igjel. 
1325 
13 e35 
13.4. 
13253 
13.6. 
Weiae 
13.8. 
13:9: 
13.10. 
13.11. 
FIGURES (Continued) 
Page 
Duffingstyperoscillatoneaassae eae eee rare rorer 383 
A(O)2 A (@) for stable: Dufting type modeliys. season eee ee 385 
Generated exponential AR model X, = (1.5 + 0.28e7 1-1) Xp] 
= 96X25 +67 re; NIONOLOZ Sona re ee toi re ore teri ere se 386 
X, = (1.5 + 0.28e%1) X,1- 0.96X;-2 + a, 
Input a; = sin{sinz fit): th; fit) is increasing by time ................. 387 
X, = (1.5 +0.28e%41) X,1—0.96X;-2 + a, 
Input a, = sin{2z f(t) -t|; ix) is decreasing by time .................. 388 
A@)s Ae) forVandellolltypeimodellmansa: sea ane 388 
X, = (1.95 + 0.23e%A)X,_; — (0.96 + 0.24e*"1)X,_9, with different 
initial values (without white noise) ............ cc cee ee eee 389 
Path of characteristic roots of threshold model and exponential 
ARS mM OE]! 5 ic, 5e soa usidocgerscelaiae a eel seouaeeye ace taooeoe aisice ane leteyenenseatenoneuanch enon eviels 390 
Schematical expression of behaviors of characteristic roots ........... 391 
Behavior of characteristic roots of a nonlinear threshold AR model ..... 391 
Time history of a nonlinear threshold model that has characteristic 
roots thatibehave asiFig-wl3 AO Myce creer rh ster irr eerie eetoier on torsioke 392) 
Xvi 
