A quantitative study of the 2D Ising model with machine learning techniques
by
Davide Vadacchino(INFN Pisa)
→
Europe/Berlin
DESY
DESY
Description
Machine Learning techniques have gained a considerable attention from the scientific community in the last few years. In this work, we analysed a particularly simple and well understood algorithm, the Support vector Machine (SVM), designed for supervised classification tasks. We used a set of raw configurations of the 2D Ising model, obtained from Monte Carlo simulations, as a testbed to understand what quantitative and qualitative informations could be obtained on the system with a minimal set of assumptions. The order parameter can be obtained, and the critical temperature and critical indices computed with an accuracy comparable with that obtained with standard methods. Moreover, the precise knowledge of the learning algorithm suggests a way to understand the global symmetry of this and other systems from a set of raw configurations.
