Glossary

RL

Reinforcment Learning. A machine learning method to learn policies trough interaction with an environment.

SARL

Single-Agent Reinforcement Learning. A reinforcement learning setting where a single agent interacts with an environment to learn an optimal policy.

MARL

Multi-Agent Reinforcement Learning. A reinforcement learning setting where multiple agents interact with an environment to learn optimal policies. In this setting, agents typically share a policy and learn collaboratively.

SB3

Stable Baselines3. A popular open-source library for RL algorithms, built on top of PyTorch, providing implementations of various RL methods for easy experimentation and benchmarking.

AFC

Active Flow Control. The use of actuators to manipulate a fluid flow in order to achieve a desired outcome, such as drag reduction or lift enhancement.

RBC

Rayleigh-Bénard Convection. A type of natural convection that occurs in a fluid layer heated from below and cooled from above, leading to the formation of convection cells.

TCF

Turbulent Channel Flow. A canonical flow configuration used to study turbulence in wall-bounded flows, characterized by a fluid flowing between two parallel plates.

PPO

Proximal Policy Optimization. A policy gradient method that improves learning stability by using a clipped surrogate objective, preventing excessive policy updates and ensuring reliable performance.

SAC

Soft Actor-Critic. An off-policy reinforcement learning algorithm that maximizes both expected reward and entropy, encouraging robust exploration and sample efficiency.