A certified gradient-based method for contact-rich manipulation that quantifies smoothing-induced errors via set-valued discrepancies and incorporates them into analytical reachable sets for robust affine feedback policies.
hub
arXiv preprint arXiv:2405.12762 (2 024)
22 Pith papers cite this work, alongside 13 external citations. Polarity classification is still indexing.
22
Pith papers citing it
13
external citations · external index
hub tools
citation-role summary
baseline 1
method 1
other 1
citation-polarity summary
representative citing papers
A learning approach trains neural networks to approximate solutions of multiparametric GNEPs using NI gap loss with value surrogates, achieving large speedups and providing new existence conditions for continuous selections.
Active context sampling algorithm for contextual linear bandits achieves instance-dependent guarantees improving over minimax rate by up to sqrt(d) and reduces samples needed in empirical tasks.
RDDP augments classical dual dynamic programming with OPP-specific backward reachable sets to deliver global optimality certificates and faster computation for both convex and non-convex optimal path parameterization.
Introduces a stochastic DDP algorithm that optimizes nominal controls and feedback gains for belief-state trajectory problems under partial observability without relying on the separation principle.
Proposes HRP-μ, HRP-Σμ, and CRISP as signal-aware extensions to HRP and Cotton-style regularization for mean-variance portfolios, with Monte Carlo results showing outperformance over baselines.
DR-DAQP is a hybrid solver using operator splitting and active-set methods that solves affine variational inequalities exactly in finite time under specified conditions and runs up to two orders of magnitude faster than the PATH solver.
Task-level ILC learns flying knot rope manipulation from one demo, achieving 100% success within 10 trials on 7 rope types with 2-5 trial transfers.
FALCON algorithm solves non-convex partially-decoupled GNEPs via SCP and potential games, claiming global convergence to open-loop Nash equilibria under mild assumptions.
LEAF learns a scalar Moreau envelope via ICNN to create MEL-ADMM and sMEL-ADMM variants that retain ADMM convergence rates while cutting per-iteration cost, with claimed order-of-magnitude empirical speedups.
A mixed-precision ripALM method on GPUs solves large-scale minimum enclosing ball problems faster than CPU geometric software and general conic solvers while maintaining high accuracy.
HUANet unrolls ADMM iterations into a trainable network that enforces equality constraints exactly via a differentiable correction layer and adds soft first-order optimality conditions during training.
Bayesian ddLQR adds posterior uncertainty to the design, decomposing expected cost into certainty-equivalence plus variance terms, proving indirect-direct equivalence, and producing a data-length-independent SDP.
A surrogate for parametric nonconvex optimization is constructed as the minimum of convex-monotonic function compositions and solved via parallel convex optimization, with a proof-of-concept on path tracking.
Establishes O(√ν ln(1/ε)) iteration complexity for path-following smoothing Newton methods on symmetric cone programs via a new self-concordant reduced barrier augmented Lagrangian function and associated central path analysis.
A reaction-wise sparsity decomposition reduces the size of semidefinite constraints in moment bounding for stochastic chemical kinetics, lowering computational cost while retaining useful bounds.
A set of simple low-cost presolve rules captures most of Gurobi's reduction and yields end-to-end speedups for GPU first-order LP solvers.
A filter line search SQP algorithm reduces iterations and computation time for nonconvex SOS programs compared to prior methods.
An integrated lander-propulsion-GNC framework using successive convexification on a test vehicle achieves sub-50-meter landing precision in Monte Carlo simulations under perturbations.
Optimal product pricing with elasticities is formulated as convex-concave maximization and solved via convex-concave procedure, quadratic programs, or nonlinear optimization, with numerical tests indicating the solutions are likely globally optimal.
Convex optimization formulations and an analytical symplectic trace expression are introduced to reconstruct physical Gaussian covariance matrices and witness genuine multipartite entanglement from experimental data.
Asymmetry PRISM-CPU achieves 4.5x-24.1x speedups over reference solvers on N=100-2000 problems and GPU completes all 500 accounts in 109.5s where OSQP completes 4.
citing papers explorer
-
Certified Gradient-Based Contact-Rich Manipulation via Smoothing-Error Reachable Tubes
A certified gradient-based method for contact-rich manipulation that quantifies smoothing-induced errors via set-valued discrepancies and incorporates them into analytical reachable sets for robust affine feedback policies.
-
Learning Approximate Solutions to Multiparametric Generalized Nash Equilibrium Problems
A learning approach trains neural networks to approximate solutions of multiparametric GNEPs using NI gap loss with value surrogates, achieving large speedups and providing new existence conditions for continuous selections.
-
Active Learning for Stochastic Contextual Linear Bandits
Active context sampling algorithm for contextual linear bandits achieves instance-dependent guarantees improving over minimax rate by up to sqrt(d) and reduces samples needed in empirical tasks.
-
Reachability-Augmented Dual Dynamic Programming for Optimal Path Parameterization
RDDP augments classical dual dynamic programming with OPP-specific backward reachable sets to deliver global optimality certificates and faster computation for both convex and non-convex optimal path parameterization.
-
Stochastic Differential Dynamic Programming for Trajectory Optimization under Partial Observability
Introduces a stochastic DDP algorithm that optimizes nominal controls and feedback gains for belief-state trajectory problems under partial observability without relying on the separation principle.
-
Beyond De Prado and Cotton: Hierarchical and Iterative Methods for General Mean-Variance Portfolios
Proposes HRP-μ, HRP-Σμ, and CRISP as signal-aware extensions to HRP and Cotton-style regularization for mean-variance portfolios, with Monte Carlo results showing outperformance over baselines.
-
\texttt{DR-DAQP}: An Hybrid Operator Splitting and Active-Set Solver for Affine Variational Inequalities
DR-DAQP is a hybrid solver using operator splitting and active-set methods that solves affine variational inequalities exactly in finite time under specified conditions and runs up to two orders of magnitude faster than the PATH solver.
-
Learning Dynamic Rope Manipulation Using Task-Level Iterative Learning Control
Task-level ILC learns flying knot rope manipulation from one demo, achieving 100% success within 10 trials on 7 rope types with 2-5 trial transfers.
-
A Fast Convergent Algorithm for Solving Non-convex Partially-Decoupled Generalized Nash Equilibrium Problems
FALCON algorithm solves non-convex partially-decoupled GNEPs via SCP and potential games, claiming global convergence to open-loop Nash equilibria under mild assumptions.
-
LEAF: A Learning-Enabled ADMM Framework for Accelerated Convex Optimization
LEAF learns a scalar Moreau envelope via ICNN to create MEL-ADMM and sMEL-ADMM variants that retain ADMM convergence rates while cutting per-iteration cost, with claimed order-of-magnitude empirical speedups.
-
Mixed-Precision GPU Acceleration for Large-Scale Minimum Enclosing Ball Problems
A mixed-precision ripALM method on GPUs solves large-scale minimum enclosing ball problems faster than CPU geometric software and general conic solvers while maintaining high accuracy.
-
HUANet: Hard-Constrained Unrolled ADMM for Constrained Convex Optimization
HUANet unrolls ADMM iterations into a trainable network that enforces equality constraints exactly via a differentiable correction layer and adds soft first-order optimality conditions during training.
-
A Bayesian Perspective on the Data-Driven LQR
Bayesian ddLQR adds posterior uncertainty to the design, decomposing expected cost into certainty-equivalence plus variance terms, proving indirect-direct equivalence, and producing a data-length-independent SDP.
-
Parametric Nonconvex Optimization via Convex Surrogates
A surrogate for parametric nonconvex optimization is constructed as the minimum of convex-monotonic function compositions and solved via parallel convex optimization, with a proof-of-concept on path tracking.
-
Polynomial iteration complexity of a path-following smoothing Newton method for symmetric cone programming
Establishes O(√ν ln(1/ε)) iteration complexity for path-following smoothing Newton methods on symmetric cone programs via a new self-concordant reduced barrier augmented Lagrangian function and associated central path analysis.
-
Acceleration of Moment Bound Optimization for Stochastic Chemical Reactions Using Reaction-wise Sparsity of Moment Equations
A reaction-wise sparsity decomposition reduces the size of semidefinite constraints in moment bounding for stochastic chemical kinetics, lowering computational cost while retaining useful bounds.
-
Presolving for GPU-Accelerated First-Order LP Solvers
A set of simple low-cost presolve rules captures most of Gurobi's reduction and yields end-to-end speedups for GPU first-order LP solvers.
-
On the Practical Implementation of a Sequential Quadratic Programming Algorithm for Nonconvex Sum-of-squares Problems
A filter line search SQP algorithm reduces iterations and computation time for nonconvex SOS programs compared to prior methods.
-
Integrated Lander-Propulsion-GNC Framework for Autonomous Lunar Powered Descent
An integrated lander-propulsion-GNC framework using successive convexification on a test vehicle achieves sub-50-meter landing precision in Monte Carlo simulations under perturbations.
-
A Note on Optimal Product Pricing
Optimal product pricing with elasticities is formulated as convex-concave maximization and solved via convex-concave procedure, quadratic programs, or nonlinear optimization, with numerical tests indicating the solutions are likely globally optimal.
-
Revisiting Gaussian genuine entanglement witnesses with modern software
Convex optimization formulations and an analytical symplectic trace expression are introduced to reconstruct physical Gaussian covariance matrices and witness genuine multipartite entanglement from experimental data.
-
Asymmetry PRISM: A CPU/GPU Portfolio Optimization Engine for Deadline-Bounded Institutional Rebalancing
Asymmetry PRISM-CPU achieves 4.5x-24.1x speedups over reference solvers on N=100-2000 problems and GPU completes all 500 accounts in 109.5s where OSQP completes 4.