Efficient storage of large symmetric matrix for quadratic programming

[Andreas Elben]

Dear all, 

My aim is to solve large scale (10^5 unknowns) unconstrained quadratic programs of the form 

minimize x^T P x + q^T x

via cvxopt.solvers.qp(P,q). The matrix P is symmetric and positive definite. Since the problem is large, is it possible to store (and hand to cvxopt.solvers.qp(P,q)) only the upper (or lower) triangular part of the symmetric matrix P?

Thank you!