We consider a general model with structural changes including both non-aggregate and
aggregate models.
We keep index numbers to indM1 and indM2 together with state number vector sind
for the case with a single break. Then we set
k1=k[indM1[.,sind[t+1]]] for the first group
and
k1=k[indM2[.,sind[t+1]]] for the second group.
If we have two breaks, we will have
k1=k[indM3[.,sind[t+1]]] for the last group.
In general, we keep index numbers to indMs that is (n x dim) x dim matrix.
We pick every n elements in order. So far we do the same things as those for
non-aggregate model.
We set differently function fc in terms of paras rather than A and Q.
That is,
![]()
In addition, we change Kstar, K, K1, and Ks by xstar, x, x1, and xs.
For three groups of parameters, we set
para1=A1~Q1
para2=A2~Q2
para3=A3~Q3
and then
paras=para1|para2|para3
for the aggregate case.