Metastable states in spin glasses and disordered ferromagnets

Physical Review E - Statistical Physics, Plasmas, Fluids, and Related Interdisciplinary Topics

Volumn 60, Issue 5, 1999, Pages 5244-5260

  Newman, C M ^a   Stein, D L ^b  

^a NEW YORK UNIVERSITY   (United States)

^b The University of Arizona   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

ARTICLE;

EID: 0013248235     PISSN: 1063651X     EISSN: None     Source Type: Journal    
DOI: 10.1103/PhysRevE.60.5244     Document Type: Article

References (79)

1. Recent Progress in Random Magnets, edited by D.H. Ryan, (World Scientific, Singapore, 1992).
2. K.H. Fischer and J.A. Hertz, Spin Glasses (Cambridge University Press, Cambridge, 1991).
3. K. Binder and A.P. Young, Rev. Mod. Phys. 58, 801 (1986).
4. P. Refrigier, E. Vincent, J. Hamman, and M. Ocio, J. Phys. (Paris) 48, 1533 (1987).
5. P. Sibani and K.-H. Hoffmann, Phys. Rev. Lett. 63, 2853 (1989).
6. K.-H. Hoffmann and P. Sibani, Z. Phys. B 80, 429 (1990).
7. M. Lederman, R. Orbach, J.M. Hamann, M. Ocio, and E. Vincent, Phys. Rev. B 44, 7403 (1991).
8. P. Svedlinh, K. Gunnarson, J.-O. Andersson, H.A. Katori, and A. Ito, Phys. Rev. B 46, 13 687 (1992).
9. E. Vincent, J. Hamann, and M. Ocio, in Recent Progress in Random Magnets (Ref. 1), pp. 207–236.
10. J.P. Bouchaud, J. Phys. I 2, 1705 (1992).
11. F. Lefloch, J. Hamann, M. Ocio, and E. Vincent, Europhys. Lett. 18, 647 (1992).
12. H. Rieger, J. Phys. A 26, L615 (1993).
13. S. Franz and M. Mézard, Physica A 210, 48 (1994).
14. E. Vincent, J.P. Bouchaud, D.S. Dean, and J. Hamann, Phys. Rev. B 52, 1050 (1995).
15. Y.G. Joh, R. Orbach, and J.M. Hamann, Phys. Rev. Lett. 77, 4648 (1996).
16. Y.G. Joh, R. Orbach, and J.M. Hamann, Philos. Mag. B 77, 231 (1998).
17. R.G. Palmer, Adv. Phys. 31, 669 (1982).
18. M.D. Ediger, C.A. Angell, and S.R. Nagel, J. Phys. Chem. 100, 13200 (1996).
19. For recent papers that study metastability as a unifying theme in glasses, spin glasses, neural network models, and other disordered systems, see G. Parisi, in 25 Years of Non-Equilibrium Statistical Mechanics, edited by J.J. Brey, J. Marro, J.M Rubi, and M. San Miguel (Springer-Verlag, Berlin, 1995), pp. 135–142;K. Binder, ibid., pp. 143–160;D. Sherrington and A.C.C. Coolen, ibid., pp. 161–176.
20. J.J. Hopfield, Proc. Natl. Acad. Sci. USA 79, 2554 (1982).
21. D.J. Amit, H. Gutfreund, and H. Sompolinsky, Phys. Rev. Lett. 55, 1539 (1982).
22. D.J. Amit, H. Gutfreund, and H. Sompolinsky, Phys. Rev. A 32, 1007 (1982).
23. C. Dasgupta, S.-K. Ma, and C.-K. Hu, Phys. Rev. B 20, 3837 (1979).
24. See, for example, K. Binder and W. Kinzel, in Heidelberg Colloquium on Spin Glasses, edited by J.L. van Hemmen and I. Morgenstern (Springer-Verlag, Berlin, 1983), pp. 279–304;I. Morgenstern, ibid., pp. 305–327;A.P. Young, ibid., pp. 328–345.
25. V. Čerńy, J. Optim. Theory Appl. 45, 41 (1985).
26. S. Kirkpatrick, C.D. Gelatt, and M.P. Vecchi, Science 220, 671 (1983).
27. M.B. Weissman, Rev. Mod. Phys. 65, 829 (1993).
28. F. Tanaka and S.F. Edwards, J. Phys. F 10, L2769 (1980).
29. A.J. Bray and M.A. Moore, J. Phys. C 13, L469 (1980). By studying the number of solutions to the TAP equations 32, the authors were also able to draw some conclusions about numbers of metastable states for (Formula presented)
30. C. De Dominicis, M. Gabay, T. Garel, and H. Orland, J. Phys. (Paris) 41, 923 (1980). This paper also contains some results on numbers of metastable states near (Formula presented)
31. D. Sherrington and S. Kirkpatrick, Phys. Rev. Lett. 35, 1792 (1975).
32. D.J. Thouless, P.W. Anderson, and R.G. Palmer, Philos. Mag. 35, 593 (1977).
33. K. Nemoto, J. Phys. A 21, L287 (1988).
34. D. Vertechi and M.A. Virasoro, J. Phys. (France) 50, 2325 (1989).
35. L.B. Ioffe and D. Sherrington, Phys. Rev. B 57, 7666 (1998).
36. V.M. de Oliveira and J.F. Fontanari, J. Phys. A 30, 8445 (1997).
37. S. Franz and G. Parisi, J. Phys. I 5, 1401 (1995).
38. E.A. Lutchinskaia and E.E. Tareyeva, Europhys. Lett. 17, 109 (1992).
39. S.A. Kauffman, The Origins of Order (Oxford University Press, Oxford, 1993).
40. T. Li, Phys. Rev. B 24, 6579 (1981).
41. R. Ettalie and M.A. Moore, J. Phys. (France) Lett. 46, L-893 (1985).
42. B. Derrida and E. Gardner, J. Phys. (Paris) 47, 959 (1986).
43. A.J. Bray and M.A. Moore, J. Phys. C 14, 1313 (1981).
44. E.S. Rodrigues and P.M.C. de Oliveira, J. Stat. Phys. 74, 1265 (1994).
45. As in previous treatments (see, however, Refs. 29 and 30), our results are confined to energetically stable states. In dynamical treatments, one finds that states that are locally stable with respect to small free energy changes are often invoked. It is not clear how to define such objects, or their companions such as free energy barriers, in a clear and unambiguous way. We do not consider this to be a serious impediment for two reasons: first, at least at low temperatures, the properties of such states are likely to be closely related to the energetically stable states discussed here; and second, we are able to consider the properties of M-spin-flip energetically stable states for arbitrarily large (finite) M. We believe that it is reasonable to expect that these states will play an important dynamical role at positive temperature, presumably with larger M becoming relevant for higher temperatures.
46. W.L. McMillan, J. Phys. C 17, 3179 (1984).
47. D.S. Fisher and D.A. Huse, Phys. Rev. Lett. 56, 1601 (1986).
48. A.J. Bray and M.A. Moore, Phys. Rev. Lett. 58, 57 (1987).
49. D.S. Fisher and D.A. Huse, Phys. Rev. B 38, 386 (1988).
50. G.J.M. Koper and H.J. Hilhorst, J. Phys. (Paris) 49, 249 (1988).
51. C.M. Newman and D.L. Stein, J. Stat. Phys. 94, 709 (1999).
52. S.F. Edwards and P.W. Anderson, J. Phys. F 5, 965 (1975).
53. Almost every (whether it’s referring to (Formula presented) or some function of them like the metastable states that are the prime subject of this paper) means a set of probability 1, where the probability measure (i.e., ensemble) in question should be clear from the context. What is left out of a set of probability 1 is insignificant in the sense that it does not contribute to any quantitites (such as overlap distributions) defined in terms of an average over the relevant measure. The distinction between almost every and every is significant and forgetting it can easily lead to confusion and seeming contradiction. So, for example, most (but not all) of the results in this paper about metastable states are only valid for almost every one and these are precisely the ones that are relevant for deep quench dynamics.
54. C.M. Newman and D.L. Stein, Phys. Rev. Lett. 82, 3944 (1999).
55. S. Nanda, C. M. Newman, and D. L. Stein, in On Dobrushin’s Way (from Probability Theory to Statistical Physics), edited by R. Minlos, S. Shlosman, and Y. Suhov (American Mathematics Society, Providence, in press).
56. J.F. Fernandez and R. Medina, Phys. Rev. B 19, 3561 (1979).
57. A. Gandolfi, C. M. Newman, and D. L. Stein (unpublished).
58. D. Stauffer and A. Aharony, Introduction to Percolation Theory, 2nd Ed. (Taylor and Francis, London, 1992).
59. T. M. Liggett, Interacting Particle Systems (Springer-Verlag, New York, 1985).
60. S. Nanda and C.M. Newman, Random Struct. Algorithms (to be published).
61. C.M. Newman and D.L. Stein, Phys. Rev. Lett. 72, 2286 (1994).
62. J.R. Banavar, M. Cieplak, and A. Maritan, Phys. Rev. Lett. 72, 2320 (1994).
63. C.M. Newman and D.L. Stein, J. Stat. Phys. 82, 1113 (1996).
64. D. Stauffer, J. Phys. A 27, 5029 (1994).
65. For recent numerical work on one such model, a 2D bond-diluted ferromagnet, see S. Jain, Phys. Rev. E 59, 2493R (1999).
66. The proof shows more than uncountability of all the M-spin-flip-stable states: the probability measure on the (Formula presented)’s (inherited from its dependence on (Formula presented) and (Formula presented) and their probability measures) is continuous—i.e., no single spin configuration has strictly positive weight.
67. C.M. Newman and D.L. Stein, Phys. Rev. E 55, 5194 (1997).
68. M. Mézard, G. Parisi, N. Sourlas, G. Toulouse, and M. Virasoro, Phys. Rev. Lett. 52, 1156 (1984).
69. M.R. Evans, Phys. Rev. B 48, 3515 (1993).
70. For example, one might consider the following situation. Allow single-spin-flip dynamics to run until convergence to a specific (Formula presented) Using this as the starting state, raise the temperature to a small nonzero value T and allow positive-temperature Glauber dynamics to run for a specified time (Formula presented) after which the temperature is reduced back to zero. It is not unreasonable to speculate that the Hamming distance of the final state (Formula presented) reached in this way from the initial (Formula presented) increases with (Formula presented) for (Formula presented) small. Such considerations may be relevant to some aging experiments 4, but our analysis makes clear that any possible correlations of this kind cannot be extended to provide information on the properties of the larger set of metastable states. This example illustrates that one needs to be careful to specify what is meant, for example, by correlations between barriers and overlaps; the meanings may be determined by the experimental situation under investigation.
71. A.C.D. van Enter and J.L. van Hemmen, Phys. Rev. A 29, 355 (1984).
72. F. Bagnoli, J. Stat. Phys. 85, 151 (1996).
73. P. Grassberger, J. Stat. Phys. 79, 13 (1995).
74. N. Jan and T.S. Ray, J. Stat. Phys. 75, 1197 (1994).
75. C.M. Newman and D.L. Stein, Phys. Rev. Lett. 76, 4821 (1996).
76. C.M. Newman and D.L. Stein, Phys. Rev. E 57, 1356 (1998).
77. The energy density results are for strongly disordered coupling distributions with finite mean.
78. A. A. Middleton, Phys. Rev. Lett. 83, 1672 (1999).
79. M. Palassini and A. P. Young (unpublished).