Optimization and Control
Operations research, linear programming, control theory, systems theory, optimal control, game theory
sort pith recommended most recent
Geometric structure of occupation measures yields primal-dual algorithm with O(T^{2/3}) regret and constraint violation bounds.
· “Learning Weakly Communicating Average-Reward CMDPs: Strong Duality and Improved Regret”
A single framework classifies stochastic and deterministic ensemble Kalman filters and supplies a route to new designs.
· “A Unified Control Theory Derivation of Discrete-Time Linear Ensemble Kalman Filters”
It proves Shannon-entropy Bregman iterates converge to critical points on nonconvex problems.
· “On the Iterate Convergence of Bregman Projected Gradient Method”
An epsilon-smoothed polar map yields O(1/T), O(1/t), and exponential rates, plus a local advantage criterion.
Bilevel view turns column scaling into accelerated mirror descent, improving unregularized approximation complexity to O(n²/ε).
· “Accelerating Sinkhorn for Entropy-Regularized Optimal Transport”
End-to-end policy tracks view poses as higher planners ensure safe paths between user-specified regions.
· “Aerial Inspection Behaviors via RL-based Quadrotor Control for Under-canopy Forest Environments”
Joint phase-space model with smooth second moment follows discrete steps for eight optimizers where prior approximations break down.
Policy gradient training yields faster solves than standard or supervised methods on electric vehicle problems and generalizes to new sizes.
· “Learning to Cut: Reinforcement Learning for Benders Decomposition”
FSBA locates approximate second-order stationary points of the hyper-objective faster than first-order baselines under nonconvex-strongly-cx
· “Second-Order Bilevel Optimization with Accelerated Convergence Rates”
Calibrated low-fidelity Darcy model lets topology search find turbulent two-fluid exchangers with better heat transfer and controlled drag
Moment-based ambiguity set plus robust chance constraint yields a deterministic reformulation solvable by nested branch-and-price for cyclic
· “The Distributionally Robust Cyclic Inventory Routing Problem”
Real-world tests show stronger economics, maintained feasibility, and faster consistent solutions than standard methods under demand and net
Decentralized momentum tracking decouples network topology from data heterogeneity and reaches the physical error floor with absolute bias.
· “Improved Convergence for Decentralized Stochastic Optimization with Biased Gradients”
ARGFree combines randomized finite differences and tracking variables to converge in expectation, with a momentum variant that smooths high-
Collaborative barrier functions let agents in coupled systems assist higher-priority peers while keeping the whole group safe through local,
· “Collaborative Altruistic Safety in Coupled Multi-Agent Systems”
Nesterov-accelerated projection with subspace embeddings preserves accuracy while cutting full-model runtime from over a minute to under 3 s
Integrated servicing strategy also trims launch and manufacturing expenses by 25% each in real-scale case study
· “On-Orbit Servicing-Integrated Maintenance Strategy for Satellite Constellation”
It solves composite nonsmooth convex problems with smooth maps on manifolds and shows cluster points are stationary when iterates stay bound
· “An inexact variable metric proximal linearization method for composite optimization on manifolds”
For uncertain nonlinear MIMO systems with input constraints, the design reduces complexity while guaranteeing performance when saturation is
· “An improved input-constrained funnel controller for nonlinear systems”
Mirror descent and proximal mirror descent with logarithmic barriers achieve a tight O(log k / k) convergence rate when the optimum lies on…
Harmonic-mean weights give the tightest majorizer of the smoothed nuclear norm; one-sided IRLS cannot match its local rate.
· “Tight Majorizations and Convergence Rates of Nuclear Norm Minimization IRLS”
Geodesic and curvature results for three spaces all reduce to one chain of maps.
Maneuver time splits into rigid-body minimum plus a bounded synchronization cost, so synthesis becomes a single scalar inversion.
The gap left open since 1996 is closed: 4/9 is the exact worst case in the plane.
A unified framework covers general separable kernels and mirror flows, closing a decades-old iterate convergence gap.
· “A Unified Framework for Iterate Convergence of Bregman Proximal Methods”
Closes the gap between the 1/2 lower bound and the series upper bound for 1D fast controls.
· “Optimal cost of fast boundary controls for the one-dimensional heat equation”
The stored previous output sample creates modes fast sampling cannot move; a simple spectral test decides when.
· “Stability of MIMO PID With Backward Differences Under Fast Sampling: An Exact Spectral Criterion”
The bound is sharp for both BB rules, in Hilbert space, and near nonlinear minimizers.
Even with the right policy class and value features, learning lands far above it until the lookahead horizon is lengthened.
· “Learning Suffers More Than the Policy Class Under Partial Observability: A Closed-Form Analysis”
Translation-mode separation yields integrable coupling and Monte Carlo rates that hold for all time.
· “Uniform-in-time propagation of chaos for Second-Order Consensus-Based Optimization”
when initial loss is small and log-sum-exp functions remain linearly independent modulo affine functions
Finite-dimensional trajectories reach primal-dual solutions with o(t^{-2}) rates for α > 3 via Bregman arguments.
Extending delay-thresholding to LMO momentum handles heterogeneous workers and matches best known bounds in smooth cases.
· “Ringmaster LMO: Asynchronous Linear Minimization Oracle Momentum Method”
Structural invariants cannot be made consistent by small perturbations, yet the conditions hold almost everywhere after generic random ones.
The minimal expected infinity-norm imbalance for high-dimensional online sign choices equals the narrowest L2-controlled Brownian interval.
· “The Mean-Field Limit of Online Stochastic Vector Balancing”
The synchronized branch is the only smooth family emerging from the uniform state and converges smoothly back to it.
· “Uniqueness of synchronized stationary equilibria in the Kuramoto mean field game”
Small-time dynamics governed by weighted squared distances to data supports, for both mixture and amplified guidance
· “Geometric Asymptotics of Score Mixing and Guidance in Diffusion Models”
In IEEE 14-bus simulations, top-ranked nodes worsen frequency and net-power deviations about 50% more than random picks.
· “Adversarially-Informed Node Criticality Identification in Power Grid Measurements”
One Lyapunov proof gives linear convergence for FGM, triple momentum, and two accelerated HB variants.
An explicit degenerate mixture refutes the assumption that the convex roof always has one.
· “On supporting affine functionals for Entanglement of Formation”
A compact square partition yields a MILP that beats direct QP on sequential bilinear programs.
· “When MILP Beats QP: Piecewise-Linear Reformulations of Sequentially Coupled Bilinear Programs”
Under convex uncertainty, beliefs about the nominal problem pay off exactly when robust minimax duality fails.
One Lyapunov estimate yields both polynomial and exponential rates for every transition family.
Closed-form projection replaces iterative fitting and unifies logarithmic, Saaty, and SVD windowing.
Trust one QKD device, treat the other as a black box: convergent bounds plus finite-size security against general attacks.
· “Computing key rates for one-sided device-independent quantum key distribution”
Trained on synthetic workloads, it migrates 7.7x fewer slices than reactive balancing at similar runtimes.
· “Scalable Multi-GPU Simulation of 3D Multicellular Growth with RNN-Based Workload Balancing”
Two theorems transfer variational convergence of data to stability of optimal values and solutions.
With only the fraction of disagreeing labels, the paper's power law sets the reach that minimizes forgetting.
Previously only suboptimality had a rate under weak growth; now the gap the algorithm checks for termination does too.
A single dissipativity condition replaces ellipticity and yields an almost-sure pseudo-trajectory property for optimization.
The normalized counts satisfy the ultra log-concavity inequality; three matroids can break it.
· “A note on the ultra log-concavity of matroid intersection”
A unified equation yields model-based, sensory, incremental, and hybrid NDI laws and shows how to choose between them.
· “Dynamic Inversion: An Incrementally Evolving Methodology for Flight Control Design”
SGHA reaches the best-known ε-dependence in a single loop using first-order gradients only.
· “SGHA: A Single-Loop Fully First-Order Algorithm for Nonconvex-Strongly-Convex Bilevel Optimization”
Coefficients may depend on the whole solution's law, including the L2 jump integrand, with no smallness restriction.
· “Well-posedness of fully coupled McKean-Vlasov FBSDEs with jumps under full-tuple law dependence”
A convex-lifting proximal method needs only O~(d epsilon^-3) function evaluations and keeps iterates feasible.
Trace-class noise keeps the convergence rate independent of Galerkin dimension, so spatial refinement preserves the guarantees.
· “Convergence of time-discrete finite particle consensus based optimization in Hilbert spaces”
Joint ordering and markdown decisions with early, gradual price cuts yield the waste–profit balance.
· “Dynamic Markdown Strategies for Perishable Products: Waste Reduction versus Profitability”
Even biased Top-k sparsification keeps the centralized O(σ/√(NK)) rate, after a transient N^3/(δ^4Δ^4).
· “CED-EF: Compressed Exact Diffusion with Error Feedback for Multi-Agent Learning”
A recourse system that is feasible at every point of a polyhedral support admits a local affine policy at every vertex if and only if the…
· “When Is Two-Stage Sample Robust Optimization Asymptotically Optimal? A Simple Perspective”
Initial-regularization SGD achieves noiseless excess risk up to O(m^{-3+epsilon}) for source parameter r=1, improving on the prior r=1/2…
A federated multiobjective algorithm using per-objective momentum variance-reduced estimators achieves O(T^{-2/3}) expected Pareto…
· “A Momentum-Based Variance-Reduced Algorithm for Federated Multiobjective Optimization”
A 3D-cell collision-risk framework for LEO shows that grid resolution changes expected impact counts nonmonotonically, and a hypothetical…
· “Korean Space Collision Environment Assessment Framework Based on 3D-Cell Model”
Every fully semimonotone matrix with positive determinant is a P_0-matrix, for arbitrary order n.
The principal's value in a finite-horizon continuous-time moral-hazard problem equals the concave envelope of the principal's utility…
· “Backward SDE characterization of the finite horizon Principal-Agent problem”
A decision-diagram framework yields ideal convex hull formulations for mixed-integer quadratic sets with indicator variables, with…
· “Convexification of mixed-integer quadratic optimization via decision diagrams”
One Newton-Schulz iteration is enough to give an infeasible Riemannian gradient method on the Stiefel manifold global convergence and the…
The paper proves convergence of distributional Bellman and moment recursions for optimal control in JMDPs, and validates them in tabular…
· “Learning to Control Coupled-Dynamics Environments with Joint Markov Decision Processes”
A generative framework using convex neural networks learns worst-case distributions for Sinkhorn-based robust hypothesis testing, enabling…
· “Generative Neural Networks for Sinkhorn Distributionally Robust Hypothesis Testing”
Every square of a polynomial is a mean polynomial, which yields a Positivstellensatz with mean polynomial certificates and a cone strictly…
· “The Cone Generated by Positive Semidefinite Mean Polynomials”
For linear-cost positive-dynamics games, feedback Nash equilibria are characterized by coupled vector ODEs or algebraic equations, giving…
A three-port admittance model and plant-level aggregation show that balanced power allocation improves stability margins of…
An existence theorem for Volterra sweeping processes with upper semicontinuous convex-valued perturbations under a…
· “Volterra Sweeping Processes with Multivalued Perturbations under Compactness Conditions”
A nonlinear forward-backward algorithm converges weakly, and under extra conditions R-linearly, for Lipschitz inclusions satisfying a…
Two online algorithms learn a latent coefficient field from a single trajectory at roughly n^{-1/2} sup-norm error.
· “Q-Learning with Stable Infinite-Dimensional Linear Function Approximation”
A strict interior condition on the terminal force yields explicit stopping bounds and a finite-convergence proximal algorithm.
With monotone safety constraints, optimal chance-constrained steering collapses to current-state feedback, losslessly.
· “On the Optimality of Markovian Policies for Chance-Constrained Covariance Steering”
New implicit-function theorem pins down explicit stability exponents for solution mappings of generalized equations.
A parameterized curvature estimate replaces line search and reaches ε-stationary points at the optimal O(ε⁻²) iteration count.
· “Adaptive Barzilai-Borwein Proximal Gradient Method for Nonconvex Optimization”
Each subsystem solves one local problem; stability follows from an M-step Lyapunov estimate and a checked prediction margin
· “Distributed model predictive control via finite-step control Lyapunov functions”
An optimistic model-based algorithm learns to maximize the tau-quantile of return, with a matching lower bound tied to a plateau width.
· “Risk-Sensitive Reinforcement Learning with Smoothed Quantile Objectives”
