Efficient stochastic thermostatting of path integral molecular dynamics

Journal of Chemical Physics

Volumn 133, Issue 12, 2010, Pages

  Ceriotti, Michele ^a   Parrinello, Michele ^a   Markland, Thomas E ^b   Manolopoulos, David E ^b  

^a ETH ZURICH   (Switzerland)

^b UNIVERSITY OF OXFORD   (United Kingdom)

Author keywords

[No Author keywords available]

Indexed keywords

COLORED NOISE; CONDENSED PHASE; EFFICIENT IMPLEMENTATION; FREE PATH; GENERALIZED LANGEVIN EQUATION; HIGH FREQUENCY; LANGEVIN EQUATION; LIQUID WATER; NORMAL MODES; NORMAL-MODE FREQUENCIES; OPTIMIZED SAMPLING; PATH-INTEGRAL MOLECULAR DYNAMICS; QUANTUM MECHANICAL; SAMPLING EFFICIENCY; SIMPLE STOCHASTIC;

DIFFERENTIAL EQUATIONS; MECHANICAL PROPERTIES; MOLECULAR DYNAMICS; PALLADIUM; PILES; QUANTUM THEORY; STOCHASTIC SYSTEMS; THERMODYNAMIC PROPERTIES; VIBRATIONS (MECHANICAL);

THERMOSTATS;

ARTICLE; CHEMICAL MODEL; MOLECULAR DYNAMICS; QUANTUM THEORY; STATISTICS; TEMPERATURE; THERMODYNAMICS; VIBRATION;

MODELS, CHEMICAL; MOLECULAR DYNAMICS SIMULATION; QUANTUM THEORY; STOCHASTIC PROCESSES; TEMPERATURE; THERMODYNAMICS; VIBRATION;

EID: 77957738954     PISSN: 00219606     EISSN: None     Source Type: Journal    
DOI: 10.1063/1.3489925     Document Type: Article

References (43)

1. R. P. Feynman and A. R. Hibbs, Quantum Mechanics and Path Integrals (McGraw-Hill, New York, 1965).
2. R. P. Feynman, Statistical Mechanics (Addison-Wesley, New York, 1972).
3. D. Chandler and P. G. Wolynes, J. Chem. Phys. JCPSA6 0021-9606 74, 4078 (1981). 10.1063/1.441588
4. J. A. Barker, J. Chem. Phys. JCPSA6 0021-9606 70, 2914 (1979). 10.1063/1.437829
5. M. Parrinello and A. Rahman, in Monte Carlo Methods in Quantum Problems, edited by, M. H. Kalos, (Reidel, Dordrecht, 1984), p. 105.
6. M. Parrinello and A. Rahman, J. Chem. Phys. JCPSA6 0021-9606 80, 860 (1984). 10.1063/1.446740
7. R. W. Hall and B. J. Berne, J. Chem. Phys. JCPSA6 0021-9606 81, 3641 (1984). 10.1063/1.448112
8. H. C. Andersen, J. Chem. Phys. JCPSA6 0021-9606 72, 2384 (1980). 10.1063/1.439486
9. B. J. Berne and D. Thirulamai, Annu. Rev. Phys. Chem. ARPLAP 0066-426X 37, 401 (1986). 10.1146/annurev.pc.37.100186.002153
10. S. Noś, J. Chem. Phys. JCPSA6 0021-9606 81, 511 (1984). 10.1063/1.447334
11. W. G. Hoover, Phys. Rev. A PLRAAN 1050-2947 31, 1695 (1985). 10.1103/PhysRevA.31.1695
12. G. J. Martyna, M. L. Klein, and M. E. Tuckerman, J. Chem. Phys. JCPSA6 0021-9606 97, 2635 (1992). 10.1063/1.463940
13. M. E. Tuckerman, B. J. Berne, G. J. Martyna, and M. L. Klein, J. Chem. Phys. JCPSA6 0021-9606 99, 2796 (1993). 10.1063/1.465188
14. G. Bussi and M. Parrinello, Phys. Rev. E PLEEE8 1063-651X 75, 056707 (2007). 10.1103/PhysRevE.75.056707 (Pubitemid 47140176)
15. G. Bussi, D. Donadio, and M. Parrinello, J. Chem. Phys. JCPSA6 0021-9606 126, 014101 (2007). 10.1063/1.2408420 (Pubitemid 46068390)
16. M. Ceriotti, G. Bussi, and M. Parrinello, Phys. Rev. Lett. PRLTAO 0031-9007 102, 020601 (2009). 10.1103/PhysRevLett.102.020601
17. M. Ceriotti, G. Bussi, and M. Parrinello, J. Chem. Theory Comput. JCTCCE 1549-9618 6, 1170 (2010). 10.1021/ct900563s
18. H. Trotter, Proc. Am. Math. Soc. PAMYAR 0002-9939 10, 545 (1959).
19. L. S. Schulman, Techniques and Applications of Path Integration (Dover, New York, 2005).
20. M. F. Herman, E. J. Bruskin, and B. J. Berne, J. Chem. Phys. JCPSA6 0021-9606 76, 5150 (1982). 10.1063/1.442815
21. D. M. Ceperley, Rev. Mod. Phys. RMPHAT 0034-6861 67, 279 (1995). 10.1103/RevModPhys.67.279
22. I. R. Craig and D. E. Manolopoulos, J. Chem. Phys. JCPSA6 0021-9606 121, 3368 (2004). 10.1063/1.1777575
23. B. J. Braams and D. E. Manolopoulos, J. Chem. Phys. JCPSA6 0021-9606 125, 124105 (2006). 10.1063/1.2357599 (Pubitemid 44484188)
24. J. M. Sanz-Serna and M. P. Calvo, Numerical Hamiltonian Problems (Chapman and Hall, London, 1994).
25. B. Leimkuhler and S. Reich, Simulating Hamiltonian Dynamics (Cambridge University Press, Cambridge, 2004).
26. L. Verlet, Phys. Rev. PHRVAO 0031-899X 159, 98 (1967). 10.1103/PhysRev.159.98
27. The number of Fourier transforms per time step can be reduced from four to three per degree of freedom in the absence of a thermostat by keeping the momenta in the normal mode representation and transforming the forces into this representation for the updates in Eqs.; this also eliminates the need for the Fourier transforms in the PILE thermostat in Eqs..
28. A. Ricci and G. Ciccotti, Mol. Phys. MOPHAM 0026-8976 101, 1927 (2003). 10.1080/0026897031000108113
29. R. Zwanzig, Nonequilibrium Statistical Mechanics (Oxford University Press, Oxford, 2001).
30. C. W. Gardiner, Handbook of Stochastic Methods, 3rd ed. (Springer, Berlin, 2003).
31. M. Ceriotti, G. Bussi, and M. Parrinello, Phys. Rev. Lett. PRLTAO 0031-9007 103, 030603 (2009). 10.1103/PhysRevLett.103.030603
32. G. J. Martyna, M. E. Tuckerman, D. J. Tobias, and M. L. Klein, Mol. Phys. MOPHAM 0026-8976 87, 1117 (1996).
33. S. Jang and G. A. Voth, J. Chem. Phys. JCPSA6 0021-9606 107, 9514 (1997). 10.1063/1.475247
34. G. Bussi and M. Parrinello, Comput. Phys. Commun. CPHCBZ 0010-4655 179, 26 (2008). 10.1016/j.cpc.2008.01.006
35. R. A. Kuharski and P. J. Rossky, J. Chem. Phys. JCPSA6 0021-9606 82, 5164 (1985). 10.1063/1.448641
36. A. Wallqvist and B. J. Berne, Chem. Phys. Lett. CHPLBC 0009-2614 117, 214 (1985). 10.1016/0009-2614(85)80206-2
37. S. Habershon, T. E. Markland, and D. E. Manolopoulos, J. Chem. Phys. JCPSA6 0021-9606 131, 024501 (2009). 10.1063/1.3167790
38. T. E. Markland and D. E. Manolopoulos, J. Chem. Phys. JCPSA6 0021-9606 129, 024105 (2008). 10.1063/1.2953308
39. T. E. Markland and D. E. Manolopoulos, Chem. Phys. Lett. CHPLBC 0009-2614 464, 256 (2008). 10.1016/j.cplett.2008.09.019
40. S. L. Adler, Phys. Rev. D PRVDAQ 0556-2821 23, 2901 (1981). 10.1103/PhysRevD.23.2901
41. X. W. Zhou, J. A. Zimmerman, B. M. Wong, and J. J. Hoyt, J. Mater. Res. JMREEE 0884-2914 23, 704 (2008). 10.1557/jmr.2008.0090 (Pubitemid 351454262)
42. 0 do indeed give nearly optimal sampling for the GLE in every case.
43. In our simulations of liquid water the use of the PILE thermostat was found to introduce an overhead of 5% over an NVE simulation. The overhead for the GLE thermostat (which requires more random numbers to be generated and matrix-vector multiplications to be performed) was about 30%. Numerical integration of the NHC equations requires the computation of several exponentials for each degree of freedom and chain variable. This effort is then further magnified by the necessity of a multiple time step integration to avoid any drift in the conserved quantity, which we found to result in a 250% overall overhead in our simulations. We should however stress that such a large thermostatting overhead will only be obtained in cases where the physical forces acting on each bead of the ring polymer can be evaluated extremely cheaply, as they can for an empirical water model using ring polymer contraction techniques. In ab initio PIMD simulations the NHC overhead will typically be a tiny fraction of the overall cost of the calculation.