I consider classical $\phi^4$-theory on the lattice. The model is defined in finite volume with Hamiltonian

\begin{align*} H(\phi) = - \sum_{x \sim y} J_{x,y} \phi_x \phi_y \end{align*} and a-priori measure \begin{align*} \rho(d \phi_x) = e^{- \lambda \phi_x^4 + b \phi_x^2} d\phi_x. \end{align*} Here the fields can be in $\mathbb{R}$, $\lambda$ is positive and I am fine with considering only ferromagnetic interactions of the form $J_{x,y} = J > 0$.

This model fits into the framework of Lebowitz and Presutti,Statistical mechanics of unbounded spins CMP. 50, 195—218 (1976). Which means that I can define the class of tempered (infinite volume) Gibbs measures for this particular model. That means that I get $\mu_{\pm}$ defined as limits of states on $\Lambda_R$ with boundary conditions $ \pm a \log(R)$ for an $a >0$ with all the tempered Gibbs measures lying in between $\mu_-$ and $\mu_+$.

However, as the paper is in larger generality and the $\phi^4$-model itself is very well studied. So I wonder what additional results can be shown for this particular model. In particular I wonder

- Whether all translation invariant Gibbs measures are tempered?
- What happens if I choose boundary conditions $\pm a R$ instead of $ \pm a \log(R)$
- In what sense is the class of tempered Gibbs measures the class that is natural to study and what classifications theorems (forming a Choquet simplex etc.) are know about tempered translation invariant Gibbs states and just translation invariant Gibbs states?

However, the litterature of the model is also a wilderness so I hope you can help me out.