Math Principles of Reinforcement Learning Notes—Windylab

Math Principles of Reinforcement Learning Notes—Windylab

Some simplified notes and thoughts when learning math foundation of RL

Wed Apr 01 2026
23 words · 1 minutes
RL Lecture RL Lecture EAI

第二课：Bellman Equation

Return: Evaluate policies

the sum of the rewards obtained along a trajectory.

return1=γ1−γreturn2return3=−0.5+γ1−γreturn_1=\frac{\gamma}{1-\gamma} \\ return_2 \\ return_3=-0.5+\frac{\gamma}{1-\gamma}

How to calculate?

  • Definition
  • Bootstrapping
v=r+γPvv=r+\gamma P v
Thanks for reading!

Math Principles of Reinforcement Learning Notes—Windylab

Wed Apr 01 2026
23 words · 1 minutes
RL Lecture RL Lecture EAI

© Napucheng | CC BY-NC-SA 4.0

Share this article

Share preview for Math Principles of Reinforcement Learning Notes—Windylab
X (Twitter) Telegram Reddit Facebook Email

Table of Contents