Efficient calculation of energy spectra using path integrals☆
Introduction
Feynman's path integrals [5], [6] provide the general mathematical framework for dealing with quantum and statistical systems. The formalism has been successfully applied in generalizing the quantization procedure from the archetypical quantum mechanical problem of the dynamics of a single particle moving in one dimension, to more particles, more dimensions, as well as to more complicated objects such as fields, strings [7], etc. Symmetries of physical systems can be more easily treated and applied in this formalism, since it gives a simple and natural setup for their use [8]. Various approximation techniques are more easily derived within the framework of this formalism, and it has been successfully used for deriving non-perturbative results. The parallel application of this formalism in both high energy and condensed matter physics makes it an important general tool [9], [10]. The analytical and numerical approaches to path integrals have by now become central to the development of many other areas of physics, chemistry and materials science, as well as to the mathematics and finance [11], [12], [13], [14]. In particular, general numerical approaches such as the path integral Monte Carlo method have made possible the treatment of a wealth of non-trivial and previously inaccessible models.
The key impediment to the development of the path integral formalism is a lack of complete understanding of the general mathematical properties of these objects. In numerical approaches limited analytical input generally translates into lower efficiency of employed algorithms. The best path generating algorithms, for example, are efficient precisely because they have built into them the kinematic consequences of the stochastic self-similarity of paths [15]. A recent series of papers [1], [2], [3] has for this reason focused on the dynamical implications of stochastic self-similarity by studying the relation between path integral discretizations of different coarseness. This has resulted in a systematic analytical construction of a hierarchy of N-fold discretized effective actions labeled by a whole number p and built up from the naively discretized action in the mid-point ordering prescription (corresponding to ). The level p effective actions lead to discretized transition amplitudes and expectation values differing from the continuum limit by a term of order .
In this Letter we extend the applicability of the above method for improving the efficiency of path integral calculations to the evaluation of energy spectra. We show how the increased convergence of path integrals translates into the speedup in the numerical calculation of energy levels. Throughout the Letter we present and comment on the Monte Carlo simulations conducted using the hierarchy of effective actions for the case of several different models including anharmonic oscillator, Pöschl–Teller potential, and Morse potential. All the numerical simulations presented were done using Grid-adapted Monte Carlo code and were run on EGEE-II and SEE-GRID-2 infrastructure [16], [17]. The effective actions and the codes used can be found on our web site [18].
Section snippets
Partition function and energy spectra
The partition function is the central object in statistical mechanics. The path integral formalism gives us an elegant framework for calculating partition functions which can be used either for deriving analytical approximation techniques or for carrying out numerical evaluation. The starting point is the expression for the partition function in the coordinate basis, where is the quantum mechanical transition amplitude for going from a to b in
Numerical results
As we have seen in the previous section, can be evaluated with arbitrary precision on any interval of inverse temperatures for any given potential by appropriately increasing and adjusting N, p, and . Let us now numerically compare the quality of different discretizations of the free energy with , the most accurate one that may be calculated on a given set . To do this we use the standard function, where M
References (18)
- et al.
Phys. Lett. A
(2005) - et al.
Phys. Lett. A
(2006) - et al.
Phys. Lett. A
(2005) - et al.
Phys. Rev. Lett.
(2005) - et al.
Phys. Rev. B
(2005) - et al.
Quantum Mechanics and Path Integrals
(1965) Statistical Mechanics
(1972)- et al.
Quantum Field Theory
(1980) Aspects of Symmetry
(1985)
Cited by (10)
Ultra-fast converging path-integral approach for rotating ideal Bose-Einstein condensates
2010, Physics Letters, Section A: General, Atomic and Solid State PhysicsCitation Excerpt :In principle, this is done by studying the high-β regime, where the short-time expansion (14) is not valid. Although this procedure works also for lower values of β [19], it requires the numerical calculation of the one-particle partition function and a detailed study of its dependence on the inverse temperature in order to obtain the ground-state energy with sufficient precision. For this reason, the algorithm becomes numerically complex and difficult to use, especially in cases where the ground state is degenerate.
Fast convergence of path integrals for many-body systems
2008, Physics Letters, Section A: General, Atomic and Solid State PhysicsFast converging path integrals for time-dependent potentials: I. Recursive calculation of short-time expansion of the propagator
2011, Journal of Statistical Mechanics: Theory and ExperimentBose-Einstein Condensation: Twenty Years After
2015, Romanian Reports in PhysicsSPEEDUP Code for calculation of transition amplitudes via the effective action approach
2012, Communications in Computational Physics
- ☆
Supported by the Ministry of Science and Environmental Protection of the Republic of Serbia through project No. 141035.