Mathematical Optimization

Convex Optimization Theory and Algorithms

Research on fundamental properties, complexity bounds, and efficient algorithms for convex optimization problems in continuous domains.

RESEARCH GAP FRONTIERS

Non-Convex Optimization Landscapes

Investigation of loss landscape geometry, critical points, and convergence guarantees in non-convex optimization problems.

RESEARCH GAP FRONTIERS

Stochastic Gradient Descent Methods

Analysis and development of variance-reduced and adaptive SGD variants for large-scale machine learning optimization.

RESEARCH GAP FRONTIERS

Distributed Optimization Algorithms

Design of optimization methods for decentralized and federated learning across multiple agents and computing nodes.

RESEARCH GAP FRONTIERS

Second-Order Optimization Methods

Development and analysis of Newton-type methods, quasi-Newton methods, and natural gradient approaches for efficient optimization.

RESEARCH GAP FRONTIERS

Integer and Combinatorial Optimization

Exact and approximate algorithms for discrete optimization problems including integer programming and NP-hard combinatorial problems.

RESEARCH GAP FRONTIERS

Linear Programming and Polyhedral Theory

Theoretical advances in simplex methods, interior-point methods, and polytope geometry for linear optimization.

RESEARCH GAP FRONTIERS

Semidefinite Programming Relaxations

Research on SDP formulations, lift-and-project hierarchies, and approximation algorithms via convex relaxations.

Variational Inequalities and Monotone Operators

Study of variational problem formulations, fixed-point theory, and operator splitting methods for optimization.

Proximal Methods and Operator Splitting

Analysis of proximal gradient, ADMM, Douglas-Rachford, and forward-backward splitting algorithms for composite optimization.

Frank-Wolfe and Conditional Gradient Methods

Development and convergence analysis of projection-free optimization algorithms for constrained convex problems.

Accelerated Optimization Methods

Theory and practice of Nesterov acceleration, momentum methods, and fast gradient techniques for convex optimization.

Online Learning and Regret Bounds

Analysis of online optimization algorithms, adversarial learning, and cumulative loss minimization with theoretical guarantees.

Bandit Optimization and Exploration

Research on multi-armed bandits, contextual bandits, and exploration-exploitation trade-offs in sequential decision-making.

Gradient-Free and Black-Box Optimization

Development of derivative-free methods including genetic algorithms, Bayesian optimization, and zeroth-order methods.

Bayesian Optimization and Gaussian Processes

Study of acquisition functions, surrogate models, and uncertainty quantification for expensive black-box function optimization.

Reinforcement Learning Control Problems

Optimization formulations of RL including Markov decision processes, policy gradient methods, and value function learning.

Robust Optimization and Uncertainty

Design of optimization approaches for problems with uncertain parameters, distributionally robust formulations, and worst-case analysis.

Bilevel Optimization Problems

Theory and algorithms for nested optimization problems arising in meta-learning, hyperparameter optimization, and game theory.

Multi-Objective Optimization Pareto Methods

Research on Pareto frontier computation, scalarization methods, and evolutionary algorithms for multiple conflicting objectives.

Network Flow and Transportation Problems

Optimization algorithms for maximum flow, minimum cost flow, and large-scale network routing and assignment problems.

Facility Location and Clustering

Approximation algorithms and optimization methods for facility location, k-means clustering, and spatial optimization problems.

Scheduling and Resource Allocation

Optimization techniques for job scheduling, machine allocation, load balancing, and temporal resource management.

Machine Learning Feature Selection

Sparse optimization methods, L0/L1 regularization, and combinatorial approaches for high-dimensional feature selection.

Deep Neural Network Training Optimization

Research on loss surface properties, convergence of SGD in deep learning, optimization dynamics, and implicit regularization effects.

Tensor and Low-Rank Matrix Optimization

Algorithms for tensor decomposition, low-rank matrix recovery, and multilinear optimization problems.

Matrix Completion and Missing Data

Optimization methods for matrix completion, collaborative filtering, and recovery from incomplete observations.

Compressed Sensing and Sparse Recovery

Theory and algorithms for sparse signal recovery from limited measurements using convex and non-convex methods.

Online Convex Optimization Games

Study of game-theoretic optimization, Nash equilibrium computation, and competitive optimization in multi-player settings.

Nash Equilibrium and Variational Approaches

Optimization methods for finding Nash equilibria in games and variational formulations of equilibrium problems.

Supply Chain and Logistics Optimization

Optimization algorithms for inventory management, vehicle routing, warehouse location, and supply network design.

Power System and Energy Optimization

Optimization methods for optimal power flow, microgrid control, renewable energy integration, and smart grid management.

Portfolio Optimization and Finance

Mathematical frameworks for portfolio selection, risk management, option pricing, and financial derivatives hedging.

Derivative-Free Trust Region Methods

Development of trust region frameworks and model-based optimization methods without gradient information.

Parameter Tuning and Hyperparameter Optimization

Methods for automated hyperparameter selection, grid search alternatives, and meta-algorithm optimization.

Quantum Optimization Algorithms

Quantum-classical hybrid approaches, variational quantum algorithms, and quantum annealing for optimization problems.

Convergence Rate Analysis and Complexity

Theoretical lower bounds, matching upper bounds, and complexity class analysis for optimization algorithms.

Federated Learning and Privacy-Preserving Optimization

Optimization methods for distributed learning with privacy constraints, communication-efficient algorithms, and differential privacy.

Neural Architecture Search Optimization

Optimization of neural network architectures, AutoML methods, and efficient exploration of architecture spaces.

Smooth and Non-Smooth Analysis

Theoretical framework comparing smooth and non-smooth optimization, subdifferential calculus, and generalized derivatives.

Variational Methods in Physics

Optimization formulations arising in physics, calculus of variations, and PDE-constrained optimization problems.

Data-Driven and Learning-Augmented Optimization

Integration of machine learning predictions with optimization algorithms for improved decision-making.

Graph and Network Optimization

Optimization on graphs, maximum cut problems, graph partitioning, and spectral methods for network optimization.

Stochastic Optimization with Constraints

Algorithms for optimization under stochastic constraints, chance-constrained programming, and scenario-based approaches.

Coordinate Descent and Decomposition Methods

Analysis of block coordinate descent, randomized coordinate methods, and decomposition approaches for large-scale problems.

Saddle Point and Minimax Optimization

Algorithms for saddle point problems, min-max optimization, and adversarial training formulations.

Generative Adversarial Network Training

Optimization theory and methods for GAN training, convergence of adversarial optimization, and stability analysis.

Constraint Qualification and Duality Theory

Study of constraint qualifications, Lagrangian duality, KKT conditions, and strong duality in constrained optimization.

Mixed-Integer Nonlinear Programming

Algorithms for MINLP problems combining discrete decisions and nonlinear constraints through cutting planes and reformulation.

Augmented Lagrangian Methods

Theory and practice of augmented Lagrangian algorithms, multiplier methods, and penalty function approaches.

Manifold Optimization and Riemannian Geometry

Studies optimization algorithms on curved manifolds using differential geometry to handle constraints naturally in non-Euclidean spaces.

Zeroth-Order Optimization and Function Evaluations

Develops optimization methods using only function value queries without gradient information for expensive or black-box objective functions.

Variance Reduction Techniques in Stochastic Methods

Investigates SVRG, SAGA, and other variance reduction strategies to improve convergence rates in stochastic optimization.

Asynchronous Distributed Optimization Networks

Analyzes convergence and efficiency of asynchronous algorithms for distributed optimization without synchronization barriers.

Mirror Descent and Bregman Divergences

Studies generalizations of gradient descent using Bregman divergences and mirror maps for non-Euclidean geometries.

Decentralized Consensus Optimization Methods

Examines peer-to-peer optimization algorithms where agents reach consensus without centralized coordination or parameter servers.

Hardness of Approximation in Optimization

Establishes computational lower bounds and approximation hardness results for NP-hard optimization problems.

Submodular Optimization and Greedy Algorithms

Develops approximation algorithms and complexity analysis for discrete submodular maximization problems.

Stochastic Mirror Descent and Online Games

Applies stochastic mirror descent to online learning and game-theoretic optimization in non-Euclidean spaces.

Extragradient Methods and Monotone Inclusion

Analyzes extragradient and forward-backward splitting methods for solving monotone inclusions and variational problems.

Zeroth-Order Momentum and Acceleration

Develops accelerated zeroth-order methods combining momentum techniques with gradient-free optimization.

Optimization Under Differential Privacy Constraints

Studies privacy-preserving optimization algorithms that provide differential privacy guarantees during learning.

Portfolio Theory and Mean-Variance Optimization

Analyzes advanced portfolio selection models beyond Markowitz including robust and multi-period optimization.

Cutting Plane Methods and Ellipsoid Algorithms

Investigates cutting plane and ellipsoid algorithm theory with applications to convex and combinatorial optimization.

Conic Programming and Self-Concordant Functions

Studies interior point methods and polynomial-time algorithms using self-concordant barrier functions for conic programs.

Derivative-Free Optimization with Surrogate Models

Develops and analyzes surrogate-based methods like space mapping and trust region models for expensive functions.

Frank-Wolfe Variants and Linear Minimization Oracle

Extends Frank-Wolfe algorithms with variance reduction, acceleration, and applications to structured constraints.

Primal-Dual Optimization and Duality Gaps

Analyzes primal-dual algorithms and their convergence properties for constrained optimization problems.

Nonconvex Landscape Characterization and Geometry

Characterizes critical point structure and landscape geometry of nonconvex problems in machine learning.

Cyclic and Block Coordinate Descent Methods

Develops theory and applications of coordinate descent with cyclic, randomized, or greedy coordinate selection.

Trust Region Methods and Newton Steps

Analyzes trust region frameworks combining gradient and second-order information with adaptive radius strategies.

Projected Gradient Descent with Constraints

Studies projected gradient methods for constrained optimization with analysis of projection complexity.

Conditional Expectation and Variance in Sampling

Analyzes importance sampling and stratified sampling methods to reduce variance in stochastic optimization.

Game Theory and Multi-Agent Optimization

Studies convergence to equilibria in multi-agent optimization and game-theoretic learning dynamics.

Sparse Optimization and L0 Minimization

Develops algorithms and relaxations for exact and approximate sparse optimization with cardinality constraints.

Time-Varying and Non-Stationary Optimization

Analyzes optimization algorithms that track solutions of time-dependent or non-stationary objectives.

Greedy Algorithms and Approximation Guarantees

Establishes approximation ratios and performance bounds for greedy algorithmic approaches to hard problems.

Convex Relaxations and SDP Hierarchies

Studies sum-of-squares and Lasserre hierarchies for computing optimal values of combinatorial problems.

Dual Averaging and Mirror Maps Theory

Investigates dual averaging algorithms and their connections to mirror descent in online optimization.

Parallel Optimization and Speedup Analysis

Analyzes speedup and communication complexity of parallel optimization algorithms across multiple processors.

Proximal Alternating Linearized Method

Studies PALM and related alternating minimization methods for nonconvex composite optimization.

Warm Starting and Transfer Learning in Optimization

Investigates leveraging solutions of related problems to accelerate convergence in new optimization instances.

Projection-Free Methods and Conditional Gradients

Extends projection-free optimization to handle complex geometries and non-Euclidean constraints efficiently.

Quasi-Newton Methods and Hessian Approximation

Develops BFGS, L-BFGS and other quasi-Newton methods with convergence guarantees and memory efficiency.

Optimization with Implicit Regularization

Studies how optimization algorithms themselves induce implicit regularization without explicit penalty terms.

Compositional Optimization and Nested Structures

Analyzes algorithms for composite objectives with nested expectations or function compositions.

Stochastic Optimization under Heavy Tails

Develops robust stochastic optimization methods for heavy-tailed noise and bounded variance conditions.

Gradient Descent Escape from Saddle Points

Analyzes how gradient descent with perturbations escapes saddle points in nonconvex optimization.

Continuous Relaxations of Discrete Programs

Studies continuous relaxations and rounding techniques for converting discrete problems to continuous optimization.

Convex Geometry and Volume Estimation

Applies convex geometry and randomization for approximating volumes and solving geometric optimization.

Stochastic Variance-Reduced Methods with Momentum

Combines variance reduction techniques with momentum acceleration for improved stochastic convergence.

Parametric and Sensitivity Analysis Optimization

Analyzes how optimal solutions change as function parameters vary using sensitivity and parametric methods.

Gradient Tracking and Gossip Algorithms

Studies decentralized gradient tracking and gossip-based communication for distributed consensus optimization.

Conditional Gradient Methods with Acceleration

Develops accelerated variants of conditional gradient methods with improved convergence rates.

Optimization with Monotone and Maximal Operators

Studies algorithms for monotone inclusions and extensions to maximal monotone operators and set-valued maps.

Consensus and Distributed Machine Learning

Develops distributed algorithms for consensus averaging and federated learning across heterogeneous nodes.

Stochastic Optimization Lower Bounds

Establishes information-theoretic and computational lower bounds for stochastic optimization complexity.

Splitting Methods and Operator Theory

Analyzes Douglas-Rachford, alternating direction, and other operator splitting schemes using monotone operator theory.

Structured Sparsity and Group Norms

Develops optimization methods for group sparsity and structured sparse recovery using group norms.

Optimal Transport and Wasserstein Distances

Studies optimization problems arising from optimal transport theory and computation of Wasserstein distances.

Adaptive Learning Rate Scheduling Optimization

Research on dynamically adjusting learning rates during optimization to balance convergence speed and solution quality across heterogeneous problem instances.

Composite Optimization with Nonsmooth Penalties

Study of optimization problems combining smooth loss functions with nonsmooth regularizers for enhanced interpretability and sparse solutions.

Asynchronous Parallel Optimization Convergence

Analysis of convergence properties in asynchronous distributed settings where communication delays and out-of-date information affect algorithm performance.

Conditional Value-at-Risk Optimization Applications

Study of risk-sensitive optimization using CVaR metrics in financial, engineering, and operational decision-making problems.

Derivative-Free Evolutionary Algorithm Analysis

Theoretical and empirical study of genetic algorithms and evolution strategies for problems where gradients are unavailable or expensive.

Zeroth-Order Optimization for High Dimensions

Development of function evaluation-based methods for large-scale optimization without gradient information in ultra-high dimensional spaces.

Decentralized Consensus-Based Optimization Networks

Study of optimization algorithms for decentralized networks where agents coordinate through local communication to reach global consensus.

Monotone Inclusion and Fixed Point Methods

Research on algorithms for solving monotone inclusion problems and finding fixed points of nonlinear operators.

Primal-Dual Splitting Schemes Convergence

Analysis of alternating primal-dual algorithms for minimax problems and convex-concave saddle point formulations.

Conic Programming and Cone Geometry

Study of optimization over general cones beyond semidefinite programming with applications to copositive and sum-of-squares hierarchies.

Submodular Function Maximization Approximation

Research on approximation algorithms and hardness results for maximizing submodular objectives in combinatorial and machine learning contexts.

Reformulation-Linearization Technique Development

Study of problem reformulation strategies that tighten linear relaxations of mixed-integer programs for improved bound quality.

Cutting Plane Methods and Branch-and-Cut

Development of advanced cutting plane generation and separation algorithms for integer programming with dynamic cut management.

Penalty Methods and Exact Penalization

Research on converting constrained problems into penalty-based unconstrained problems with theoretical exactness guarantees.

Natural Gradient and Riemannian Optimization

Study of optimization on Riemannian manifolds using curvature-aware geometry for machine learning and statistical applications.

Oracle-Based Lower Bounds Complexity Theory

Fundamental research establishing lower bounds on algorithm query complexity and convergence rates for optimization problems.

Momentum-Based Acceleration Mechanisms Analysis

Investigation of heavy ball methods, Nesterov acceleration, and polyak momentum in various optimization settings and variants.

Robust Optimal Transport and Wasserstein

Study of optimal transport computation with robustness to adversarial perturbations and approximation guarantees.

Trust Region Methods Globalization Strategies

Research on trust region frameworks balancing local quadratic approximation with global convergence guarantees for nonconvex problems.

Alternating Direction Method of Multipliers Variants

Development and analysis of ADMM extensions for general nonconvex, distributed, and asynchronous problem settings.

Constraint Reduction and Infeasibility Detection

Development of preprocessing and identification techniques for removing redundant constraints and detecting infeasible problems early.

Stochastic Dual Averaging Online Algorithms

Research on dual averaging methods for online optimization with improved regret bounds under various feedback models.

Nonconvex Landscape Escaping Local Minima

Study of algorithms and analysis techniques for escaping local minima in nonconvex optimization problems effectively.

First-Order Methods for Minimax Optimization

Development of gradient-based algorithms for finding approximate Nash equilibria in zero-sum games and minimax formulations.

Sketching and Randomized Dimensionality Reduction

Research on using random projections and sketch matrices to accelerate large-scale optimization through dimension reduction.

Coordinate Minimization and Block Updates

Study of algorithms that optimize over coordinate blocks or subsets sequentially for distributed and scalable computation.

Nonlinear Programming Constraint Qualification

Theoretical analysis of constraint qualifications enabling strong duality and error bounds in nonlinear programs.

Optimistic Gradient Methods Best Possible Rates

Study of algorithms exploiting problem smoothness with optimistic updates to achieve near-optimal convergence rates.

Stochastic Variance-Reduced Gradient Methods

Research on SVRG and SAGA algorithms achieving linear convergence for strongly convex problems with finite sum structure.

Extragradient and Forward-Backward Variants

Analysis of extragradient methods and forward-backward algorithms for monotone inclusions and variational inequalities.

Adaptive Subgradient Methods Convergence Guarantees

Study of AdaGrad, Adam, and other adaptive first-order methods with provable convergence in smooth and composite settings.

Dual Decomposition and Lagrangian Relaxation

Research on relaxing complex optimization problems through dual formulations for solving large-scale and structured instances.

Approximate Message Passing Algorithm Analysis

Study of approximate message passing algorithms in factor graphs with applications to compressed sensing and signal recovery.

Smoothing Techniques for Nonsmooth Problems

Development of Moreau envelope and other smoothing methods transforming nonsmooth problems into smooth approximations.

Majorization-Minimization Algorithms Convergence

Research on surrogate-based optimization using majorizers with applications to signal processing and statistics.

Nonconvex Regularized Regression Phase Transitions

Study of computational and statistical phase transitions in nonconvex M-estimation with sparsity or low-rank constraints.

Implicit Differentiation for Optimization Layers

Research on differentiating through optimization solvers for end-to-end learning of prediction models with embedded optimization.

Online Gradient Descent Dynamic Regret Analysis

Study of regret bounds for online convex optimization in time-varying environments with drifting comparators.

Composite Convex Function Optimization Acceleration

Development of optimal algorithms for problems with composite structure combining smooth and nonsmooth convex functions.

Sparse Principal Component Analysis Optimization

Research on efficient algorithms for computing principal components with sparsity constraints and theoretical guarantees.

Sequential Convex Programming Methods Development

Study of successive approximation of nonconvex problems by convex subproblems with convergence to stationary points.

Optimal Control and Trajectory Optimization Methods

Research on optimization algorithms for trajectory planning and control problems in robotics and autonomous systems.

Graph Neural Network Training Optimization

Study of specialized optimization methods for training graph neural networks accounting for graph structure and scalability.

Randomized Block Coordinate Descent Rates

Analysis of convergence rates for randomized coordinate descent with various block selection strategies and step sizes.

Subsampled Newton Methods Large-Scale Learning

Research on Newton-type methods using subsampled Hessian information for efficient second-order optimization at scale.

Exponential Family and Natural Parameter Estimation

Study of optimization algorithms for maximum likelihood estimation in exponential family models with constraints.

Inexact Gradient Descent Perturbation Analysis

Theoretical framework for analyzing convergence when gradients are computed inexactly or approximately.

Variance Reduction in Stochastic Optimization

Development of techniques to reduce gradient estimator variance in stochastic algorithms for improved convergence rates and sample efficiency.

Asynchronous Parallel Optimization Methods

Design and analysis of optimization algorithms that operate without synchronization barriers across distributed computing systems.

Manifold Optimization and Riemannian Methods

Optimization on curved geometries and non-Euclidean spaces using Riemannian geometry and differential geometry frameworks.

Conditional Gradient Dynamics and Flows

Continuous-time analysis and differential equation models of conditional gradient algorithms and their limiting behaviors.

Monotone Inclusion Problems and Splitting

Theory and algorithms for solving systems of monotone inclusions using operator splitting and decomposition techniques.

Finite-Time Convergence Rate Analysis

Characterization of non-asymptotic convergence rates for optimization algorithms under various smoothness and convexity assumptions.

Zeroth-Order Optimization and Gradient Estimation

Optimization using only function value evaluations without access to gradient information through randomized probing methods.

Proximal Point Algorithms and Regularization

Convergence analysis and acceleration of proximal point methods for ill-posed and regularized optimization problems.

Adaptive Learning Rate and Preconditioning

Design of adaptive step size rules and preconditioning strategies that dynamically adjust to problem geometry and data properties.

Compositional Optimization and Nested Functions

Optimization of composite functions where objectives involve nested function compositions requiring specialized algorithmic approaches.

Decentralized Consensus Optimization Networks

Optimization algorithms where agents communicate locally without central coordination to reach collective consensus on solutions.

Non-Ergodic and Last-Iterate Convergence

Analysis of optimization algorithms focusing on convergence of final iterates rather than time-averaged solutions.

Implicit Differentiation and Implicit Bias

Study of optimization''s implicit tendency to find solutions with specific properties without explicit regularization.

Hamiltonian and Symplectic Optimization Dynamics

Differential equation models of optimization based on Hamiltonian mechanics and symplectic structure preservation.

Heavy Ball and Momentum Acceleration Methods

Analysis and improvement of momentum-based methods inspired by physics for accelerated convergence in optimization.

Dual Averaging and Mirror Descent Methods

First-order methods using Bregman divergences and mirror maps to handle non-Euclidean geometries and constraints.

Submodular Function Maximization Algorithms

Efficient algorithms for optimizing non-linear submodular objectives arising in machine learning and combinatorics.

Constraint Handling and Active Set Methods

Algorithms for constrained optimization that identify and maintain active constraints during the optimization process.

Regularized Empirical Risk Minimization Theory

Theoretical foundations connecting statistical learning theory to optimization of regularized loss functions.

Wasserstein Gradient Flows and Optimal Transport

Optimization via gradient flows in Wasserstein distance spaces with applications to machine learning and PDEs.

Differential Privacy in Optimization Algorithms

Design of optimization methods that maintain rigorous privacy guarantees while achieving convergence.

Decoupled and Separable Optimization Structures

Exploitation of separable problem structures through decoupling techniques for improved computational efficiency.

Smoothing and Regularization Techniques Optimization

Methods for approximating non-smooth objectives with smooth surrogates to enable gradient-based optimization.

Universal Gradient Methods and Adaptivity

Gradient methods that automatically adapt to unknown problem parameters and achieve optimal rates across problem classes.

Prox-Linear and Composite Gradient Methods

Optimization algorithms combining proximal steps with gradient steps for composite objectives with non-smooth components.

Stochastic Approximation and Recursive Algorithms

Theory of iterative algorithms using noisy gradient observations for root-finding and optimization problems.

Localization and Local Convergence Analysis

Detailed convergence analysis of optimization algorithms in neighborhoods of optimal solutions and critical points.

Extragradient and Extrapolation Methods

Optimization algorithms using prediction and extrapolation steps to improve convergence in monotone and variational problems.

Sketching and Dimensionality Reduction Optimization

Optimization using random projections and sketched gradients to reduce computational and memory requirements.

Projected and Constrained Optimization Dynamics

Continuous-time differential equations and discrete algorithms for optimization on constrained manifolds and feasible sets.

Averaged Gradient and SAG Methods

Stochastic optimization using averaged gradient information from past iterations for variance reduction.

Interior Point and Barrier Methods Modern Theory

Contemporary analysis of interior point methods for conic programming with focus on iteration complexity.

Stochastic Variational Inequalities Solutions

Algorithms for solving variational inequality problems subject to stochastic or noisy observations.

Escaping Saddle Points in Non-Convex Optimization

Methods for efficiently escaping saddle points and finding local minima in non-convex landscapes.

Gradient Clipping and Norm-Constrained Optimization

Analysis of gradient clipping strategies and norm constraints for robustness and convergence in optimization.

Distributed Gradient Compression and Quantization

Optimization algorithms using quantized and compressed gradient communication for bandwidth reduction.

Free Lunch Theorems and Lower Bounds

Fundamental limits and impossibility results characterizing lower bounds on convergence rates for optimization classes.

Orthogonal and Structured Matrix Optimization

Optimization algorithms for matrices constrained to specific structures like orthogonal or symmetric groups.

Implicit Regularization and Generalization

Investigation of how optimization algorithms induce implicit regularization affecting generalization performance.

Consensus-Based Sampling and Derivative-Free Methods

Derivative-free optimization using consensus mechanisms and ensemble-based sampling techniques.

Nonlocal Operators and Fractional Calculus Optimization

Optimization using nonlocal differential operators and fractional order dynamics.

Continuous-Time Optimization and ODEs

Analysis of continuous differential equation models of optimization algorithms and their discretization.

Inexact Proximal and Approximate Methods

Convergence theory for algorithms where proximal operators and subproblems are solved approximately.

Gradient Tracking and Dual Correction Methods

Distributed optimization using gradient tracking and dual variable updates for exact convergence.

Submodular Optimization and Greedy Approximations

Study of optimization problems with submodular objective functions and development of approximation algorithms with provable performance guarantees.

Distributed Federated Learning and Decentralized Training

Investigation of optimization algorithms for training machine learning models across distributed systems without centralizing data while preserving privacy.

Polyak-Lojasiewicz Condition and Quadratic Growth

Optimization under Polyak-Lojasiewicz and quadratic growth conditions for accelerated convergence analysis.

Implicit Differentiation and Bilevel Gradient Computation

Development of efficient gradient computation methods for nested optimization problems using implicit function theorems for hyperparameter optimization.

Peer-to-Peer and Gossip Optimization Networks

Distributed optimization on peer-to-peer networks using gossip and local communication protocols.

Stochastic Variance-Reduced Methods and Catalyst Acceleration

Analysis of variance reduction techniques combined with acceleration schemes to achieve faster convergence in large-scale stochastic optimization.

Monotone Operator Splitting and Proximal Algorithms

Theory and algorithms for solving structured optimization problems through splitting techniques applied to monotone inclusion problems and operator compositions.

Bilevel and Hierarchical Game-Theoretic Learning

This area investigates optimization frameworks where multiple decision-makers interact strategically across hierarchical levels, combining game theory with learning dynamics to analyze equilibrium convergence and algorithmic solutions in competitive and cooperative optimization settings.

Zeroth-Order Optimization for Black-Box Functions

Research on optimization algorithms that access only function values without gradient information, addressing derivative-free optimization in high-dimensional spaces and noisy black-box settings where gradient computation is prohibitively expensive or unavailable.

Optimal Transport and Wasserstein Distance Methods

Research on computational methods for solving optimal transport problems and leveraging Wasserstein distances as loss functions in machine learning, with applications to generative modeling, domain adaptation, and statistical inference.