SOTA-RL

This repository provides a high-performance Python framework for solving Reliable Routing Problems in stochastic networks. Unlike standard shortest-path algorithms, R2L uses Reinforcement Learning (Q-Learning) to find policies that maximize the probability of reaching a destination within a strict time budget, where edge costs are uncertain.

📁 Project Structure

📁 SOTA-RL
	📁 docs
    📁 external
        📁 SOTA-py
	📁 instances
    📁 notebook
	📁 src
	   🔹 main.py
	   🔹 R2L_plot.py
	   🔹 R2L_test.py
	   🔹 R2L_train.py
		   🔸GridNet(ABC)
           🔸afGridNet
           🔸reachGridNet
           🔸embeddedGridNet
	   🔹 R2L_utilities.py
		   🔸R2LTestFunctions

All modules' full documentation can be found in docs folder:

• Train Module Documentation
• Test Module Documentation
• Utilities Module Documentation

Requirements

Recommended packages:

numpy
networkx
scipy
math
abc
pickle
matplotlib

This module also requires the SOTA-py library (included in the external folder of this repository), specifically:

• stochastic_graph
• deterministic_algorithms
• preprocessing (Reach and Arcflags logic)