I can only from dense least square problem example in eigen?
can i use the build in sparse matrix to solve the least square problem?
if yes, then how to do it?
thank you
P.s. MS VC++ 2012

Two options, either form the normal equation and solve it with either SimplicialLDLT or SparseLU or ConjugateGradients solvers or directly use the SparseQR solver. See these two pages for examples:

http://eigen.tuxfamily.org/dox-devel/gr ... parse.html
http://eigen.tuxfamily.org/dox-devel/gr ... stems.html