3
$\begingroup$

I recently came across the Feynman-Kac formula which states that given an open domain $\Omega\in\mathbb{R}^n$ and $f \in L^2(\Omega)$ where $x \in \Omega$ and $t > 0$, then $e^{t\Delta_D}f(x) = E_x(f(\omega(t))\psi_{\Omega}(\omega, t))$, where t > 0 is arbitrary, $\omega(t)$ denotes an element of the probability space of Brownian motions starting in $x$ and $E_x$ is to be understood with regards to the measure of that probability space and $\psi_{\Omega}(\omega, t) = 1$ if $\omega([0, t]) \subset \Omega$ and $0$ otherwise. I am curious if there are any same kind of formula or generalisations of Feymann–Kac available for other operators for example for fractional Laplacians or higher order Laplacians or even for $p$-Laplacians?

Any insight will be very helpful.

$\endgroup$
5
  • $\begingroup$ TeX note: instead of 1\ if\ ω([0, t]) \subset \Omega $1\ if\ ω([0, t]) \subset \Omega$, use 1\text{ if $\omega([0, t]) \subset \Omega$} $1\text{ if …}$ (although there are differences of opinion about the scope of \text). In fact, there's no need to keep math mode going at all, so $1$ if $\omega([0, t]) \subset \Omega$ is best of all. Also, in $p-$ Laplacians, the - goes out of math mode (it's a hyphen, not a minus). I edited accordingly. You might also like to know about the {cases} environment: \psi = \begin{cases} 1 & \text{if …} \\ 0 & \text{if …} \end{cases}. $\endgroup$
    – LSpice
    Commented Apr 7 at 18:56
  • $\begingroup$ For fractional Laplacians it is $\alpha$ stable Levy processes. Also pretty general elliptic operators correspond to diffusions. $\endgroup$
    – user479223
    Commented Apr 7 at 18:56
  • $\begingroup$ I don't know about a Feynman-Kac formula specifically, but the $p$-Laplacian has a nice probabilistic interpretation which generalizes the Laplacian's interpretation as the infinitesimal generator of Brownian motion, see arxiv.org/abs/math/0607761 $\endgroup$ Commented Apr 7 at 21:25
  • $\begingroup$ "higher order Laplacians" For differential operators of order $>2$ one can't have such a formula because the semigroup generated by such an operator is not positivity preserving. $\endgroup$ Commented Apr 8 at 5:21
  • $\begingroup$ @JochenGlueck -- the high-order generalizations I linked to in my answer (in particular #4) seem to work around this obstruction by introducing a positive probability measure on a complex space of paths. $\endgroup$ Commented Apr 8 at 7:46

1 Answer 1

2
$\begingroup$

Some pointers to the (extensive) literature on generalized Feyman-Kac formulas:

$\endgroup$
1
  • $\begingroup$ is there any overview paper, where the various directions of research on this subject would be discussed jointly? (Optimally, but not necessarily, also taking into account some relationships with other lines of research that started out of the Feynman--Kac formula, such as Nelson's approach to QFT.) $\endgroup$
    – rpk
    Commented Apr 8 at 1:29

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .