Mathematical models have been very successful in the study of the
physical world. Galilei and Newton introduced point particles moving
without friction under the action of simple forces as the basis for the
description of concrete motions like the ones of the planets. This
approach was sustained by appro- priate mathematical methods, namely
infinitesimal calculus, which was being developed at that time. In this
way classical analytical mechanics was able to establish some general
results, gaining insight through explicit solution of some simple cases
and developing various methods of approximation for handling more
complicated ones. Special relativity theory can be seen as an extension
of this kind of modelling. In the study of electromagnetic phenomena and
in general relativity another mathematical model is used, in which the
concept of classical field plays the fundamental role. The equations of
motion here are partial differential equations, and the methods of study
used involve further developments of classical analysis. These models
are deterministic in nature. However it was realized already in the
second half of last century, through the work of Maxwell, Boltzmann,
Gibbs and others, that in the discussion of systems involving a great
number of particles, the deterministic description is not by itself of
great help, in particu- lar a suitable "weighting" of all possible
initial conditions should be considered.