3
$\begingroup$

I am right now working on some linear parabolic problems studying the behaviour of its solutions for large initial data. To do this, I have needed to use some estimates of the Dirichlet and Neumann heat kernels (for the Laplacian operator) in exterior domains (i.e. $\Omega=\mathbb{R}^N\backslash K$ where $K$ is a regular compact set).

I have found multiple results from Saloff-Coste, Grigor'yan, Davies, Q. S. Zhang... (See the references below). Most of them concern the problem in complete Riemann manifolds. Some of them study the problem in exterior domains (or in more general (if $K$ is regular enough) inner uniform domains) [2,4,5]. A good general text about this is this one from Saloff-Coste: [1].

However, most of the results I have found are bounds like the following: $$ \phi(x,y,t)\frac{ e^{-c_1\frac{d(x,y)^2}{4t}}}{(4\pi t)^{N/2}}\leq k(x,y,t)\leq \Phi(x,y,t)\frac{ e^{-c_2\frac{d(x,y)^2}{4t}}}{(4\pi t)^{N/2}} $$ where $d(x,y)$ is the geodesic distance in $\Omega$ between $x$ and $y$ and $\phi,\Phi$ are known functions. See for example [2] Th 1.1. and [4] Th 1.3.1, 1.3.3.

I would like to know if there are some sharp results in which $c_1=1$ and/or $c_2=1$, like the heat kernel in the whole space. I have not found almost any result in this way. One I found is [6] Cor 3.1, Th 4.1 in which they get $c_1=1+\varepsilon$ and $c_2=1-\varepsilon$ with $\varepsilon$ as small as one would like (but then $\phi$ and $\Phi$ change obviously) but it is for complete manifolds.

Any comment, suggestion or reference related to this would be really helpful. In particular my main questions are:

  1. Are there any sharp result of that type? It is useful for me even if it is not the case of exterior domain, but other settings.
  2. Are there any result in which I can make $c_1$ and $c_2$ as close to $1$ as we want, although not $1$?
  3. If you don't know about the first two questions, do you think it is a plausible result to obtain that type of bounds for the Dirichlet Heat Kernel in exterior domains?

I am also interested in Neumann heat kernel, so any results for it are also welcomed.


References:

[1] The heat kernel and its estimates. L. Saloff Coste.

[2] The global behavior of heat kernels in exterior domains. Q. S. Zhang.

[3] Heat Kernels and Spectral Theory. E. B. Davies.

[4] Heat kernel estimates for inner uniform subsets of Harnack-type Dirichlet space. P. Gyrya.

[5] Dirichlet heat kernel in the exterior of a compact set Grigor'yan, Saloff-Coste.

[6] On the parabolic kernel of the Schrödinger operator P. Li, S. T. Yau

PS: This is my first question here, do not hesitate to correct me if the question is not well formulated.

$\endgroup$

2 Answers 2

2
$\begingroup$

The upper bound with $c_2=1$ and a polynomial $\Phi$ should be in "Coulhon-Sikora: Gaussian upper bounds via Phragmen-Lindelof principle", Proceedings London Math. Soc. 96 (2008), 507-544. The setting is quite general and should cover exterior domains, too.

$\endgroup$
1
  • $\begingroup$ Thank you very much! It seems a very good text for the upper bound by itself and its references . I have to check all the conditions because it is written in a very general setting, but it seems that it works. I even think that the polynomial dependence can be removed with Theorem 4.2 but I'm not quite sure because I have not treated so much with complex time. $\endgroup$
    – joaquindt
    Commented Apr 8, 2022 at 10:34
1
$\begingroup$

Unfortunately I don't know about any estimates for exterior domains, but there are recent results for the ball and a certain class of convex domains (including unbounded):

Dirichlet Heat Kernel for the Laplacian in a Ball by J. Małecki, G. Serafin

Laplace Dirichlet heat kernels in convex domains by G. Serafin

It seems that convexity is vital for their methods (see p. 703 in the second paper), which does not sound very optimistic for exterior domains.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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