1: \begin{abstract}

2: Starting with an infinite measure Lebesgue space $\left(X,\mathcal{A},\mu\right)$,

3: we consider two norms on $L^{1}\left(\mu\right)$ that are both weaker

4: than $\left\Vert \cdot\right\Vert _{1}$. The first one is associated

5: to a particular Orlicz space and allows to obtain general ergodic

6: theorems for a measure preserving transformation, involving the convergence

7: of its Birkhoff sums, that are true for $\left(L^{p}\left(\mu\right),\left\Vert \cdot\right\Vert _{p}\right)$,

8: $1<p<\infty$, but fail for $p=1$ as the measure is infinite. With

9: the second, we obtain non-trivial $T$-invariant vectors in the completion

10: of $L^{1}\left(\mu\right)$ with respect to this new norm and this

11: leads to a $L^{1}\left(\mu\right)$-characterization of ergodicity.

12: Both norms are in fact related to the Poisson process over $\left(X,\mathcal{A},\mu\right)$

13: and their family of stochastic integrals whose inherent integrability

14: constraints are the main ingredient in the proofs. A new characterization

15: of stochastic integrals is obtained and additional results on coboundaries

16: come out as a natural byproduct.

17: \end{abstract}

18: