(here approximately equal to 27), as is explained in Section 2.5.5 is shown as vertical 
lines for reference. 
2.50 5 
2.00 4 oe = 1 
Fig. 5.5a. Theoretical AR(0). 
2.50 
6. = 1.046490 
0.50 
0.00 
Fig. 5.5b. Estimated by model fitting as optimum AR(0). 
Fig. 5.5. Spectrum for pure random process AR(0) 
X= €1, Ee: N[O, 1]. 
This difference tells us that the process might not actually be white in this short 
period of 600 and also that it is hard to get the ‘real’ character of the spectrum by the 
nonparametric method from this short record. It is interesting to note, however, that by 
the model fitting method, even from this short record, the fluctuations disappeared and 
the spectrum appeared to be white. 
102 
