- Physical Review Journals
All JournalsPhysics Magazine

Physical Review D

Reuse & Permissions

It is not necessary to obtain permission to reuse this article or its components as it is available under the terms of the Creative Commons Attribution 4.0 International license. This license permits unrestricted use, distribution, and reproduction in any medium, provided attribution to the author(s) and the published article's title, journal citation, and DOI are maintained. Please note that some figures may have been included with permission from other third parties. It is your responsibility to obtain the proper permission from the rights holder directly for these figures.

Export citation

Export citation

Choose format for download:

Download Citation
  • Letter
  • Open Access

Reconstructing spectral functions via automatic differentiation

Lingxiao Wang1, Shuzhe Shi2,3,*, and Kai Zhou1,†

  • 1Frankfurt Institute for Advanced Studies, Ruth Moufang Strasse 1, D-60438, Frankfurt am Main, Germany
  • 2Department of Physics, McGill University, Montreal, Quebec H3A 2T8, Canada
  • 3Center for Nuclear Theory, Department of Physics and Astronomy, Stony Brook University, Stony Brook, New York 11784, USA

  • *shuzhe.shi@stonybrook.edu
  • †zhou@fias.uni-frankfurt.de
PDF

Phys. Rev. D 106, L051502 – Published 16 September, 2022

DOI: https://doi.org/10.1103/PhysRevD.106.L051502

Abstract

Reconstructing spectral functions from Euclidean Green’s functions is an important inverse problem in many-body physics. However, the inversion is proved to be ill-posed in the realistic systems with noisy Green’s functions. In this paper we propose an automatic differentiation (AD) framework as a generic tool for the spectral reconstruction from propagator observable. Exploiting the neural networks’ regularization as a nonlocal smoothness regulator of the spectral function, we represent spectral functions by neural networks and use the propagator’s reconstruction error to optimize the network parameters unsupervisedly. In the training process, except for the positive-definite form for the spectral function, there are no other explicit physical priors embedded into the neural networks. The reconstruction performance is assessed through relative entropy and mean square error for two different network representations. Compared to the maximum entropy method, the AD framework achieves better performance in the large-noise situation. It is noted that the freedom of introducing nonlocal regularization is an inherent advantage of the present framework and may lead to substantial improvements in solving inverse problems.

View figure in article

Physics Subject Headings (PhySH)

Article Text

Supplemental Material

References (54)

  1. M. Jarrell and J. E. Gubernatis, Phys. Rep. 269, 133 (1996).
  2. S. I. Kabanikhin, Inverse and Ill-Posed Problems: Theory and Applications (De Gruyter, Gottingen, Germany, 2011).
  3. M. Asakawa, T. Hatsuda, and Y. Nakahara, Prog. Part. Nucl. Phys. 46, 459 (2001).
  4. A. Rothkopf, Proc. Sci. Confinement2018 (2018) 026 [arXiv:1903.02293].
  5. P. J. Caudrey, Physica (Amsterdam) 6D, 51 (1982).
  6. A. N. Tikhonov, A. V. Goncharsky, V. V. Stepanov, and A. G. Yagola, Numerical Methods for the Solution of Ill-Posed Problems (Springer Netherlands, Dordrecht, 1995).
  7. J. G. McWhirter and E. R. Pike, J. Phys. A 11, 1729 (1978).
  8. S. Shi, L. Wang, and K. Zhou, arXiv:2201.02564.
  9. M. Bertero, in Advances in Electronics and Electron Physics, edited by P. W. Hawkes (Academic Press, New York, 1989), Vol. 75, pp. 1–120.
  10. Y. Burnier and A. Rothkopf, Phys. Rev. Lett. 111, 182003 (2013).
  11. Y. Burnier, O. Kaczmarek, and A. Rothkopf, Phys. Rev. Lett. 114, 082001 (2015).
  12. R. K. Bryan, Eur. Biophys. J. 18, 165 (1990).
  13. M. Asakawa, arXiv:2001.10205.
  14. A. Rothkopf, Data 5, 85 (2020).
  15. L. Kades, J. M. Pawlowski, A. Rothkopf, M. Scherzer, J. M. Urban, S. J. Wetzel, N. Wink, and F. P. G. Ziegler, Phys. Rev. D 102, 096001 (2020).
  16. H. Yoon, J.-H. Sim, and M. J. Han, Phys. Rev. B 98, 245101 (2018).
  17. R. Fournier, L. Wang, O. V. Yazyev, and Q. Wu, Phys. Rev. Lett. 124, 056401 (2020).
  18. H. Li, J. Schwab, S. Antholzer, and M. Haltmeier, Inverse Probl. 36, 065005 (2020).
  19. S. Y. Chen, H. T. Ding, F. Y. Liu, G. Papp, and C. B. Yang, arXiv:2110.13521.
  20. M. Zhou, F. Gao, J. Chao, Y.-X. Liu, and H. Song, Phys. Rev. D 104, 076011 (2021).
  21. J. Horak, J. M. Pawlowski, J. Rodríguez-Quintero, J. Turnwald, J. M. Urban, N. Wink, and S. Zafeiropoulos, Phys. Rev. D 105, 036014 (2022).

  22. The universal approximation theorem ensures that ANNs can approximate any kind of continuous function with nonlinear activation functions [23, 24].

  23. I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning (MIT Press, Cambridge, MA, 2016).
  24. L. Wu, Z. Zhu, and W. E, arXiv:1706.10239.
  25. S. Forte, L. s. Garrido, J. I. Latorre, and A. Piccione, J. High Energy Phys. 05 (2002) 062.
  26. T. N. Collaboration, L. D. Debbio, S. Forte, J. I. Latorre, A. Piccione, and J. Rojo, J. High Energy Phys. 03 (2007) 039.
  27. J. Carrasquilla and R. G. Melko, Nat. Phys. 13, 431 (2017).
  28. L.-G. Pang, K. Zhou, N. Su, H. Petersen, H. Stöcker, and X.-N. Wang, Nat. Commun. 9, 210 (2018).
  29. L. Wang, Y. Jiang, L. He, and K. Zhou, arXiv:2005.04857.
  30. R. Wang, Y.-G. Ma, R. Wada, L.-W. Chen, W.-B. He, H.-L. Liu, and K.-J. Sun, Phys. Rev. Research 2, 043202 (2020).
  31. L. Jiang, L. Wang, and K. Zhou, Phys. Rev. D 103, 116023 (2021).
  32. K. Zhou, G. Endro˝di, L.-G. Pang, and H. Stöcker, Phys. Rev. D 100, 011501 (2019).
  33. D. Boyda, G. Kanwar, S. Racanière, D. J. Rezende, M. S. Albergo, K. Cranmer, D. C. Hackett, and P. E. Shanahan, Phys. Rev. D 103, 074504 (2021).
  34. G. Kanwar, M. S. Albergo, D. Boyda, K. Cranmer, D. C. Hackett, S. Racanière, D. J. Rezende, and P. E. Shanahan, Phys. Rev. Lett. 125, 121601 (2020).
  35. M. S. Albergo, G. Kanwar, and P. E. Shanahan, Phys. Rev. D 100, 034515 (2019).
  36. M. Omana Kuttan, J. Steinheimer, K. Zhou, A. Redelbach, and H. Stoecker, Phys. Lett. B 811, 135872 (2020).
  37. P. Thaprasop, K. Zhou, J. Steinheimer, and C. Herold, Phys. Scr. 96, 064003 (2021).
  38. F. Li, Y. Wang, H. Lü, P. Li, Q. Li, and F. Liu, J. Phys. G 47, 115104 (2020).
  39. A. Andreassen, S.-C. Hsu, B. Nachman, N. Suaysom, and A. Suresh, Phys. Rev. D 103, 036001 (2021).
  40. M. O. Kuttan, K. Zhou, J. Steinheimer, A. Redelbach, and H. Stoecker, J. High Energy Phys. 10 (2021) 184.
  41. H. Huang, B. Xiao, Z. Liu, Z. Wu, Y. Mu, and H. Song, Phys. Rev. Research 3, 023256 (2021).
  42. M. E. Peskin and D. V. Schroeder, An Introduction to Quantum Field Theory, 1st ed. (Westview Press, Reading, MA, 1995).
  43. See Supplemental Material at http://link.aps.org/supplemental/10.1103/PhysRevD.106.L051502 for machine learning terminology about automatic differentiation and neural networks. Then the details of training set-ups are listed. We also explain how to prepare the mock lattice QCD correlators for doing the tests shown in the section of extensions. Detailed comparisons with the classical MEM method are added in the final part.

  44. It can be conveniently implemented in many deep learning frameworks. In our case, main computations are deployed in Pytorch and released on Github, but also validated in Tensorflow.

  45. It is a stochastic gradient-based algorithm that is based on adaptive estimations of first-order and second-order moments [46].

  46. D. P. Kingma and J. Ba, in Proceeding of the 3rd International Conference for Learning Representations, San Diego, 2015 [arXiv:1412.6980].
  47. M. Rosca, T. Weber, A. Gretton, and S. Mohamed, in Proceedings on “I Can’t Believe It’s Not Better!” At NeurIPS Workshops, Proceedings of the Machine Learning Research Vol. 137, edited by J. Zosa Forde, F. Ruiz, M. F. Pradier, and A. Schein (PMLR, 2020), pp. 21–32.
  48. D. P. Kingma and J. Ba, arXiv:1412.6980.
  49. R.-A. Tripolt, P. Gubler, M. Ulybyshev, and L. Von Smekal, Comput. Phys. Commun. 237, 129 (2019).
  50. H.-T. Ding, F. Karsch, and S. Mukherjee, Int. J. Mod. Phys. E 24, 1530007 (2015).
  51. C. Ratti, Rep. Prog. Phys. 81, 084301 (2018).
  52. M. Hobson and A. Lasenby, Mon. Not. R. Astron. Soc. 298, 905 (1998).
  53. A. Rothkopf, Phys. Rev. D 95, 056016 (2017).
  54. D. Dudal, O. Oliveira, M. Roelfs, and P. Silva, Nucl. Phys. B952, 114912 (2020).

Outline

Information

Phys. Rev. D 106, L051502– Published 16 September, 2022
  • Received 17 January 2022
  • Revised 29 July 2022
  • Accepted 1 September 2022
Crossmark - Check for updates button
Creative Commons Logo

Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI. Funded by SCOAP3.

Published by the American Physical Society

Sign In to Your Journals Account

Filter

Filter

Article Lookup

Enter a citation