Machine Learning Phase Transitions

Analyse phase transitions with neural networks

This is an exam project for a master course on ''Statistical Methods for Learning''.

The repository includes C++ simulations of physical systems, Python Keras/Tensorflow code for the neural networks part and the LaTeX code for the exam report (in Italian, but with nice pictures!).

Simulated physical systems:

• Ising on square lattice, with 4 nearest neighbours (Wolff algorithm)
• Ising on honeycomb lattice, with 3 nearest neighbours (Wolff algorithm)
• Ising on triangular lattice, with 6 nearest neighbours (Metropolis algorithm)
• XY model on square lattice, with 4 nearest neighbours (Wolff algorithm)

Sources and References

Keras and Tensorflow

Carrasquilla J, Melko RG. 2017. ''Machine learning phases of matter''. Nature Physics.

Beach MJS, Golubeva A, Melko RG. 2018. ''Machine learning vortices at the Kosterlitz-Thouless transition''. Physical Review B.

Suchsland P, Wessel S. 2018. ''Parameter diagnostics of phases and phase transition learning by neural networks''. Physical Review B.

Authors

Martina Crippa [email protected]

Pietro F. Fontana [email protected]

License

The code is released under MIT license, see LICENSE file for further information.