This nice paper (unfortunately not crossposted to astroph) studies the welldefinedness of classical interactions which go like a powerlaw in the longrange limit, in [tex]d[/tex] dimensions. Newtonian gravity in three dimensions is a particular marginally pathological case. The fact that Newtonian gravity is not welldefined for an infinite system is of course well known in cosmology, but this puts the problem nicely in context. (It would be interesting to know what are the statistical properties of similar systems in general relativity, where there is no problem with infinite systems and the physics is quite different.)
As a minor issue, the authors also argue that systems where the absolute force diverges can be considered welldefined is the relative forces between particles remain finite, which seems to me odd, since absolute acceleration is measurable.
[1003.5680] A dynamical classification of the range of pair interactions
Authors:  Andrea Gabrielli, Michael Joyce, Bruno Marcos, Francois Sicard 
Abstract:  We formalize and discuss the relevance of a classification of pair interactions based on the convergence properties of the {\it forces} acting on particles as a function of system size. We do so by considering the behavior of the probability distribution function (PDF) P(F) of the force field F in a particle distribution in the limit that the size of the system is taken to infinity at constant particle density, i.e., in the "usual" thermodynamic limit. For a pair interaction potential V(r) with V(r \to \infty) ~ 1/r^\gamma defining a {\it bounded} pair force, we show that P(F) converges continuously to a welldefined and rapidly decreasing PDF if and only if the {\it pair force} is absolutely integrable, i.e., for \gamma > d1, where d is the spatial dimension. We refer to this case as {\it dynamically shortrange}, because the dominant contribution to the force on a typical particle in this limit arises from particles in a finite neighborhood around it. For the {\it dynamically longrange} case, i.e., \gamma </ d1, on the other hand, the dominant contribution to the force comes from the mean field due to the bulk, which becomes undefined in this limit. We discuss also how, for \gamma </ d1, P(F) may be defined in a weaker sense, using a regularization of the force summation which is a generalization of the socalled "Jeans swindle" employed to define Newtonian gravitational forces in an infinite static universe. We explain that the distinction of primary relevance in this context is, however, between pair forces with \gamma > d2 (or \gamma < d2), for which the PDF of the {\it difference in forces} is defined (or not defined) in the infinite system limit, without any regularization. 
[PDF] [PS] [BibTex] [Bookmark] 

 Posts: 119
 Joined: March 02 2005
 Affiliation: University of Helsinki