Abstract
A simple algorithm is described to sample permutations of identical particles in Path Integral Monte Carlo (PIMC) simulations of continuum many-body systems. The sampling strategy illustrated here is fairly general, and can be easily incorporated in any PIMC implementation based on the staging algorithm. Although it is similar in spirit to an existing prescription, it differs from it in some key aspects. It allows one to sample permutations efficiently, even if long paths (e.g., hundreds, or thousands of slices) are needed. We illustrate its effectiveness by presenting results of a PIMC calculation of thermodynamic properties of superfluid 4He, in which a very simple approximation for the high-temperature density matrix was utilized.
Similar content being viewed by others
References
D. M. Ceperley (1995) Rev. Mod. Phys. 67 275 Occurrence Handle10.1103/RevModPhys.67.279
D. M. Ceperley E. L. Pollock (1986) Phys. Rev. Lett. 56 351
See, for instance, D. M, Ceperley, in Monte Carlo and Molecular Dynamics of Condensed Matter Systems, K. Binder and G. Ciccotti, eds., Editrice Compositori Bologna, Italy, (1996).
Our permutation sampling algorithm is equally applicable to fermions, although it does not solve the sign problem.
S. A. Chin (1997) Phys, Lett. A 226 344
S. Jang S. Jang G. A. Voth (2001) J. Chem. Phys. 115 7832 Occurrence Handle10.1063/1.1410117
See, for instance, J. M. Thijssen, Computational Physics, Cambridge University Press Cambridge, UK (1999), pp. 278–280.
Note that r0 need not coincide with r L for any given particle, due to the presence of permutations. Therefore, it is necessary that path entanglement, arising from permutations, be properly taken into account. That means that, on proceeding past the Lth position, one must be sure to reconnect the path of the particle, back at the first slice, in a way that is consistent with the current permutation of particle labels.
E. L. Pollock D. M. Ceperley (1984) Phys. Rev. B 30 2555 Occurrence Handle10.1103/PhysRevB.30.2555
It all can be derived using the identity (a − b)2+(b − c)2=(a − c)2/2+2(b − \(\bar{b}\))2, where \(\bar{b}\)=(a+c)/2.
We emphasize that all distances in (17) are computed by taking the periodic image of each particle that is closest to particle ν. Thus, as a result of permutation sampling, paths cross over into adjacent simulation cells. This aspect is crucial in estimating the superfluid density, or imaginary time diffusion of individual particles. (for details; see Ref. 1).
R. A. Aziz V. P. S. Nain J. S. Carley W. L. Taylor G. T. McConville (1979) J. Chem. Phys. 70 4330 Occurrence Handle10.1063/1.438007
C. Chakravarty M. C. Gordillo D. M. Ceperley (1998) J. Chem. Phys. 109 2123 Occurrence Handle10.1063/1.476725
For “permute” type moves, we define the optimal value of s that for which the rate of acceptance of attempted permutations is greatest.
E. L. Pollock M. Ceperley D. (1987) Phys. Rev. B 36 8343 Occurrence Handle10.1103/PhysRevB.36.8343
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Boninsegni, M. Permutation Sampling in Path Integral Monte Carlo. J Low Temp Phys 141, 27–46 (2005). https://doi.org/10.1007/s10909-005-7513-0
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/s10909-005-7513-0