Extracting a large submatrix with column and row names

[Shae Cetinel]

Hi,

Variance-covariance matrix of the estimators has a dimension of 9000 x 9000. I am trying to find a way to extract a submatrix (32x32). It is not very easy to find the indices of the variables. The variables to be extracted from the matrix are sequential, so slicing may also work. I can also make a list of variable names to be extracted from the matrix, but I am not familiar with looping in Stata.
What is a clean and fast way of handling this problem in Stata?

[John Mullahy]

Is the indexing used to create msub and msub1 useful for your purposes?

set seed 2345
mata st_matrix("mall",floor(runiform(5,5):*5))
matrix list mall
matrix msub=mall["r2".."r4","c2".."c4"]
matrix list msub

matrix mall1=mall

matrix rownames mall1 = myrow1 myrow2 myrow3 myrow4 myrow5
matrix colnames mall1 = mycol1 mycol2 mycol3 mycol4 mycol5
matrix list mall1

matrix msub1=mall1["myrow2".."myrow4","mycol2".."mycol4"]
matrix list msub1

Results:

. matrix list mall

mall[5,5]
    c1  c2  c3  c4  c5
r1   1   0   0   1   2
r2   1   1   1   4   3
r3   0   2   3   3   2
r4   2   4   4   3   4
r5   1   0   0   1   3

. matrix msub=mall["r2".."r4","c2".."c4"]

. matrix list msub

msub[3,3]
    c2  c3  c4
r2   1   1   4
r3   2   3   3
r4   4   4   3

.
. matrix mall1=mall

.
. matrix rownames mall1 = myrow1 myrow2 myrow3 myrow4 myrow5

. matrix colnames mall1 = mycol1 mycol2 mycol3 mycol4 mycol5

. matrix list mall1

mall1[5,5]
        mycol1  mycol2  mycol3  mycol4  mycol5
myrow1       1       0       0       1       2
myrow2       1       1       1       4       3
myrow3       0       2       3       3       2
myrow4       2       4       4       3       4
myrow5       1       0       0       1       3

.
. matrix msub1=mall1["myrow2".."myrow4","mycol2".."mycol4"]

. matrix list msub1

msub1[3,3]
        mycol2  mycol3  mycol4
myrow2       1       1       4
myrow3       2       3       3
myrow4       4       4       3

.

[Shae Cetinel]

Hi John,

Thank you for the help. I saved the variance-covariance matrix. It has a size of [8646, 8646].

mata : st_matrix("Co", st_matrix("e(V)"))

Unfortunately I cannot figure out what the indices of columns and rows I am trying to extract because the matrix is too large. There are 38 variables that I am trying to extract from matrix Co.

[Shae Cetinel]

Thanks John for the instructions. Here is what I did that worked for me.

*save variance-covariance matrix
mat Co=e(V) 

*save a submatrix by locating variables with their names
 matrix sub_co = co["tt0".."26.cat_t1", "tt0".."26.cat_t1"]