense to differential equations without having to worry about single points in the domain. You can check intuition behind weak solution , https://en.wikipedia.org/wiki/Weak_solution The idea is to use the integral formulation of the equation, that in your case is: \begin{equation} y(t)-y(t_0)=\int^t_{t_0}f(s,y)d s \end{equation} if you are willing to adopt this as a definition of your solution then you can dispense with the C1 assumption and even the "definition" of the function on a set of measure zero.


The answer is no. At least there is no such definable solution, because, according to the definition, a solution of the IVP $$ y'=f(t,y), \quad y(t_0)=y_0, $$ is a $C^1$ function $\psi$ which is defined in an open interval $I$, with $t_0\in I$, and satisfying the both the initial condition and the ODE, i.e., $$ \psi(t_0)=y_0\,\,\,\text{and}\,\,\,\psi'(t)=f\big(t,\psi(t)\big), $$ for all $t\in I$. Satisfaction of the ODE implies that $$ \big(t,\psi(t)\big) \in D, \quad \text{for all $t\in I$}, $$ where $D$ is the domain of definition (and continuity) of $f$. Hence $(t_0,y_0)\in D$.


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.