Dear gretl-users,
I have the following script:
"
matrix ctest = {}
loop i = 0..150
eleje = 1 + 25*$i
vege = 250 + 25*$i
smpl eleje vege
garch 1 1 ; r_DJ r_DJ(-1)
c = $coeff
uev = c[3]/(1-c[4]-c[5])
iter = $i + 1
matrix identif = eleje ~ vege ~ uev
matrix ctest = ctest | identif
endloop
print ctest
"
I want to "push a time window" through my observations, fit a model and store
some values.
The problem is I get errors like:
loop: i = 21
? eleje = 1 + 25*21
? vege = 250 + 25*21
? smpl eleje vege
Full data range: 1970/12/31 - 2010/07/27 (n = 10324)
Current sample: 1973/01/04 - 1973/12/19 (n = 250)
? garch 1 1 ; r_DJ r_DJ(-1)
Matrix is not positive definite
It does not matter what model or time series I use, the iteration gets stuck somewhere.
In this case in loop 21, sometimes sooner, sometimes later.
I tried to use different window sizes etc, without any luck.
Please let me know if you have any suggestions.
Thank you:
Daniel
Show replies by date