Utilizing Unsupervised Learning in the Domany-Kinzel Model for Phase Transitions

Investigating the Domany-Kinzel Model

The Domany-Kinzel (DK) model illustrates several types of nonequilibrium phase transitions based on specific parameters.

Application of Learning Methods

• Supervised learning and semi-supervised learning techniques are used to estimate critical points.
• Unsupervised methods help analyze unlabeled configurations of the DK model.

We explored the efficiency of two prominent unsupervised techniques:

1. Principal Component Analysis (PCA)
2. Autoencoder

Both methods have demonstrated strong correlation with simulated stationary particle number density.

Conclusion

This research highlights the potential of unsupervised learning techniques in effectively studying complex phase transitions, furthering the understanding of critical behavior in physical systems.