To analyze the high energy heavy ion reactions performed at Lawrence B
erkeley BEVALAC, Brookhaven AGS, and CERN SPS we present a dynamical m
odel that is entirely Poincare covariant. The description of strong in
teractions by Lorentz-scalar quasipotentials makes possible a relativi
stic extension of the cascade concept without losing its inherent simp
licity. No field degrees of freedom appear explicitly, neither in elas
tic nucleon scattering nor in particle creation processes. Although va
rious formalisms describe directly interacting relativistic particles
with identical solutions for the two-particle case, they all suffer fr
om different problems in many-particle systems. The basis of our appro
ach is a Hamiltonian formulation for N pointlike nucleons, moving unco
nstrained in an 8N-dimensional phase space. One additional Lorentz sca
lar is introduced for an appropriate parametrization of all trajectori
es, defining also the connection to the proper times of the individual
interacting nucleons. The creation of particles, for which Hamiltonia
n dynamics does not provide a generic mechanism, is incorporated pheno
menologically as a ''perturbative'' process. We describe heavy ion col
lisions as a sequence of two-particle reactions without any additional
fit to experimental data. Pion yield and mass spectra of the heavy fr
agments agree reasonably well with BEVALAC results. We do not compare
to experimental high energy data because the employed mechanism of par
ticle production is certainly improper in that regime. However, the nu
merical results give an impression on how the reaction evolves.