A Hybrid High-Order Method for the Gross–Pitaevskii Eigenvalue Problem

Moritz Hauck ∗ , Yizhou Liang † · 2025

Read A Hybrid High-Order Method for the Gross–Pitaevskii Eigenvalue Problem by Moritz Hauck ∗ , Yizhou Liang † (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-19944.txt.

Introduction

Moritz Hauck and Yizhou Liang introduce a hybrid high-order method for finding the ground state of a nonlinear eigenvalue problem from quantum physics, published in 2025. The problem, the Gross-Pitaevskii equation, models systems such as ultracold gases. Standard conforming methods give upper bounds on the ground state energy, which is only half of the picture.

The new method is proved to converge at an optimal rate for the ground state, its eigenvalue, and its energy. Unlike classical methods, it yields guaranteed lower bounds on the energy that grow exact in the limit. Those bounds come directly from the method, with no post-processing step. The authors say this gives more accurate guaranteed lower bounds in practice than prior approaches that needed extra work.

Catalog record

CREATORMoritz Hauck ∗ , Yizhou Liang †
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE89 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-19944

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A Hybrid High-Order Method for the Gross–Pitaevskii Eigenvalue Problem Moritz Hauck ∗ , Yizhou Liang † Abstract . We introduce a hybrid high-order method for approximating the ground state of the nonlinear Gross–Pitaevskii eigenvalue problem. Optimal convergence rates are proved for the ground state approximation, as well as for the associated eigenvalue and energy approximations. Unlike classical conforming methods, which inherently provide upper bounds on the ground state energy, the proposed approach gives rise to guaranteed and asymptotically exact lower energy bounds. Importantly, and in contrast to previous works, they are obtained directly without the need of post-processing, leading to more accurate guaranteed lower energy bounds in practice. 1991 Mathematics Subject Classification 65N12, 65N15, 65N25, 65N30 1. Introduction The Gross–Pitaevskii eigenvalue problem (GP-EVP) arises in quantum physics, where it describes stationary quantum states of bosonic particles at ultracold temperatures, known as Bose–Einstein condensates; see, e.g., [ 13 , 4 , 19 ] . The problem involves a non-negative trapping potential V\in L^{\infty}(\Omega) and a parameter \kappa>0 describing the strength of the repulsive particle interactions. As computational domain we consider a bounded convex Lipschitz domain \Omega\subset\mathbb{R}^{d} , {d\in\{2,3\}} , where we note that the restriction to a sufficiently large domain, along with homogeneous Dirichlet boundary conditions, is a standard and physically reasonable assumption for modeling low-energy quantum states, cf. [ 5 ] . Mathematically, the GP-EVP seeks L^{2} -normalised eigenfunctions \{u_{j}\,:\,j=1,2,\dots\}\subset H^{1}_{0}(\Omega) and corresponding eigenvalues \{\lambda_{j}\,:\,j=1,2,\dots\} such that (1.1) -\Delta u_{j}+Vu_{j}+\kappa|u_{j}|^{2}u_{j}=\lambda_{j}u_{j} holds in the weak sense. The function |u_{j}|^{2} represents the density of the stationary quantum state u_{j} , and the eigenvalue \lambda_{j} is typically called the chemical potential. It is a classical result that all eigenvalues of Equation 1.1 are real and positive, and the smallest eigenvalue is simple; see, e.g., [ 7 ] . Without loss of generality, we assume that the eigenvalues are ordered nondecreasingly, i.e., 0<\lambda_{1}<\lambda_{2}\leq\lambda_{3}\leq\dots . The GP-EVP arises as the Euler–Lagrange equations for critical points of the Gross–Pitaevskii energy, defined for all v\in H^{1}_{0}(\Omega) as, (1.2) \mathcal{E}(v)\coloneqq\tfrac{1}{2}(\nabla v,\nabla v)_{L^{2}}+\tfrac{1}{2}(Vv,v)_{L^{2}}+\tfrac{\kappa}{4}(|v|^{2}v,v)_{L^{2}}, subject to the L^{2} -normalisation constraint. The ground state, which represents the stationary quantum state of lowest energy, is characterised as the minimiser of the Gross–Pitaevskii energy, i.e., (1.3) u\in\argmin_{v\in H^{1}_{0}(\Omega)\,:\,\|v\|_{L^{2}}=1}\mathcal{E}(v). We emphasise that, under the above assumptions on \Omega and V , the constrained minimisation problem Equation 1.3 admits a unique global minimiser (up to sign). Moreover, this ground state u coincides, up to sign, with the eigenfunction u_{1} associated with the smallest eigenvalue \lambda_{1} of Equation 1.1 . The minimal energy E:=\mathcal{E}(u) is related to the smallest eigenvalue by the identity \lambda_{1}=2E+\tfrac{\kappa}{2}\|u\|_{L^{4}}^{4} . For further theoretical results on the Gross–Pitaevskii problem, see, for example, [ 7 ] . A wide range of discretisation methods has been proposed in the literature to approximate the Gross–Pitaevskii ground state. Most existing approaches are based on H^{1}_{0} -conforming discretisations, including standard continuous finite elements [ 33 , 7 , 10 ] , spectral and pseudospectral methods [ 7 , 5 ] , multiscale methods [ 24 , 28 , 26 , 31 ] , and mesh-adaptive methods [ 14 , 27 ] . Recently, also non-standard conforming finite element methods using mass lumping techniques to preserve certain positivity properties of the ground state have been introduced; see [ 11 , 23 ] . A characteristic property of conforming discretisations is that they approximate the ground state energy from above, as the energy functional is minimised over a subset. In the context of linear eigenvalue problems, where the terms energy and eigenvalue can be used interchangeably, several strategies have been developed to compute lower eigenvalue bounds. Among these, mixed finite element methods have proven effective [ 20 ] , and this approach has recently been extended to the present nonlinear GP-EVP in [ 21 ] . However, this method typically requires an additional postprocessing step to obtain asymptotically exact lower bounds. An alternative approach that gives guaranteed lower eigenvalue bounds without the need for post-processing is provided by hybrid high-order (HHO) methods; see [ 17 ] for an overview. In the linear setting, such bounds have been established, for example in [ 8 , 9 , 32 ] ; see also [ 12 ] for a similar result based on the related hybridisable discontinuous Galerkin method. In this work, we extend the HHO methodology to the GP-EVP, aiming to obtain guaranteed lower bounds on the ground state energy without the need for post-processing. To this end, we introduce a modified version of the HHO method that employs a lowest-order quadrature rule for the nonlinearity. This modification is crucial for estimating the nonlinear term in the proof of the guaranteed lower energy bound via Jensen’s inequality. Note that the lack of higher-order generalisations of Jensen’s inequality prevents us from deriving high-order guaranteed lower energy bounds for the Gross–Pitaevskii problem. Although the guaranteed lower energy bounds we obtain are only of lowest order, they can still be significantly more accurate than those from the post-processed lowest-order mixed discretization in [ 21 ] . The improved accuracy is due to the absence of post-processing, which can degrade the quality of the bounds, particularly for smooth problems where discretization errors are comparatively small. Numerical experiments clearly demonstrate this improvement, showing that the guaranteed lower energy bounds of the proposed modified HHO method are more accurate by up to two orders of magnitude than the post-processed mixed discretisation from [ 21 ] . In addition to providing guaranteed lower energy bounds, HHO methods offer improved convergence rates for the reconstructed unknowns compared to classical finite element methods. Specifically, when using k -th order polynomials on the mesh faces and assuming sufficient smoothness, the reconstructed ground state approximation exhibits optimal convergence rates of order k+1 in the H^{1} -seminorm and k+2 in the L^{2} -norm. We provide a rigorous convergence analysis establishing these rates, along with optimal convergence of order 2k+2 for the energy and eigenvalue approximations. Furthermore, we analyse the convergence of the proposed modified HHO method and prove first-order H^{1} -convergence of the ground state approximation, second-order L^{2} -convergence, and second-order convergence for the energy and eigenvalue approximations. Notably, the presented error analysis differs substantially from standard techniques in the linear setting. Instead, it is inspired by the analysis conducted in [ 7 , 21 ] for classical conforming and mixed discretisations of the Gross–Pitaevskii problem, respectively. The remainder of the paper is organised as follows. Section 2 introduces the HHO method for the Gross–Pitaevskii problem. Guaranteed lower energy bounds for a modified HHO method are established in Section 3 . Section 4 presents an a priori convergence analysis of the HHO method and its modified version. Numerical experiments supporting our theoretical results are presented in Section 5 . 2. Hybrid high-order method Consider a hierarchy of simplicial meshes \{\mathcal{T}_{h}\}_{h>0} of the domain \Omega , which we assume to be geometrically conforming and shape regular. We denote by T the elements of \mathcal{T}_{h} , and define the maximal mesh size of \mathcal{T}_{h} as h\coloneqq\max_{T\in\mathcal{T}_{h}}h_{T} , where h_{T}\coloneqq\operatorname{diam}(T) . The set of mesh faces of \mathcal{T}_{h} , denoted by \mathcal{F}_{h} , is partitioned into the set of interior faces \mathcal{F}_{h}^{i} and boundary faces \mathcal{F}_{h}^{b} . For any T\in\mathcal{T}_{h} , we denote by \mathcal{F}_{\partial T}

Read the complete text