I'm currently looking through the following paper which examines some dynamics of the Airy$_2$ process: https://arxiv.org/pdf/1106.2717.pdf

On page 2, there appears a PDE of the form

$\partial_t u + Hu = 0, \quad x \leq g(t) $

with boundary conditions

$u(t,x) = 0, \quad x > g(t)$,

$u(l, x) = f(x)$ if $x < g(t)$

where $H = - \partial_x^2 + x$ is the Airy Hamiltonian.

Now, this means that $g$ describes the domain in space - however I don't understand why one might require it to be in $H^1$. The authors point to the proof of Proposition 3.2, but there only a generalisation from a particular solution to the $g\in H^1$ case is done - nothing indicates that that integrability should be necessary for the result to hold.

$H^1$ in one dimension doesn't seem sufficient to get a lot of regularity - if I'm not mistaken, we get that $g\in L^\infty \cap C^{0,\frac{1}{2}}$, but nothing like being locally Lipschitz or other conditions you might want in connection with a PDE that would allow one to, for example, apply divergence theorem.

So, are there situations where this kind of $H^1$ regularity on a domain is important, or was it just some class for which the proof was done in this paper.

An additional doubt I have is that in context, the statement should work even if just $g(l), \ g(r)$ are prescribed, and $g(t) = \infty$ everywhere else, and obviously that lacks any integrability or differentiability whatsoever.


Your Answer

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

Browse other questions tagged or ask your own question.