Research projects

Physical dynamics

Engineering

Water-energy dynamics

Data center dynamics connect computing demand, energy supply, and cooling to sustainable operation.Computing demandData centerEnergy supplyCoolingheatWaterDynamic optimization of operationEnergy use · Water use · Emissions

Science

Materials process dynamics

Processing conditions drive material structure evolution. Dynamic models guide the optimization of those conditions.Processing conditionsMaterial formationt₀t₁t₂Initial stateEvolutionTarget structureOptimize the processing trajectory

Water-energy systems

My engineering interests focus on water-energy systems and the sustainable operation of multi-energy systems serving data centers.

Materials processing

I work with Professor Fengqi You on neural operators for dynamic modeling and optimization in precision materials manufacturing.

Learning dynamics

Water demand and pressure

Figure 2 from KbBcM: lag-penalty attention, AR-LSTM, and posterior correction Figure 2 from CritiCoder: coupled Coder and Critic networks with two training loops

KbBcM · CritiCoder

Rolling energy prediction

Figure 3 from the rolling energy paper. Two DCLS-Net stages predict intermediate variables and rolling energy.

Paper

Model2Action

Water-energy scheduling

Figure 2 from SPTO: probabilistic forecasting, water-energy scheduling, and surrogate-loss training

SPTO

Nonlinear control

Figure 1 from the constrained RL controller paper: NMPC initialization and actor-critic feedback Figure 1 from the eligibility traces paper. Actor and critic networks use truncated returns and target critic networks.

Constrained control · Eligibility traces

Robot trajectory generation

Figure 2 from DRIFT: local LiDAR point clouds and sparse graph nodes with a magnified scene detail

DRIFT