SearcharxivSearch

arXiv subjects

Dengyu Zheng

Publications and source records attributed to Dengyu Zheng.

2 recordsLinked to original sources

A Regularized Newton-Type Method for Manifold--Affine Intersection Problems under Intrinsic Transversality

We propose Regularized Newton-SLRA (RN-SLRA), a regularized Newton-type method for local manifold--affine intersection problems motivated by structured low-rank approximation. Classical Newton-SLRA achieves fast local convergence under transversality, but its tangent-space intersection step may become ill-defined, singular, or severely ill-conditioned when transversality fails. RN-SLRA overcomes this difficulty by replacing the exact tangent-space intersection step with a regularized quadratic subproblem over the affine space. Under intrinsic transversality, RN-SLRA with regularization parameter $μ_k=c r_k^ρ$ converges locally with Q-order at least $1+ρ$ for $0<ρ\le1$, including quadratically for $ρ=1$, and linearly for $ρ=0$. We also analyze an inexact variant based on $σ$-quasioptimal manifold projections. It retains the $1+ρ$ order for $0<ρ\leq1$, and converges linearly for fixed regularization when $qσ<1$, where $q$ is the local one-step alternating-projection factor. Experiments support the predicted quadratic behavior on an analytically clean but nontransversal SLRA instance, where the classical Newton tangent system is severely ill-conditioned, and show reduced projection costs for the inexact method on large-scale Hankel problems.

math.OC

Descent-Net: Learning Descent Directions for Constrained Optimization

Deep learning approaches, known for their ability to model complex relationships and fast execution, are increasingly being applied to solve large optimization problems. However, existing methods often face challenges in simultaneously ensuring feasibility and achieving an optimal objective value. To address this issue, we propose Descent-Net, a neural network designed to learn an effective descent direction from a feasible solution. By updating the solution along this learned direction, Descent-Net improves the objective value while preserving feasibility. Our method demonstrates strong performance on both synthetic optimization tasks and the real-world AC optimal power flow problem, while also exhibiting effective scalability to large problems, as shown by portfolio optimization experiments with thousands of assets.

math.OC