I am looking for a multidimensional version of Kingman's subadditive theorem. I found this but it is not exactely what I need.

I would rather have something like that:

Let us consider $\mathbb{Z}^2_+$ and a family $(X)$ of random variables indexed by pairs of points in $\mathbb{Z}^2$ i.e. $X_{z_1, z_2}$ is a random variable associated with subgrid of $\mathbb{Z}^2$ "starting from" $z_1$ and ending on $z_2$. Assume that for any rectangular subgrid $\Lambda \subset \mathbb{Z}^2_+$ and point $x\in \Lambda $ we have

$X_{\Lambda}\leq X_{\Lambda_1} + X_{\Lambda_2}+X_{\Lambda_3}+X_{\Lambda_4},$

where $\Lambda_1,\Lambda_2,\Lambda_3,\Lambda_4$ is $\Lambda$ split in point $x$ into four subgrids. I suppose that this together with some "usual" ergodic theorem conditions should imply that

$X_{(0,0),(n,n)}/n^2$ converges in $L^1$ and a.s.

May be some one you could give me some references.

Being in the topic of subadditivity. I found a multidimensional version of Fekete lemma. It is surprising for me that it was not done before 2007. But may be it was. Again I will be happy to know any.


Have a look at the paper by Nguyen, Xuan-Xanh Ergodic theorems for subadditive spatial processes, Z. Wahrsch. Verw. Gebiete 48 (1979), no. 2, 159–176, MR0534842 (82c:60056). The setup there is somewhat different, but it seems that the main result there should contain your claim.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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