Many fundamental questions for the understanding of polymer networks a
re more suitably addressed by current computer simulations than by exp
eriments. Details of the microscopic topology, such as the elastically
active cluster or loop entanglements, can be identified as well as co
ntrolled. In particular, it is possible to isolate and quantify their
effects on macroscopic observables such as the elastic modulus. The co
nstraints due to connectivity and conserved topology are more clearly
present for networks than for melts. Already for strand lengths betwee
n crosslinks which are relatively short, the effect of the conserved t
opology is important. The mode relaxation in a network is significantl
y different from that of a melt. For weakly crosslinked systems the me
lt entanglement length is the relevant scaling parameter. The elastic
modulus of a long chain network under ideal conditions reaches an asym
ptotic value which is about 2.2 times smaller than the prediction of a
n affine model for a network made of strands of the melt entanglement
length. An analysis of the stress reveals that in the linear regime th
e contribution from the excluded volume is dominant compared to that f
rom the connectivity along the strands. For larger elongations, howeve
r, the non-linear elastic response is dominated by the chemically and
topologically shortest paths through the system, where chemical crossl
inks and topological entanglements between meshes of the network play
a similarly important role.