Form the correlation matrix
of order order
whose
correlations follow the ar3 pattern. The resulting matrix
is banded.
mat.ar3(ARparameters, order)
A numeric
containing the three autoregressive parameter values
of the process, being the weights given to the lag 1, lag 2 and lag 3
response values.
The order of the matrix
to be formed.
A banded correlation matrix
whose elements follow an ar3 pattern.
The correlations in the correlation matrix, corr
say, are calculated
from the autoregressive parameters, ARparameters
.
Let omega = 1 - ARparameters[2] - ARparameters[3] * (ARparameters[1] + ARparameters[3])
.
Then the values in
the diagonal of corr
(k = 1
) are one;
the first subdiagonal band (k = 2
) of corr
are equal to
(ARparameters[1] + ARparameters[2]*ARparameters[3]) / omega
;
the second subdiagonal band (k = 3
) of corr
are equal to
(ARparameters[1] * (ARparameters[1] + ARparameters[3]) +
ARparameters[2] * (1 - ARparameters[2])) / omega
;
the subsequent subdiagonal bands, (k = 4:order
), of corr
are equal to
ARparameters[1]*corr[k-1] + ARparameters[2]*corr[k-2] + ARparameters[3]*corr[k-3]
.
mat.I
, mat.J
, mat.banded
, mat.exp
, mat.gau
,
mat.ar1
, mat.ar2
, mat.sar2
,
mat.ma1
, mat.ma2
, mat.arma
# NOT RUN {
corr <- mat.ar3(ARparameters = c(0.4, 0.2, 0.1), order = 4)
# }
Run the code above in your browser using DataCamp Workspace