n this work by C++ programs different numeric investigations were
accomplished to discrete models of the Boltzmann equation. In the chapter
II. were shown at the smallest collision model (6-Punkte of models) the
characteristics and afterwards for larger models generalized in chapter
III. ' numeric investigations ' first the dependence of the solution on the
lattice was more near examined. We came to the conclusion, the more near
the bend point (peak) of the function $ M_G(v_x^{(i)}, v_y^{(i)}) $ at a
grid point lie, the better the energy $ E_N $ are approximated. Increases
one the points number of models (from the 24-Punkte to the 96-Punkte model)
the energy $E_N$ is likewise better approximated. Problems arose only if
the temperature became $ T$ very small or very high. With the test of the
initial conditions it was stated that the choice of the initial conditions
have a large influence on the computations. In a further point of the work
the fading times were compared by the dog by the m-functional. We stated
that the h-functional fades away faster than the m-functional.With the
m-functional the funktionsverlauf depends $ and $ very strongly on the
choice of the variable $ \alpha \beta $. The choice of the initial speeds
does not play a large role with the two functional ones. In the next point
the work was occupied with the construction of discrete solutions to given
moments. As basis for this the Newton procedure served. Here it showed up
that the numerically calculated borders in some cases deviate from the
theoretically determined borders. In the section III.5 we examined the
qualitative behavior of different models. With the help of the Newton
procedure after the choice by initial conditions a limit function was
computed. Afterwards with the initial conditions the main program was
started. In each step that became 4. Moment computes and with that 4.
Moment of the limit function compared. As result we received and stated the
function $\phi(t)$ that them have a linear process within the ranges in
accordance with table 1 (chapter III.1). The optimal process showed up in
the 96-Punkte model with the initial condition $\theta=1$ and the Vorfaktor
$\alpha_1=1$ before the impact operator. These values were then transferred
in the section III.3. Here it was shown that the solution the
characteristics BKW type fulfill the chapter IV. place the programs, which
were used during the numeric investigations forwards in the appendix are
closer the illustrations, which are only with difficulty recognizable in
the text again represented. Further the tables and the source codes of the
individual programs are angegeben.