In [7] we proposed a general numerical approach to the (linear) radiat
ive transfer equation which resulted in a high-dimensional linear syst
em of equations. Using the concept of the generalized mean intensity,
the dimension of the system can be drastically diminished, without los
ing any information. Additionally, the corresponding system matrices a
re positive definite under appropriate conditions on the choice of the
discrete ordinates and, therefore, the classical conjugate gradient-i
teration (CG) is converging. In connection with local preconditioners,
we develop robust and efficient methods of conjugate gradient type wh
ich are superior to the classical approximate LAMBDA-iteration, but wi
th about the same numerical effort. For some numerical tests, which si
mulate the astrophysically interesting case of radiation of stars in d
ust clouds, we compare the methods derived and give some examples for
their efficiency.