Comprehensive Flow Control Suite
From canonical 2D cylinder flows to complex 3D turbulent scenarios, HydroGym provides standardized environments for systematic RL algorithm development and evaluation.
A Reinforcement Learning Platform for Fluid Dynamics Control
42+ Validated Environments | 3 Numerical Backends | Easy access
From canonical 2D cylinder flows to complex 3D turbulent scenarios, HydroGym provides standardized environments for systematic RL algorithm development and evaluation.
Choose the right solver for your research needs—from high-performance computing to gradient-enhanced optimization
High-Performance CFD for Large-Scale Simulations
Built on RWTH Aachen's m-AIA framework, this backend enables massive parallel simulations with efficient CPU/GPU acceleration.
Gradient-Enhanced RL with Automatic Differentiation
GPU-accelerated differentiable simulations enable gradient-based policy optimization, dramatically improving sample efficiency.
Transparent, Extensible & Customizable Python Framework
Python-based finite element solver prioritizing code clarity and research extensibility with full access to solver internals.
Validated environments with systematic Reynolds number progressions
2D & 3D flows at Re=100-3,900. Drag reduction via rotation and jet actuation. Achieves >20% drag reduction.
Multi-body wake interactions at Re=30-150. Coordinated control of three cylinders demonstrates chaos suppression.
Shear layer stabilization at Re=4,140-7,500. Acoustic feedback disruption through leading-edge control.
Gust mitigation at Re=100-50,000. Transverse gust encounters with load alleviation strategies.
Extreme event mitigation in 2D turbulence. Control energy bursts and enhance mixing efficiency.
Wall-shear stress reduction at Re_τ=180. Gradient-enhanced training with JAX for efficient optimization.