To determine the diffusion constant Ciccotti et al.
[J. Stat. Phys. 21(1979)1] suggest the following Hamiltonian
nonequilibrium method:
Let

(8.43) 
where for even/odd numbered particles is a
``color charge'' coupling to the applied ``color field'' .
In other words, particles are drawn in opposite directions depending
on their number; otherwise, they are completely identical.
Keeping constant we have

(8.44) 
where is the ``color current''.
Again, the external field does work on the system, heating it:

(8.45) 
The heating may once more be avoided by Gaussian terms in the eqs. of
motion which now read, in 2 dimensions,



(8.46) 



(8.47) 



(8.48) 



(8.49) 
In this way a ``color conductivity''

(8.50) 
may be evaluated. Applying Linear Response Theory one may show that
is closely related to the diffusion constant:

(8.51) 
F. J. Vesely / University of Vienna