Uniqueness of the ODE solutions

123 Views Asked by At

Say we have a continuous function (perhaps not everywhere differentiable) that satisfies an ODE $y^\prime(x)=h(y(x),x)$ for almost all $x$ in $[0,1]$.

Are the any references for that deal with basic ODE questions (existence, uniqueness) for these class of solutions? If so which ones you would recommend?

1

There are 1 best solutions below

0
On BEST ANSWER

Mere continuity is not enough for viable theory. Indeed, the Cantor function is continuous and satisfies $y'(x)=0$ for almost all $x$. So do linear combinations of its shifts, and a countless number of other similar functions $y$. So we don't have anything like a uniqueness theorem in this context.

The appropriate assumption on $y$ is absolute continuity. This is equivalent to $y$ being an indefinite integral of a Lebesgue integrable function (which agrees with $y'$, of course). We can recast the problem as an integral equation $$y(x) = y_0+ \int_{x_0}^x h(y(t),t)\,dt $$ and perhaps attempt the Picard iteration.

A classical existence result for equations with discontinuous $h$ is Carathéodory's existence theorem. Under additional assumption one can get uniqueness: see the notes On Discontinuous Differential Equations by Bressan and Shen, where further references may be found.