cholesky of a matrix with deleted rows and columns: chol(Q*A*Q')

For a large matrix A, Cholesky factor L exists

L = chol(A)

B is a matrix of smaller dimension obtained by deleting rows and columns via a 0/1 filter matrix:

B = Q*A*Q'

Is there a way of computing chol(B) or chol(Q*A*Q') directly from A, Q and L, without actually computing B?

Thanks

Answer 1

[Matt J]

 [~,L_new] = qr(L*Q.',0);

[Walter Roberson]

That does not involve A, the filter matrix, so it is not the solution Sue was hoping for.

[Matt J]

Sue's post stipulates that the solution be "directly from A, Q and L, without actually computing B". My proposal does directly use Q and also L and does not compute B. I didn't require A.

[Sue]

Thanks, this works. I simply needed to compute chol(B). ~S