Large-scale Optimization with Linear Equality Constraints using Reduced Compact Representation

For optimization problems with sparse linear equality constraints, we observe that the (1,1) block of the inverse KKT matrix remains unchanged when projected onto the nullspace of the constraints. We develop reduced compact representations of the limited-memory BFGS Hessian to compute search directions efficiently. Orthogonal projections are implemented by sparse QR factorization or preconditioned LSQR iteration. In numerical experiments two proposed trust-region algorithms improve in computation times, often significantly, compared to previous implementations and compared to IPOPT.