This paper aims to introduce an algorithm for solving large scale least squ
ares problems subject to quadratic inequality constraints. The algorithm re
casts the least squares problem in terms of a parameterized eigenproblem. A
variant of k-step Arnoldi method is determined to be well suited for compu
ting the parameterized eigenpair. A two-point interpolating scheme is devel
oped for updating the parameter. A local convergence theory for this algori
thm is presented. It is shown that this algorithm is superlinearly converge
nt.