Using Monte Carlo simulations, we study the relaxation and short-time diffu
sion of polymer chains in two-dimensional periodic arrays of obstacles with
random point defects. The displacement of the center of mass follows the a
nomalous scaling law r(c.m.)(t)(2)=4D*t(beta), with beta<1, for times t<t(S
S), where t(SS) is the time required to attain the steady state. The relaxa
tion of the autocorrelation function of the chain's end-to-end vector, on t
he other hand, is well described by the stretched exponential form C(t)= ex
p[-(t/tau*)(alpha)], where 0 < alpha less than or equal to 1 and tau*much l
ess than t(SS). However, our results also obey the functional form C(r(c.m.
))= exp(-[r(c.m.)/lambda](2)), implying the coupling alpha=beta even though
these exponents vary widely from system to system. We thus propose that it
is lambda, and not the traditional length (D tau*)(1/2), that is the relev
ant relaxation polymer length scale in disordered systems. [S1063-651X(99)1
0909-7].