Numerical comparison of mass-conservative schemes for the Gross-Pitaevskii equation

Abstract

In this paper we present a numerical comparison of various mass-conservative discretizations for the time-dependent Gross-Pitaevskii equation. We have three main objectives. First, we want to clarify how purely mass-conservative methods perform compared to methods that are additionally energy-conservative or symplectic. Second, we shall compare the accuracy of energy-conservative and symplectic methods among each other. Third, we will investigate if a linearized energy-conserving method suffers from a loss of accuracy compared to an approach which requires to solve a full nonlinear problem in each time-step. In order to obtain a representative comparison, our numerical experiments cover different physically relevant test cases, such as traveling solitons, stationary multi-solitons, Bose-Einstein condensates in an optical lattice and vortex pattern in a rapidly rotating superfluid. We shall also consider a computationally severe test case involving a pseudo Mott insulator. Our space discretization is based on finite elements throughout the paper. We will also give special attention to long time behavior and possible coupling conditions between time-step sizes and mesh sizes. The main observation of this paper is that mass conservation alone will not lead to a competitive method in complex settings. Furthermore, energy-conserving and symplectic methods are both reliable and accurate, yet, the energy-conservative schemes achieve a visibly higher accuracy in our test cases. Finally, the scheme that performs best throughout our experiments is an energy-conserving relaxation scheme with linear time-stepping proposed by C. Besse (SINUM,42(3):934-952,2004).

A note on optimal $$H^1$$-error estimates for Crank-Nicolson approximations to the nonlinear Schrödinger equation

“…When solving a NLS numerically it is therefore of great importance to also reproduce this conservation on the discrete level. This aspect was emphasized by various numerical studies [21,22], where it was also found that the complexity of the physical setup (or low-regularity) can stress this issue even further .…”

“…when effects close to phase transitions are studied), then the performance of these methods can drop dramatically. Here we refer exemplarily to the recent numerical experiments reported in [19][20][21]. To overcome this issue, Ostermann and Schratz proposed new low-regularity time-integrators [19,20] which improve the convergence in low regularity regimes significantly.…”

“…For the subclass of power law nonlinearities of the form γ (|u| 2 ) = K k=1 κ k |u| 2σ k for σ k ≥ 0 and α k ∈ R, a mass and energy conserving relaxation scheme was proposed and analyzed by Besse [23,24]. Thanks to its properties, the scheme shows a very good performance in realistic physical setups [21]. Despite the large variety of different numerical approaches for solving the time-dependent NLS (cf.…”

Energy preserving relaxation method for space-fractional nonlinear Schrödinger equation

“…They have been applied to different NLS equation for example in the context of plasma physics [22]. For cubic nonlinearities (σ " 1 in (1)), they are linearly implicit hence very popular [4,15,6,18,20]. They preserve the L 2 -norm and a discrete analogue of (2).…”

Convergence of exponential Lawson-multistep methods for the MCTDHF equations

“…Semi-implicit finite difference schemes lose most of the desired properties. A relaxation method based on a finite element spatial discretization was recently found to perform well in [24] for problems with low spatial regularity. Extensive numerical comparisons in the literature show the accuracy and efficiency of splitting methods for various quantum mechanical models under a number of different spatial discretizations, see for instance [8,16].…”