Hi Jingjing,
Didn't we already do this the other day:
http://www.stata.com/statalist/archive/2009-03/msg00366.html?
I am wondering if -M= = st_matrix("M")- here make sure it is a *symmetric*
matrix with the elements h11 h21... that I just generated.
-st_matrix("M")- will return whatever is in your stata matrix M. If your
stata matrix M is symmetric, then it will return a symmetric matrix. If not,
then it won't. You've said that your stata matrix is Tx10, so that's what
you'll get.
But I don't think you want M to be symmetric anyway; you want mt to be
symmetric.
The expression
mt = invvech(M[., t])
extracts the tth column of M and tries to turn it into a symmetric matrix.
But that won't work, because it's the *rows* of M that contain the h* values
you want to use.
mt = invvech(M[t,.]')
should give you what you want (the transpose is necessary because
-invvech()- requires a colvector argument).
Alternatively, if you only want to transpose once, it may be slightly more
efficient to code:
M = st_matrix("M")'
At the top of your routine, and leave the rest as is.
Best wishes,
Glenn.
[email protected] wrote:
Thanks Maarten,
I know what's "double" now.
And I have to corrected my expression in the previous email.
I first use -gen-... to calculate the elements of matrix M.
Then -mkmat h*, mat(M)- creat matrix M as T*10, where T is the mumver of
observations in the sample and there are 10 elements in the lower triangle
of the matrix for a specific period
Now I want to calculate the eigenvalues of the matrix for each time period:
mata:
M = st_matrix("M")
lambda = J(cols(M), 4, .)
for(t=1; t<=cols(M); t++) {
mt = invvech(M[., t])
lambda[t, .] = symeigenvalues(mt)
}
lambda
end
I am wondering if -M= = st_matrix("M")- here make sure it is a *symmetric*
matrix with the elements h11 h21... that I just generated.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
