J. Phys. Soc. Jpn. 91, 044001 (2022) [9 Pages]

Randomized-Gauge Test for Machine Learning of Ising Model Order Parameter

+ Affiliations
1Department of Physics, The University of Tokyo, Bunkyo, Tokyo 113-0033, Japan2Institute for Physics of Intelligence, The University of Tokyo, Bunkyo, Tokyo 113-0033, Japan3Institute for Solid State Physics, The University of Tokyo, Kashiwa, Chiba 277-8581, Japan

Recently, machine learning has been applied successfully for identifying phases and phase transitions of the Ising models. The continuous phase transition is characterized by spontaneous symmetry breaking, which can not be detected in general from a single spin configuration. To investigate if neural networks can extract correlations among spin snapshots, we propose a new test using the random-gauge Ising model. We show that neural networks can extract the order parameter or the energy of the random-gauge model as in the ferromagnetic case. We also discuss how and where the information of random gauge is encoded in neural networks and attempt to reconstruct the gauge from the neural network parameters. We find that the fully connected network encodes the effect of random gauge to its weights naturally. In contrast, the convolutional network copes with the randomness by assigning different network parts to local gauge patterns. This observation indicates that although the latter demonstrates higher performance than the former for the present randomized-gauge test, the former is more effective and suitable for dealing with models with spatial randomness.

©2022 The Physical Society of Japan


  • 1 G. E. Hinton and R. R. Salakhutdinov, Science 313, 504 (2006). 10.1126/science.1127647 CrossrefGoogle Scholar
  • 2 Y. LeCun, Y. Bengio, and G. Hinton, Nature 521, 436 (2015). 10.1038/nature14539 CrossrefGoogle Scholar
  • 3 M. I. Jordan and T. M. Mitchell, Science 349, 255 (2015). 10.1126/science.aaa8415 CrossrefGoogle Scholar
  • 4 C. M. Bishop, Pattern Recognition and Machine Learning (Springer, New York, 2006). Google Scholar
  • 5 G. Hinton, L. Deng, D. Yu, G. E. Dahl, A.-r. Mohamed, N. Jaitly, A. Senior, V. Vanhoucke, P. Nguyen, T. N. Sainath, and B. Kingsbury, IEEE Signal Process. Mag. 29, 82 (2012). 10.1109/MSP.2012.2205597 CrossrefGoogle Scholar
  • 6 G. Carleo, I. Cirac, K. Cranmer, L. Daudet, M. Schuld, N. Tishby, L. Vogt-Maranto, and L. Zdeborová, Rev. Mod. Phys. 91, 045002 (2019). 10.1103/RevModPhys.91.045002 CrossrefGoogle Scholar
  • 7 P. Broecker, F. F. Assaad, and S. Trebst, arXiv:1707.00663. Google Scholar
  • 8 P. Ponte and R. G. Melko, Phys. Rev. B 96, 205146 (2017). 10.1103/PhysRevB.96.205146 CrossrefGoogle Scholar
  • 9 C. Wang and H. Zhai, Phys. Rev. B 96, 144432 (2017). 10.1103/PhysRevB.96.144432 CrossrefGoogle Scholar
  • 10 Y.-H. Liu and E. P. L. van Nieuwenburg, Phys. Rev. Lett. 120, 176401 (2018). 10.1103/PhysRevLett.120.176401 CrossrefGoogle Scholar
  • 11 P. Zhang, H. Shen, and H. Zhai, Phys. Rev. Lett. 120, 066401 (2018). 10.1103/PhysRevLett.120.066401 CrossrefGoogle Scholar
  • 12 M. J. S. Beach, A. Golubeva, and R. G. Melko, Phys. Rev. B 97, 045207 (2018). 10.1103/PhysRevB.97.045207 CrossrefGoogle Scholar
  • 13 E. van Nieuwenburg, Y.-H. Liu, and S. Huber, Nat. Phys. 13, 435 (2017). 10.1038/nphys4037 CrossrefGoogle Scholar
  • 14 S. J. Wetzel, Phys. Rev. E 96, 022140 (2017). 10.1103/PhysRevE.96.022140 CrossrefGoogle Scholar
  • 15 S. J. Wetzel and M. Scherzer, Phys. Rev. B 96, 184410 (2017). 10.1103/PhysRevB.96.184410 CrossrefGoogle Scholar
  • 16 J. Carrasquilla and R. G. Melko, Nat. Phys. 13, 431 (2017). 10.1038/nphys4035 CrossrefGoogle Scholar
  • 17 A. Morningstar and R. G. Melko, J. Mach. Learn. Res. 18, 1 (2018). Google Scholar
  • 18 P. Suchsland and S. Wessel, Phys. Rev. B 97, 174435 (2018). 10.1103/PhysRevB.97.174435 CrossrefGoogle Scholar
  • 19 S. Iso, S. Shiba, and S. Yokoo, Phys. Rev. E 97, 053304 (2018). 10.1103/PhysRevE.97.053304 CrossrefGoogle Scholar
  • 20 D. Kim and D.-H. Kim, Phys. Rev. E 98, 022138 (2018). 10.1103/PhysRevE.98.022138 CrossrefGoogle Scholar
  • 21 K. Kashiwa, Y. Kikuchi, and A. Tomiya, Prog. Theor. Exp. Phys. 2019, 083A04 (2019). 10.1093/ptep/ptz082 CrossrefGoogle Scholar
  • 22 H. A. Kramers and G. H. Wannier, Phys. Rev. 60, 252 (1941). 10.1103/PhysRev.60.252 CrossrefGoogle Scholar
  • 23 L. Onsager, Phys. Rev. 65, 117 (1944). 10.1103/PhysRev.65.117 CrossrefGoogle Scholar
  • 24 J. Villain, J. Phys. C 10, 1717 (1977). 10.1088/0022-3719/10/10/014 CrossrefGoogle Scholar
  • 25 R. H. Swendsen and J.-S. Wang, Phys. Rev. Lett. 58, 86 (1987). 10.1103/PhysRevLett.58.86 CrossrefGoogle Scholar
  • 26 B. Kastening, Phys. Rev. E 64, 066106 (2001). 10.1103/PhysRevE.64.066106 CrossrefGoogle Scholar
  • 27 Exact Solutions, https://github.com/todo-group/exact. Google Scholar
  • 28 I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning (MIT Press, Cambridge, MA, 2016). Google Scholar
  • 29 Keras, https://keras.io. Google Scholar
  • 30 TensorFlow, https://www.tensorflow.org. Google Scholar
  • 31 D. P. Kingma and J. Ba, 3rd International Conference on Learning Representations, ICLR 2015. Google Scholar
  • 32 J. S. Bridle, in Neurocomputing, ed. F. F. Soulié and J. Hérault (Springer, Berlin/Heidelberg, 1990) p. 227. 10.1007/978-3-642-76153-9_28 CrossrefGoogle Scholar
  • 33 X. Glorot, A. Bordes, and Y. Bengio, Proc. fourteenth Int. Conf. Artif. Intell. Stat., 2011, p. 315. Google Scholar