# Research by Tongxin Li My research asks how learning can improve decisions when predictions, models, and information are imperfect. The work connects mathematical guarantees with control, AI agents, and the operation of energy infrastructure. ## Themes ### Learning with predictions When should a decision maker trust a prediction? Algorithms that benefit from accurate advice while limiting the cost of errors. This thread studies consistency, robustness, competitive analysis, and the value of different forms of predictive information. https://tongxin.me/research/themes/predictions/ ### Safe and distributed control How can a controller learn while preserving reliable operation? Predictive control with performance and feasibility guarantees. The work spans perturbation analysis, nonlinear policy certification, aggregate flexibility, and distributed system level synthesis. https://tongxin.me/research/themes/control/ ### Language and decision agents How can language and world models support dependable decisions? Language models provide context, preferences, and predictions for sequential decisions. Algorithmic guidance and structural certification connect these capabilities to explicit performance guarantees. https://tongxin.me/research/themes/agents/ ### Energy and infrastructure How can intelligent infrastructure use energy more effectively? Control and learning for electric vehicle charging, renewable integration, demand forecasting, water systems, and grid coordination. Physical constraints and changing operating conditions shape the algorithms. https://tongxin.me/research/themes/energy/ ### Strategic and equitable learning What changes when decisions affect other decision makers? Learning and optimization with interacting agents. This work examines uncertain beliefs, strategic advice, equilibrium, and the distribution of benefits from shared predictive models. https://tongxin.me/research/themes/games/ ### Information and network learning What information and structure are needed to learn reliably? Information limits and structured learning, from group testing and communication channels to graph recovery and adaptive graph geometry. These results connect limited measurements to useful representations. https://tongxin.me/research/themes/structure/ ## Papers ### Adaptive Riemannian Graph Neural Networks Xudong Wang, Chris Ding, Tongxin Li, Jicong Fan. Adaptive Riemannian Graph Neural Networks. Proceedings of the AAAI Conference on Artificial Intelligence, 2026. https://doi.org/10.1609/aaai.v40i31.39869 How can a graph model adapt to different local geometries? ARGNN learns a node specific Riemannian metric instead of imposing one fixed curvature on an entire graph. A diagonal metric parameterization and geometric regularization support efficient training, with convergence analysis and experiments on homophilic and heterophilic graphs. https://tongxin.me/research/papers/argnn/ ### Counterfactual load forecasting with LLM-structured events and representation learning Yujie Chen, Yifei Gao, Runyao Yu, Yuhe Wu, Guangyu Wang, Yue Chen, Tongxin Li. Counterfactual load forecasting with LLM-structured events and representation learning. Applied Energy, 2026. https://doi.org/10.1016/j.apenergy.2026.128554 How might electricity demand change under a different news event? NACF turns news into structured treatments and estimates load trajectories under alternative event conditions. Reweighting and representation balancing address observed confounding. Experiments examine factual accuracy and interpretable demand perturbations without claiming that unobserved counterfactual outcomes can be directly validated. https://tongxin.me/research/papers/counterfactual-load/ ### Energy Injection Identification enabled Disaggregation with Deep Multi-Task Learning Xudong Wang, Guoming Tang, Junyu Xue, Srinivasan Keshav, Tongxin Li, Chris Ding. Energy Injection Identification enabled Disaggregation with Deep Multi-Task Learning. ACM International Conference on Future and Sustainable Energy Systems, 2026. https://doi.org/10.1145/3744255.3798113 Can appliances be identified when solar and storage obscure meter readings? DualNILM jointly recognizes appliance states and identifies energy injected behind the meter. Its transformer architecture combines temporal learning tasks to separate consumption from injections, with evaluation on measured and synthesized datasets. https://tongxin.me/research/papers/dualnilm/ ### Leveraging Machine-Learned Advice in Strategic Interactions with No-Regret Learners Tinashe Handina, Tongxin Li, Kishan Panaganti, Eric Mazumdar, Adam Wierman. Leveraging Machine-Learned Advice in Strategic Interactions with No-Regret Learners. Proceedings of the International Conference on Artificial Intelligence and Statistics, 2026. https://arxiv.org/abs/2606.10261 How useful is imperfect advice against an adaptive opponent? A measure of advice quality connects simulators and payoff predictions to strategic performance. The paper establishes benefits of reliable advice for approximate Stackelberg play and limitations on simultaneously exploiting accurate advice and protecting against inaccurate advice. https://tongxin.me/research/papers/strategic-advice/ ### PEARL: A Physics-Enhanced Adaptive Residual Learning Framework for PV Modeling Yikai Lu, Yujie Chen, Tongxin Li. PEARL: A Physics-Enhanced Adaptive Residual Learning Framework for PV Modeling. IEEE PES International Meeting, 2026. https://doi.org/10.1109/PESIM67009.2026.11438881 Can learned residuals improve physical models of solar generation? PEARL separates a physical photovoltaic simulator from a learned residual correction. A controlled benchmark compares several machine learning architectures and a coupled modeling baseline, examining accuracy while retaining the physical model as an interpretable reference. https://tongxin.me/research/papers/pearl/ ### Prediction-Specific Design of Learning-Augmented Algorithms Sizhe Li, Nicolas Christianson, Tongxin Li. Prediction-Specific Design of Learning-Augmented Algorithms. Proceedings of the ACM on Measurement and Analysis of Computing Systems, 2026. https://doi.org/10.1145/3788100 Can an algorithm be optimal for the particular prediction it receives? Strong optimality refines the usual consistency and robustness tradeoff to account for each prediction. A bilevel optimization framework constructs algorithms for ski rental and one max search, with case studies in power management and trading. https://tongxin.me/research/papers/prediction-specific/ ### Quantum Learning and Estimation for Coordinated Operation between Distribution Networks and Energy Communities Yingrui Zhuang, Lin Cheng, Yuji Cao, Tongxin Li, Ning Qi, Yan Xu, Yue Chen. Quantum Learning and Estimation for Coordinated Operation between Distribution Networks and Energy Communities. CSEE Journal of Power and Energy Systems, 2026. https://doi.org/10.17775/CSEEJPES.2025.05720 Can quantum methods assist coordination with limited local information? A hybrid quantum learning model estimates energy community responses to prices, while quantum amplitude estimation supports uncertainty calculations. Numerical studies assess accuracy and resource requirements. Reported timing advantages assume ideal quantum devices rather than deployed hardware. https://tongxin.me/research/papers/quantum-coordination/ ### VigilMPC: Certified Online Tube Model Predictive Control for Safe Real-Time Neuromodulation Yu Mao, Zhichao Liang, Junxiang Zhang, Quanying Liu, Tongxin Li. VigilMPC: Certified Online Tube Model Predictive Control for Safe Real-Time Neuromodulation. IEEE Control Systems Letters, 2026. https://doi.org/10.1109/LCSYS.2026.3711599 Can a predictive controller safely update its learned dynamics online? VigilMPC accepts a predictor update only when a backup plan remains feasible under tightened constraints. Recursive feasibility and constraint satisfaction follow under the tube MPC assumptions. Evaluation uses simulated neural dynamics and a switching benchmark, rather than clinical deployment. https://tongxin.me/research/papers/vigilmpc/ ### World Models in Pieces: Structural Certification for General Agents Yikai Lu, Yifei Wu, Xinyu Lu, Tongxin Li. World Models in Pieces: Structural Certification for General Agents. Proceedings of the International Conference on Machine Learning, 2026. https://arxiv.org/abs/2606.24842 Which parts of an agent's world model can support reliable planning? Structural certification links performance on compositional goals to local guarantees on world model transitions. The results identify reliable pieces of a model without requiring universal competence, and establish limits on the precision of such guarantees. https://tongxin.me/research/papers/world-models/ ### Beyond Numeric Rewards: In-Context Dueling Bandits with LLM Agents Fanzeng Xia, Hao Liu, Yisong Yue, Tongxin Li. Beyond Numeric Rewards: In-Context Dueling Bandits with LLM Agents. Findings of the Association for Computational Linguistics, 2025. https://doi.org/10.18653/v1/2025.findings-acl.519 Can language agents learn reliably from pairwise preferences? The study identifies a gap between quick preference discovery and sustained exploitation by language agents. LEAD combines LLM reasoning with dueling bandit algorithms to obtain weak and strong regret guarantees, with evaluations under noisy and adversarial prompts. https://tongxin.me/research/papers/lead/ ### Hierarchical Game for Coupled Power System With Energy Sharing and Transportation System Dongxiang Yan, Tongxin Li, Changhong Zhao, Han Wang, Yue Chen. Hierarchical Game for Coupled Power System With Energy Sharing and Transportation System. IEEE Transactions on Transportation Electrification, 2025. https://doi.org/10.1109/TTE.2024.3398988 How do charging, traffic, and local energy trading influence one another? A hierarchical game couples an energy sharing mechanism with transportation decisions and power network constraints. The paper establishes properties of the energy sharing equilibrium and reformulates the coupled problem for numerical solution. https://tongxin.me/research/papers/coupled-networks/ ### INSTRUCT MPC: A Human-LLM-in-the-Loop Framework for Context-Aware Control Ruixiang Wu, Jiahao Ai, Tongxin Li. INSTRUCT MPC: A Human-LLM-in-the-Loop Framework for Context-Aware Control. IEEE Conference on Decision and Control, 2025. https://doi.org/10.1109/CDC57313.2025.11312156 How can human instructions improve predictive control? A language to distribution module converts contextual instructions into disturbance predictions for MPC. The framework closes the loop between instructions, predictions, and control, with a regret analysis for linear dynamics under the stated training assumptions. https://tongxin.me/research/papers/instruct-mpc/ ### Learning-Augmented Online Control for Decarbonizing Water Infrastructures Jianyi Yang, Pengfei Li, Tongxin Li, Adam Wierman, Shaolei Ren. Learning-Augmented Online Control for Decarbonizing Water Infrastructures. ACM International Conference on Future and Sustainable Energy Systems, 2025. https://doi.org/10.1145/3679240.3734595 Can learned pump control reduce carbon costs while meeting safety constraints? LAOC restricts online decisions to safe action sets while exploiting learned predictions for pump scheduling. Analysis characterizes the relationship between safety requirements and operating costs, with a building water supply case study. https://tongxin.me/research/papers/water-control/ ### Open In-Context Energy Management Platform Yikai Lu, Tinko Sebastian Bartels, Ruixiang Wu, Fanzeng Xia, Xudong Wang, Yifei Wu, Haoxiang Yang, Tongxin Li. Open In-Context Energy Management Platform. ACM International Conference on Future and Sustainable Energy Systems, 2025. https://doi.org/10.1145/3679240.3734678 What would a shared benchmark for contextual energy management provide? OpenCEM presents a platform design connecting energy time series, events, human context, and simulation. An on campus solar and battery installation motivates evaluation of context sensitive control. The paper describes planned data and API capabilities. https://tongxin.me/research/papers/opencem/ ### Out-of-Distribution-Aware Electric Vehicle Charging Tongxin Li, Chenxi Sun. Out-of-Distribution-Aware Electric Vehicle Charging. IEEE Transactions on Transportation Electrification, 2025. https://doi.org/10.1109/TTE.2024.3434750 How should charging schedules respond when user behavior changes? OOD Charging adapts an awareness radius using temporal difference errors to balance learned scheduling with a robust baseline. Evaluation on ACN data examines behavioral changes during COVID 19 and the resulting distribution shift. https://tongxin.me/research/papers/ood-charging/ ### PredSLS: A System-Level Framework for Distributed Predictive Control Yifei Wu, Jing Yu, Tongxin Li. PredSLS: A System-Level Framework for Distributed Predictive Control. IEEE Conference on Decision and Control, 2025. https://doi.org/10.1109/CDC57313.2025.11312851 How can distributed controllers use predictions with limited communication? PredSLS integrates forecasts directly into system level synthesis with locality constraints. Temporal decay of prediction effects supports finite horizon approximations, while spatial and temporal decomposition enable scalable computation across a network. https://tongxin.me/research/papers/predsls/ ### Reinforcement Learning with Imperfect Transition Predictions: A Bellman-Jensen Approach Chenbei Lu, Zaiwei Chen, Tongxin Li, Chenye Wu, Adam Wierman. Reinforcement Learning with Imperfect Transition Predictions: A Bellman-Jensen Approach. Advances in Neural Information Processing Systems, 2025. https://papers.nips.cc/paper_files/paper/2025/hash/940a7634dab556b67af15bacd337f7db-Abstract-Conference.html How can reinforcement learning use imperfect forecasts beyond one step? A Bayesian value function and Bellman Jensen gap quantify the value of imperfect transition forecasts. BOLA separates offline value learning from online adaptation, with sample efficiency analysis and experiments in synthetic environments and wind energy storage control. https://tongxin.me/research/papers/bellman-jensen/ ### Robust machine-learned algorithms for efficient grid operation Nicolas Christianson, Christopher Yeh, Tongxin Li, Mehdi Hosseini, Mahdi Torabi Rad, Azarang Golmohammadi, Adam Wierman. Robust machine-learned algorithms for efficient grid operation. Environmental Data Science, 2025. https://doi.org/10.1017/eds.2024.28 Can learned dispatch policies remain reliable under distribution shift? RobustML combines a learned dispatch algorithm with a bound on worst case cost. A combined cycle cogeneration case study examines dispatch and ramping costs as renewable penetration increases, including performance under distribution shift. https://tongxin.me/research/papers/robust-grid/ ### Robust Solar Hosting Capacity for Pasadena's Decarbonization: A Mathematical Program for Robustness Against Uncertainty Han Xu, Yiheng Xie, Wenqi Cui, Tongxin Li, Steven Low. Robust Solar Hosting Capacity for Pasadena's Decarbonization: A Mathematical Program for Robustness Against Uncertainty. IEEE Energy Sustainability Magazine, 2025. https://doi.org/10.1109/ESM.2025.3606174 How much distributed solar can a grid accommodate under uncertainty? The Pasadena case study formulates solar hosting capacity planning as a robust optimization problem. It connects uncertainty in renewable generation and demand to distribution network constraints and practical choices along a municipal decarbonization pathway. https://tongxin.me/research/papers/solar-hosting/ ### A Data-Driven Approach for Optimizing Early-Stage Electric Vehicle Charging Station Placement Chenxi Sun, Tongxin Li, Xiaoying Tang. A Data-Driven Approach for Optimizing Early-Stage Electric Vehicle Charging Station Placement. IEEE Transactions on Industrial Informatics, 2024. https://doi.org/10.1109/TII.2023.3245633 How can early charging infrastructure encourage electric vehicle adoption? Two proximity models represent user satisfaction with station locations. Population, travel, road network, and point of interest data inform a submodular optimization problem, solved approximately with a greedy method and evaluated in Haikou. https://tongxin.me/research/papers/early-stations/ ### Building Socially-Equitable Public Models Yejia Liu, Jianyi Yang, Pengfei Li, Tongxin Li, Shaolei Ren. Building Socially-Equitable Public Models. Proceedings of the International Conference on Machine Learning, 2024. https://proceedings.mlr.press/v235/liu24bw.html Who benefits when many decision makers share one predictive model? The Equitable Objective trains a public predictive model around the outcomes of heterogeneous downstream agents. A policy gradient method and theoretical analysis study a more equitable distribution of performance, beyond optimizing prediction accuracy alone. https://tongxin.me/research/papers/equitable-models/ ### Disentangling Linear Quadratic Control with Untrusted ML Predictions Tongxin Li, Hao Liu, Yisong Yue. Disentangling Linear Quadratic Control with Untrusted ML Predictions. Advances in Neural Information Processing Systems, 2024. https://papers.nips.cc/paper_files/paper/2024/hash/9dff3b83d463fab213941bfee23341ba-Abstract-Conference.html Can a controller learn which components of a forecast to trust? DISC learns confidence in predictions of latent disturbance components. Competitive analysis covers linear and more general mixing functions, showing how online confidence adaptation can exploit accurate forecasts while maintaining protection against large errors. https://tongxin.me/research/papers/disc/ ### Safe Exploitative Play with Untrusted Type Beliefs Tongxin Li, Tinashe Handina, Shaolei Ren, Adam Wierman. Safe Exploitative Play with Untrusted Type Beliefs. Advances in Neural Information Processing Systems, 2024. https://papers.nips.cc/paper_files/paper/2024/hash/858e773f5c99a4186b548f3f531fb07e-Abstract-Conference.html How much risk comes from exploiting a predicted opponent type? The paper formalizes a risk and opportunity tradeoff for agents with uncertain beliefs about other players. Upper and lower bounds characterize the Pareto frontier in normal form and stochastic Bayesian games, with numerical illustrations of the tradeoff. https://tongxin.me/research/papers/safe-play/ ### Anytime-Competitive Reinforcement Learning with Policy Prior Jianyi Yang, Pengfei Li, Tongxin Li, Adam Wierman, Shaolei Ren. Anytime-Competitive Reinforcement Learning with Policy Prior. Advances in Neural Information Processing Systems, 2023. https://papers.nips.cc/paper_files/paper/2023/hash/f53437debdd397c42929d929614bc705-Abstract-Conference.html Can reinforcement learning enforce cost protection throughout an episode? ACRL constrains cumulative costs relative to a policy prior at each round, rather than only in expectation over episodes. The analysis establishes cost guarantees and regret relative to the constrained optimum, with experiments in carbon aware computing. https://tongxin.me/research/papers/anytime-rl/ ### Beyond Black-Box Advice: Learning-Augmented Algorithms for MDPs with Q-Value Predictions Tongxin Li, Yiheng Lin, Shaolei Ren, Adam Wierman. Beyond Black-Box Advice: Learning-Augmented Algorithms for MDPs with Q-Value Predictions. Advances in Neural Information Processing Systems, 2023. https://papers.nips.cc/paper_files/paper/2023/hash/8e806d3c56ed5f1dab85d601e13cbe38-Abstract-Conference.html Does the structure of advice improve robust decision making? Q value predictions expose more information than an opaque policy recommendation. For a single trajectory MDP, the analysis characterizes consistency and robustness tradeoffs and shows how structured advice improves the guarantees available from a robust baseline. https://tongxin.me/research/papers/q-value-advice/ ### Certifying Black-Box Policies With Stability for Nonlinear Control Tongxin Li, Ruixiao Yang, Guannan Qu, Yiheng Lin, Adam Wierman, Steven H. Low. Certifying Black-Box Policies With Stability for Nonlinear Control. IEEE Open Journal of Control Systems, 2023. https://doi.org/10.1109/OJCSYS.2023.3241486 How can model based advice stabilize a learned nonlinear policy? Simply blending two stabilizing policies can cause instability. An adaptive confidence policy instead uses approximate model information to certify stability, with competitive guarantees under bounded nonlinearity and evaluations on Cart Pole and charging under distribution shift. https://tongxin.me/research/papers/certified-policies/ ### Bounded-Regret MPC via Perturbation Analysis: Prediction Error, Constraints, and Nonlinearity Yiheng Lin, Yang Hu, Guannan Qu, Tongxin Li, Adam Wierman. Bounded-Regret MPC via Perturbation Analysis: Prediction Error, Constraints, and Nonlinearity. Advances in Neural Information Processing Systems, 2022. https://papers.nips.cc/paper_files/paper/2022/hash/eadeef7c51ad86989cc3b311cb49ec89-Abstract-Conference.html How do forecast errors translate into regret for MPC? A general analysis pipeline converts perturbation bounds for finite horizon control into dynamic regret bounds for MPC. It handles prediction errors in costs, dynamics, and disturbances, and extends the analysis to constrained and nonlinear settings under suitable regularity conditions. https://tongxin.me/research/papers/bounded-regret-mpc/ ### Robustness and Consistency in Linear Quadratic Control with Untrusted Predictions Tongxin Li, Ruixiao Yang, Guannan Qu, Guanya Shi, Chenkai Yu, Adam Wierman, Steven Low. Robustness and Consistency in Linear Quadratic Control with Untrusted Predictions. Proceedings of the ACM on Measurement and Analysis of Computing Systems, 2022. https://doi.org/10.1145/3508038 Can predictive control balance consistency and robustness automatically? A confidence parameter governs how much a linear quadratic controller trusts disturbance predictions. Competitive bounds describe the tradeoff between accurate and inaccurate advice, and a self tuning policy adapts confidence online using observed prediction quality. https://tongxin.me/research/papers/robust-lqc/ ### Towards balanced three-phase charging: Phase optimization in adaptive charging networks Zixin Ye, Tongxin Li, Steven Low. Towards balanced three-phase charging: Phase optimization in adaptive charging networks. Electric Power Systems Research, 2022. https://doi.org/10.1016/j.epsr.2022.108322 How should electric vehicle demand be assigned across electrical phases? PXA solves a relaxation of the phase optimization problem, with convergence guarantees under specified conditions. Its MPC implementation updates decisions online. Experiments use adaptive charging network data to examine energy delivery, charging costs, and phase balance. https://tongxin.me/research/papers/phase-optimization/ ### Data-driven Electric Vehicle Charging Station Placement for Incentivizing Potential Demand Chenxi Sun, Tongxin Li, Xiaoying Tang. Data-driven Electric Vehicle Charging Station Placement for Incentivizing Potential Demand. IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids, 2021. https://doi.org/10.1109/SmartGridComm51999.2021.9632309 Where should a city place charging stations with a limited budget? A proximity based satisfaction model turns station placement into monotone submodular maximization. A greedy algorithm has a provable approximation ratio, with a Haikou case study using population, travel, and point of interest data. https://tongxin.me/research/papers/station-placement/ ### Decentralized Provision of Renewable Predictions Within a Virtual Power Plant Yue Chen, Tongxin Li, Changhong Zhao, Wei Wei. Decentralized Provision of Renewable Predictions Within a Virtual Power Plant. IEEE Transactions on Power Systems, 2021. https://doi.org/10.1109/TPWRS.2020.3035174 Can local renewable forecasts support decentralized market coordination? A decentralized prediction provision scheme lets consumers obtain local forecasts and exchange summaries with a virtual power plant. The analysis studies social surplus, convergence, and the demand gap relative to centralized information sharing. https://tongxin.me/research/papers/vpp-predictions/ ### Information Aggregation for Constrained Online Control Tongxin Li, Yue Chen, Bo Sun, Adam Wierman, Steven Low. Information Aggregation for Constrained Online Control. Proceedings of the ACM on Measurement and Analysis of Computing Systems, 2021. https://doi.org/10.1145/3460085 How much feasibility information must a local controller share? Entropic aggregation summarizes a local controller's feasible actions for a remote decision maker. Penalized Predictive Control can approach the regret of full information control under causal invariance and a sufficiently long prediction window. https://tongxin.me/research/papers/information-aggregation/ ### Learning-Based Predictive Control via Real-Time Aggregate Flexibility Tongxin Li, Bo Sun, Yue Chen, Zixin Ye, Steven H. Low, Adam Wierman. Learning-Based Predictive Control via Real-Time Aggregate Flexibility. IEEE Transactions on Smart Grid, 2021. https://doi.org/10.1109/TSG.2021.3094719 How can an aggregator communicate flexibility in real time? Maximum entropy feedback summarizes the feasible actions of controllable loads. Reinforcement learning approximates this feedback for Penalized Predictive Control, reducing information and computation requirements. Charging data demonstrates the coordination approach. https://tongxin.me/research/papers/aggregate-flexibility/ ### Classification of electric vehicle charging time series with selective clustering Chenxi Sun, Tongxin Li, Steven H. Low, Victor O. K. Li. Classification of electric vehicle charging time series with selective clustering. Electric Power Systems Research, 2020. https://doi.org/10.1016/j.epsr.2020.106695 What battery behavior can be learned from imperfect charging records? An iterative procedure extracts and clusters the tail portions of charging curves despite missing observations, variable lengths, scheduling effects, and measurement noise. Experiments on ACN data illustrate how these patterns can support useful charging models. https://tongxin.me/research/papers/charging-clusters/ ### Competitive Algorithms for the Online Multiple Knapsack Problem with Application to Electric Vehicle Charging Bo Sun, Ali Zeynali, Tongxin Li, Mohammad Hajiesmaili, Adam Wierman, Danny H. K. Tsang. Competitive Algorithms for the Online Multiple Knapsack Problem with Application to Electric Vehicle Charging. Proceedings of the ACM on Measurement and Analysis of Computing Systems, 2020. https://doi.org/10.1145/3428336 How should limited resources be allocated to requests arriving online? A general fractional multiple knapsack model captures assignment and rate constraints. An online primal dual algorithm obtains near optimal competitive guarantees and is evaluated using charging traces. This work provides an online allocation foundation without requiring learned predictions. https://tongxin.me/research/papers/online-knapsack/ ### Learning Graphs From Linear Measurements: Fundamental Trade-Offs and Applications Tongxin Li, Lucien Werner, Steven H. Low. Learning Graphs From Linear Measurements: Fundamental Trade-Offs and Applications. IEEE Transactions on Signal and Information Processing over Networks, 2020. https://doi.org/10.1109/TSIPN.2020.2975368 What determines the sample complexity of network reconstruction? Bounds for noisy and noiseless graph recovery connect sparsity, graph distributions, and linear measurements. A three stage recovery scheme and practical algorithm are studied on canonical graphs and electrical grid admittance matrices. https://tongxin.me/research/papers/graph-recovery/ ### ORC: An Online Competitive Algorithm for Recommendation and Charging Schedule in Electric Vehicle Charging Network Bo Sun, Tongxin Li, Steven H. Low, Danny H. K. Tsang. ORC: An Online Competitive Algorithm for Recommendation and Charging Schedule in Electric Vehicle Charging Network. ACM International Conference on Future Energy Systems, 2020. https://doi.org/10.1145/3396851.3397727 How can charging recommendations and schedules be coordinated online? ORC jointly recommends stations and prices while scheduling charging for sequential arrivals. A customizable value function supports competitive guarantees despite uncertainty in arrivals and user choices. Experiments compare operator revenue with benchmark scheduling methods. https://tongxin.me/research/papers/orc/ ### Learning Graph Parameters from Linear Measurements: Fundamental Trade-offs and Application to Electric Grids Tongxin Li, Lucien Werner, Steven H. Low. Learning Graph Parameters from Linear Measurements: Fundamental Trade-offs and Application to Electric Grids. IEEE Conference on Decision and Control, 2019. https://doi.org/10.1109/CDC40024.2019.9029949 How many measurements are needed to recover a network? Information theoretic bounds relate measurement requirements to graph complexity and recovery error. A recovery algorithm estimates topology and parameters from linear measurements, with demonstrations on graph families and electrical grid test cases. https://tongxin.me/research/papers/graph-parameters/ ### Maximum Likelihood Upper Bounds on the Capacities of Discrete Information Stable Channels Tongxin Li. Maximum Likelihood Upper Bounds on the Capacities of Discrete Information Stable Channels. IEEE Information Theory Workshop, 2018. https://doi.org/10.1109/ITW.2018.8613485 How can channel capacity be bounded through maximum likelihood? The paper derives a maximum likelihood upper bound using optimality conditions for discrete information stable channels. It recovers capacity for selected memoryless channels and develops a counting based bound for the binary deletion channel. https://tongxin.me/research/papers/channel-capacity/ ### Quadratically Constrained Channels with Causal Adversaries Tongxin Li, Bikash Kumar Dey, Sidharth Jaggi, Michael Langberg, Anand D. Sarwate. Quadratically Constrained Channels with Causal Adversaries. IEEE International Symposium on Information Theory, 2018. https://doi.org/10.1109/ISIT.2018.8437839 What rates are possible when a jammer observes transmissions causally? Capacity is characterized through a sequence of optimization problems under quadratic power constraints. Analytical and numerical bounds show that nonuniform power allocation can outperform uniform allocation in certain signal to noise regimes. https://tongxin.me/research/papers/causal-channels/ ### Robust and Consistent Clustering Recovery via SDP Approaches Chenxi Sun, Tongxin Li, Victor O. K. Li. Robust and Consistent Clustering Recovery via SDP Approaches. IEEE Data Science Workshop, 2018. https://doi.org/10.1109/DSW.2018.8439918 Can clusters be recovered from noisy similarities without knowing their number? A semidefinite programming approach recovers clustering structure in the presence of noise and outliers. The analysis gives conditions for exact recovery and consistency in a Gaussian similarity model, supported by synthetic experiments. https://tongxin.me/research/papers/clustering-sdp/ ### Fundamental limits and achievable strategies for low energy compressed sensing with applications in wireless communication Tongxin Li, Mayank Bakshi, Pulkit Grover. Fundamental limits and achievable strategies for low energy compressed sensing with applications in wireless communication. IEEE International Workshop on Signal Processing Advances in Wireless Communications, 2016. https://doi.org/10.1109/SPAWC.2016.7536775 What energy is required to decode a sparse signal? A communication complexity model measures the energy required to move information during support recovery. Upper and lower bounds show how adaptive message scheduling can reduce decoding energy, with an application to multiuser detection in wireless systems. https://tongxin.me/research/papers/energy-sensing/ ### Group Testing with Prior Statistics Tongxin Li, Chun Lam Chan, Wenhao Huang, Tarik Kaced, Sidharth Jaggi. Group Testing with Prior Statistics. IEEE International Symposium on Information Theory, 2014. https://doi.org/10.1109/ISIT.2014.6875253 How can prior probabilities reduce the number of group tests? Adaptive and nonadaptive testing algorithms exploit heterogeneous prior probabilities of defective items. Under the stated independence and sparsity assumptions, their measurement requirements approach information theoretic lower bounds up to explicit constant factors. https://tongxin.me/research/papers/group-testing/ ## Connections across the work ### From prediction error to control performance Trust adaptation in linear quadratic control and perturbation bounds for MPC connect the quality of a forecast to the cost of acting on it. Nonlinear certification adds stability to this picture. Disentangling Linear Quadratic Control with Untrusted ML Predictions https://tongxin.me/research/papers/disc/ Certifying Black-Box Policies With Stability for Nonlinear Control https://tongxin.me/research/papers/certified-policies/ Bounded-Regret MPC via Perturbation Analysis: Prediction Error, Constraints, and Nonlinearity https://tongxin.me/research/papers/bounded-regret-mpc/ Robustness and Consistency in Linear Quadratic Control with Untrusted Predictions https://tongxin.me/research/papers/robust-lqc/ ### From language context to constrained actions INSTRUCT MPC translates human instructions into disturbance predictions. World model certification and VigilMPC study complementary ways of deciding which learned models can support reliable planning and safe updates. VigilMPC: Certified Online Tube Model Predictive Control for Safe Real-Time Neuromodulation https://tongxin.me/research/papers/vigilmpc/ World Models in Pieces: Structural Certification for General Agents https://tongxin.me/research/papers/world-models/ INSTRUCT MPC: A Human-LLM-in-the-Loop Framework for Context-Aware Control https://tongxin.me/research/papers/instruct-mpc/ ### Algorithms guide learned decision makers Q value advice and imperfect transition forecasts provide structured information for reinforcement learning. LEAD applies a related principle by supporting language agents with dueling bandit algorithms. Beyond Numeric Rewards: In-Context Dueling Bandits with LLM Agents https://tongxin.me/research/papers/lead/ Reinforcement Learning with Imperfect Transition Predictions: A Bellman-Jensen Approach https://tongxin.me/research/papers/bellman-jensen/ Beyond Black-Box Advice: Learning-Augmented Algorithms for MDPs with Q-Value Predictions https://tongxin.me/research/papers/q-value-advice/ ### Guarantees meet physical infrastructure Aggregate flexibility communicates feasible actions for coordinated charging. Robust dispatch and safe water control connect performance guarantees to operational costs and physical constraints. Learning-Augmented Online Control for Decarbonizing Water Infrastructures https://tongxin.me/research/papers/water-control/ Robust machine-learned algorithms for efficient grid operation https://tongxin.me/research/papers/robust-grid/ Information Aggregation for Constrained Online Control https://tongxin.me/research/papers/information-aggregation/ Learning-Based Predictive Control via Real-Time Aggregate Flexibility https://tongxin.me/research/papers/aggregate-flexibility/ ### Context becomes an energy system input OpenCEM motivates shared contextual energy benchmarks. INSTRUCT MPC uses instructions for control, while counterfactual load forecasting represents news events as interventions on electricity demand. Counterfactual load forecasting with LLM-structured events and representation learning https://tongxin.me/research/papers/counterfactual-load/ INSTRUCT MPC: A Human-LLM-in-the-Loop Framework for Context-Aware Control https://tongxin.me/research/papers/instruct-mpc/ Open In-Context Energy Management Platform https://tongxin.me/research/papers/opencem/ ### Prediction errors become strategic risks In strategic environments, advice quality affects both opportunity and risk. The Bayesian game and repeated game results expose limits on combining exploitation with protection against inaccurate beliefs. Leveraging Machine-Learned Advice in Strategic Interactions with No-Regret Learners https://tongxin.me/research/papers/strategic-advice/ Safe Exploitative Play with Untrusted Type Beliefs https://tongxin.me/research/papers/safe-play/ ### Coordination through information and incentives Virtual power plants use renewable predictions to coordinate consumers. Energy sharing and transportation games model how charging demand, prices, and network constraints interact. Hierarchical Game for Coupled Power System With Energy Sharing and Transportation System https://tongxin.me/research/papers/coupled-networks/ Building Socially-Equitable Public Models https://tongxin.me/research/papers/equitable-models/ Decentralized Provision of Renewable Predictions Within a Virtual Power Plant https://tongxin.me/research/papers/vpp-predictions/ ### Recovering structure from limited measurements Graph recovery estimates electrical network parameters. Charging curve clustering and energy disaggregation recover useful structure from observed behavior and aggregate meter readings. Energy Injection Identification enabled Disaggregation with Deep Multi-Task Learning https://tongxin.me/research/papers/dualnilm/ Classification of electric vehicle charging time series with selective clustering https://tongxin.me/research/papers/charging-clusters/ Learning Graphs From Linear Measurements: Fundamental Trade-Offs and Applications https://tongxin.me/research/papers/graph-recovery/ ## Open research directions These are research questions motivated by the papers, not established results. ### Certified contextual control Can contextual and world model updates be certified before they affect a physical system? Connect language conditioned predictions with local world model guarantees and feasibility checks. An open challenge is preserving safety when both the context and the learned dynamics change during operation. Starting point VigilMPC: Certified Online Tube Model Predictive Control for Safe Real-Time Neuromodulation https://tongxin.me/research/papers/vigilmpc/ Starting point World Models in Pieces: Structural Certification for General Agents https://tongxin.me/research/papers/world-models/ Starting point INSTRUCT MPC: A Human-LLM-in-the-Loop Framework for Context-Aware Control https://tongxin.me/research/papers/instruct-mpc/ ### Prediction quality as a decision resource Which prediction errors actually matter for the decision being made? Move from a single forecast accuracy score to guarantees that reflect the prediction, the task, and the affected transitions. This suggests adaptive confidence allocation across sources, horizons, and latent disturbances. Starting point Prediction-Specific Design of Learning-Augmented Algorithms https://tongxin.me/research/papers/prediction-specific/ Starting point Reinforcement Learning with Imperfect Transition Predictions: A Bellman-Jensen Approach https://tongxin.me/research/papers/bellman-jensen/ Starting point Disentangling Linear Quadratic Control with Untrusted ML Predictions https://tongxin.me/research/papers/disc/ Starting point Beyond Black-Box Advice: Learning-Augmented Algorithms for MDPs with Q-Value Predictions https://tongxin.me/research/papers/q-value-advice/ ### Coordinated low carbon infrastructure Can distributed energy and water resources adapt together under changing conditions? Combine local feasibility summaries with distributed predictive control and contextual energy models. Open questions include communication limits, distribution shift, and guarantees across coupled physical networks. Starting point Learning-Augmented Online Control for Decarbonizing Water Infrastructures https://tongxin.me/research/papers/water-control/ Starting point Open In-Context Energy Management Platform https://tongxin.me/research/papers/opencem/ Starting point Out-of-Distribution-Aware Electric Vehicle Charging https://tongxin.me/research/papers/ood-charging/ Starting point PredSLS: A System-Level Framework for Distributed Predictive Control https://tongxin.me/research/papers/predsls/ Starting point Learning-Based Predictive Control via Real-Time Aggregate Flexibility https://tongxin.me/research/papers/aggregate-flexibility/ ### Accountable learning across agents How should shared models balance efficiency, strategic risk, and equity? Connect the downstream objectives of public models to robust strategic decision making. A key challenge is evaluating advice when its benefits and failures are distributed unevenly across participating agents. Starting point Leveraging Machine-Learned Advice in Strategic Interactions with No-Regret Learners https://tongxin.me/research/papers/strategic-advice/ Starting point Beyond Numeric Rewards: In-Context Dueling Bandits with LLM Agents https://tongxin.me/research/papers/lead/ Starting point Building Socially-Equitable Public Models https://tongxin.me/research/papers/equitable-models/ Starting point Safe Exploitative Play with Untrusted Type Beliefs https://tongxin.me/research/papers/safe-play/