Framework for testing random numbers in parallel calculations

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

Volumn 59, Issue 6, 1999, Pages 7200-7204

  Vattulainen, I ^a,b  

^a TECHNICAL UNIVERSITY OF DENMARK   (Denmark)

^b UNIVERSITY OF HELSINKI   (Finland)

Author keywords

[No Author keywords available]

Indexed keywords

ARTICLE;

EID: 0001359130     PISSN: 1063651X     EISSN: None     Source Type: Journal    
DOI: 10.1103/PhysRevE.59.7200     Document Type: Article

References (44)

1. K. Binder and D.W. Heermann, Monte Carlo Simulation in Statistical Physics (Springer-Verlag, Berlin, 1988)
2. M. P. Allen and D. J. Tildesley, Computer Simulation of Liquids (Oxford, New York, 1993)
3. D. Frenkel and B. Smit, Understanding Molecular Simulation: From Algorithms to Applications (Academic Press, San Diego, 1996).
4. P. L’Écuyer, Ann. Oper. Res. 53, 77 (1994).
5. P. Hellekalek, Math. Comput. Simul. 46, 485 (1998).
6. R.M. Ziff, Comput. Phys. 12, 385 (1998)
7. R.M. ZiffPhys. Rev. Lett. 69, 2670 (1992).
8. G. Marsaglia, in Computer Science and Statistics: The Interface, edited by L. Billard (Elsevier, Amsterdam, 1985), p. 3.
9. A. Compagner, Phys. Rev. E 52, 5634 (1995).
10. I. Vattulainen, T. Ala-Nissila, and K. Kankaala, Phys. Rev. Lett. 73, 2513 (1994)
11. I. VattulainenT. Ala-NissilaK. KankaalaPhys. Rev. E 52, 3205 (1995).
12. A.M. Ferrenberg, D.P. Landau, and Y.J. Wong, Phys. Rev. Lett. 69, 3382 (1992).
13. W. Selke, A.L. Talapov, and L.N. Shchur, Pis’ma Zh. Éksp. Teor. Fiz. 85, 1144 (1983) [JETP Lett. 58, 665 (1993)].
14. P.D. Coddington, Int. J. Mod. Phys. C 5, 547 (1994)
15. Int. J. Mod. Phys. CP.D. Coddington7, 295 (1996).
16. P. Grassberger, J. Phys. A 26, 2769 (1993)
17. P. GrassbergerPhys. Lett. A 181, 43 (1993).
18. R.M. D’Souza, Y. Bar-Yam, and M. Kardar, Phys. Rev. E 57, 5044 (1998).
19. F.J. Resende and B.V. Costa, Phys. Rev. E 58, 5183 (1998).
20. J. Stat. Phys. 30, 2 (1983), special issue, edited by G.W. Weiss and R.J. Rubin; E.W. Montroll and M.F. Shlesinger, in Nonequilibrium Phenomena II: From Stochastics to Hydrodynamics, edited by J.L. Lebowitz and E.W. Montroll (Elsevier, Amsterdam, 1984).
21. There are many ways to construct the sequences (Formula presented) for the processors one through m. We used random numbers (Formula presented) generated by a single pseudorandom number generator to make nonoverlapping sequences (Formula presented) (Formula presented) and so forth. Other possibilities for constructing the sequences are given in, e.g.
22. S. L. Anderson, SIAM Rev. 32, 221 (1990).
23. J. Krug, Adv. Phys. 46, 139 (1997).
24. B. Duplantier and K.-H. Kwon, Phys. Rev. Lett. 61, 2514 (1988).
25. K. Burdzy and G.F. Lawler, Ann. Prob. 18, 981 (1990)
26. G. Slade, Am. Sci. 84, 146 (1996).
27. Use of the intersection test to study more than two random walkers is also possible, although rigorous mathematical bounds for (Formula presented) in this case are not available (see Refs. 1718).
28. H. Larralde, P. Trunfio, S. Havlin, H.E. Stanley, and G.H. Weiss, Phys. Rev. A 45, 7128 (1992).
29. Equation (2) is just one rational approach to consider the convergence of an exponent which characterizes power-law behavior. Another possibility given in Ref. 17 was found to yield consistent results in the asymptotic limit.
30. T.G. Lewis and W.H. Payne, J. Assoc. Comput. Mach. 20, 456 (1973).
31. F. James, Comput. Phys. Commun. 60, 329 (1990).
32. G. Marsaglia, A. Zaman, and W.-W. Tsang, Stat. Prob. Lett. 9, 35 (1990).
33. G. Marsaglia and A. Zaman, Comput. Phys. 8, 117 (1994).
34. M. Lüscher, Comput. Phys. Commun. 79, 100 (1994)
35. Comput. Phys. Commun.F. James, 79, 111 (1994).
36. To determine the exponent, there are various ways which yield essentially identical results. Estimation of the error bars is a different matter. Use of linear regression is one possibility, but this approach provides an error estimate that is unreasonably small. To obtain a more realistic estimate, we determined the error bars (given in Table I) on the basis of fluctuations in the running exponents.
37. L.N. Shchur, J.R. Heringa, and H.W.J. Blöte, Physica A 241, 579 (1997).
38. I. Vattulainen, K. Kankaala, J. Saarinen, and T. Ala-Nissila, Comput. Phys. Commun. 86, 209 (1995).
39. L.N. Shchur and P. Butera, Int. J. Mod. Phys. C 9, 607 (1998).
40. D.E. Knuth, The Art of Computer Programming, Volume 2: Seminumerical Algorithms, 2nd ed. (Addison-Wesley, Reading, MA, 1981).
41. To determine (Formula presented), we calculated a given correlation function (Formula presented) of RANLUX4 ten times with (Formula presented) samples each up to (Formula presented). For each of these, we calculated (Formula presented) (Formula presented), with respect to (an independent calculation of) (Formula presented) of RANLUX4 with M samples. Then (Formula presented) is simply the mean of (Formula presented). The same procedure was carried out for all correlation functions (Formula presented) (Formula presented) and (Formula presented)
42. K.G. Hamilton, Comput. Phys. Commun. 81, 237 (1994).
43. N. Madras and G. Slade, The Self-Avoiding Walk (Birkhäuser, Boston, 1993).
44. For theoretical work on parallel random number generation, see K. Entacher, ACM Trans. Model. Comput. Simul. 8, 61 (1998), and references therein.