I'd like to get a right null vector of a large, sparse matrix.  My plan is 
to use SPQR (although I'm happy to consider alternatives).  The 
documentation for SPQR states that if QR factorization is applied to a 
matrix that does not have full column rank, then a 'squeezed' R factor is 
computed from which a right null vector can be deduced.

However, as far as I can see, the Julia interface to SPQR in 
base/sparse/spqr.jl (Julia version 0.4.6) does not provide access to any of 
the factors but only yields an opaque handle to a C data structure.

Could someone give me a clue how to get access to the R factor of the QR 
factorization computed by SPQR?  Ideally I would obtain R as a 
SparseMatrixCSC object.

Thanks,
Steve Vavasis

Reply via email to