"s 
1d 
he 
= - oc 
F7 26044 D BE Ten tly (1) 
Peas ed and r, are successive parameters belonging to 
the steps k-2, k-1, k of the process and the values n, 
denote uncorrelated (and normally distributed) random 
variables of expectation zero and arbitrary, but equal 
variance 9, 
The second order GMP (1) generates random variables P. of 
expectation zero and assymptotically constant variance: 
3 
2,2 = on (2407) / (178°) (2) 
The covariance 9p Prt] is assymptotically time independent 
as well and depends only on the number of steps i, the 
generated variablesr, and i are seperated from each 
Tti 
other. 
2 1 2 2 
Sra T Tr (1*i(1-8 )/(1*8^)) (3) 
Two typical theoretical covariance functions are shown in 
figure 1 and figure 2.