The case $\lambda=1$ is well known free field kernel. What about $\lambda$ in between 0 and 1 ?? ... for $\lambda>1$ I have a proof that the kernel is not reflection positive , but for $\lambda<1$ there is the so called fractional Brownian motion representation of the corresponding semigroup describing the fractional diffusion generalized ( if space dimension is greater zero ) processes..
1 Answer
Good question, about something which is not very well known.
For $\lambda=0$, the two-point function is a delta function and the random field corresponding to the Euclidean functional integral is called white noise. The model is trivially reflection positive, i.e., this is a free (Gaussian) unitary QFT, albeit an ultralocal one which is not very interesting.
For $0<\lambda<1$, the model is also reflection positive, i.e., this is a unitary Euclidean QFT, often called a generalized free field.
The reason for this unitarity property is that one has a convergent integral representation $$ \frac{1}{(p^2+m^2)^{\lambda}}= \frac{1}{C_{\lambda}}\times \int\limits_{0}^{\infty} \frac{1}{p^2+m^2+u}\times \frac{du}{u^{\lambda}} $$ where $$ C_{\lambda}=\int\limits_{0}^{\infty}\frac{du}{u^{\lambda}(u+1)}\ . $$ The above formula is an explicit Källén-Lehmann representation. So, reflection positivity for the fractional case $0<\lambda<1$ follows, by continuous superposition, from the reflection positivity for the usual case $\lambda=1$. For more details, see Theorem 7.1 of the article "A PDE construction of the Euclidean $\Phi_{3}^{4}$ quantum field theory" by Gubinelli and Hofmanová.
-
$\begingroup$ Thank you very much for pointing out the paper by Gubinelli and Hofmanowa-indeed this is something like Lehman-Kallen representation of the ( Euclidean -time e )Wightman two-point function $\endgroup$ Commented Sep 21, 2020 at 7:31