🔢 mathematics

The weak Galerkin method for a class of Gross-Pitaevskii type eigenvalue problems

This paper proposes a weak Galerkin method for solving a class of Gross-Pitaevskii type nonlinear eigenvalue problems, proving that the scheme yields lower bounds for both the energy and the ground state eigenvalue via post-processing, and validates these theoretical results through numerical experiments.

Original authors: Wei Lu, Qilong Zhai

Published 2026-05-29

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Wei Lu, Qilong Zhai

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to find the most stable, lowest-energy state of a swirling cloud of particles (like a Bose-Einstein condensate, which is what the Gross-Pitaevskii equation models). In the real world, this cloud settles into a specific shape and energy level. Mathematically, finding this "ground state" is like trying to find the absolute bottom of a very bumpy, complex valley.

The paper you provided is about a new mathematical tool called the Weak Galerkin (WG) method designed to solve this specific problem on a computer. Here is a breakdown of what they did, using simple analogies:

1. The Problem: Finding the Bottom of the Valley

The authors are dealing with a "nonlinear eigenvalue problem."

The Analogy: Imagine you are blindfolded in a vast, foggy landscape trying to find the lowest point (the ground state energy). The landscape isn't just a smooth bowl; it has hills, bumps, and the shape of the ground changes depending on how you stand on it (this is the "nonlinear" part).
The Goal: You need to calculate exactly how low the valley is and what the shape of the ground looks like at that lowest point.

2. The Tool: The Weak Galerkin Method (The "Digital Mesh")

To solve this on a computer, you can't look at the whole landscape at once. You have to break it into small pieces (like a mosaic or a grid).

The Innovation: The "Weak Galerkin" method is a specific way of drawing this grid. Unlike traditional methods that demand the pieces fit together perfectly like a jigsaw puzzle, this method allows the pieces to be a bit "loose" or disconnected at the edges, as long as they fit together mathematically in a "weak" sense.
Why it helps: It's like using a flexible net to scoop up the shape of the valley. It allows the computer to handle complex shapes and irregular grids much more easily than rigid, traditional methods.

3. The Big Discovery: The "Safety Net" (Lower Bounds)

This is the most exciting part of the paper. Usually, when you use a computer to estimate the depth of a valley, you might accidentally dig a little too deep or not deep enough. You don't know if your answer is an overestimate or an underestimate.

The Claim: The authors proved that their specific Weak Galerkin method acts like a safety net. It guarantees that the energy level the computer calculates will never be higher than the true energy. It will always be slightly lower (or equal).
The Metaphor: Imagine you are trying to guess the depth of a swimming pool. Most methods might guess 5 feet, but the pool is actually 4.8 feet (you guessed too deep). This new method guarantees you will never guess 4.8 feet if the pool is actually 5 feet; you will always guess 5.1 feet or 5.0 feet. You are always "under" the true value, giving you a guaranteed lower bound.

4. Polishing the Result (Post-Processing)

The authors didn't stop at just getting a lower bound. They used a "post-processing" technique.

The Analogy: Think of the initial calculation as a rough sketch of the valley. The post-processing is like taking that sketch and smoothing it out with a fine brush.
The Result: By combining their rough sketch with a few extra calculations, they were able to create a very precise estimate that sits below the true answer, giving them a very tight range for where the true answer must be.

5. The Proof: Does it Work?

The paper is heavy on math proofs (theorems and lemmas), which are the authors' way of saying, "We proved this works for all possible scenarios, not just one lucky example."

The Experiments: They ran computer simulations on different types of "landscapes" (different mathematical potentials and shapes).
The Outcome: The computer results matched their theory perfectly. As they made the grid finer (more pieces in the mosaic), the error shrank rapidly, and the "lower bound" property held true every time.

Summary

In short, this paper introduces a new, flexible way to solve complex physics problems on a computer. Its superpower is that it doesn't just give you an answer; it gives you an answer that you can trust is guaranteed to be on the safe side (a lower bound) of the true energy. This is crucial for scientists who need to know the absolute minimum energy a system can have, ensuring they don't accidentally overestimate stability.

Technical Summary: The Weak Galerkin Method for a Class of Gross-Pitaevskii Type Eigenvalue Problems

Problem Statement
This paper addresses the numerical solution of a class of nonlinear eigenvalue problems arising from Gross-Pitaevskii type models. The problem is formulated as a variational minimization of an energy functional $E(v)$ over the Sobolev space $H^1_0(\Omega)$ subject to an $L^2$ -norm constraint. The energy functional includes a linear part involving a symmetric matrix $A(x)$ and a potential $V(x)$ , and a nonlinear part defined by a function $F(v^2)$ . The specific goal is to approximate the ground state eigenpair $(\lambda, u)$ , where $u$ is the minimizer and $\lambda$ is the corresponding eigenvalue.

Methodology
The authors employ the Weak Galerkin (WG) finite element method to discretize the nonlinear eigenvalue problem. The methodology proceeds through the following steps:

Discretization: A WG finite element space $V_h$ is constructed using piecewise polynomials. For an integer $k \ge 1$ , the discrete function $v_h = \{v_0, v_b\}$ consists of a polynomial $v_0$ of degree $k$ on each element and a polynomial $v_b$ of degree $k-1$ on the edges.
Weak Gradient: A discrete weak gradient operator $\nabla_w$ is defined via a weak formulation involving $L^2$ projections, allowing for discontinuous approximations across element boundaries.
Stabilization: A stabilization bilinear form $s(v_h, w_h)$ is introduced to control the difference between the interior and boundary components of the discrete functions.
Nonlinear Scheme: The continuous nonlinear operator is replaced by a discrete operator $A_{u_h}^h$ . The WG scheme seeks a discrete eigenpair $(u_h, \lambda_h)$ satisfying a discrete variational equation involving the discrete energy functional $E_h$ .
Post-Processing: To derive lower bounds for the eigenvalue, the authors utilize a post-processing technique that combines the WG approximation with conforming Lagrange element approximations (linear and quadratic).

Key Contributions and Theoretical Results
The paper provides a rigorous error analysis and establishes specific bounds for the WG approximation:

Lower Bound for Energy: The authors prove that the discrete energy $E_h(u_h)$ produced by the WG scheme serves as a lower bound for the exact ground state energy $E(u)$ , i.e., $E(u) - E_h(u_h) \ge 0$ .
Error Estimates: Under specific assumptions on the nonlinearity $F$ $F$ (Assumptions A1–A5) and the regularity of the ground state $u \in H^{k+1}(\Omega)$ $u \in H^{k + 1} (Ω)$ , the paper derives the following convergence rates for sufficiently small mesh size $h$ $h$ :
- $L^2$ -norm error: $\|u - u_h\| \lesssim h^{k+1-2\varepsilon}$ .
- Discrete $H^1$ -norm error: $\|u - u_h\|_{1,h} \lesssim h^{k-\varepsilon}$ .
- Eigenvalue error: $|\lambda - \lambda_h| \lesssim h^{k-\varepsilon}$ .
  Here, $\varepsilon \in (0, 1)$ is a small constant arising from the stabilization term.
Lower Bound for Eigenvalue: By combining the WG approximation with conforming element approximations, the authors construct a corrected eigenvalue $\tilde{\lambda}_h$ that satisfies a lower bound property: $0 \le \lambda - \tilde{\lambda}_h \lesssim h^{2-2\varepsilon}$ (for $k=1$ ).
Upper Bound for Eigenvalue: Similarly, an upper bound for the eigenvalue is derived using linear conforming elements, showing $0 \le \tilde{\lambda}_h - \lambda \lesssim h^2$ .

Numerical Results
The theoretical findings are validated through numerical experiments on unit square domains with various potentials $V(x,y)$ and nonlinearities $f(u^2)$ .

Convergence Rates: The experiments confirm the theoretical convergence orders. For $k=1$ , the observed order is approximately 2.0 for both eigenvalues and energy. For $k=2$ , the observed order is approximately 4.0.
Lower Bound Verification: The numerical data consistently shows that the computed WG energy and the post-processed eigenvalues are lower than the reference values (approximated by highly refined meshes), supporting the theoretical lower bound claims.
Robustness: The method demonstrates stability across different values of the nonlinearity parameter $\beta$ and different potential configurations.

Significance
The paper claims significance in establishing that the Weak Galerkin method not only provides accurate approximations for nonlinear eigenvalue problems but also inherently produces lower bounds for the ground state energy. This property is particularly valuable in physics and engineering applications where underestimating energy levels is often safer or more desirable than overestimation. Furthermore, the development of a post-processing technique to obtain lower and upper bounds for the eigenvalue offers a practical tool for error estimation and solution verification without requiring the exact solution. The work extends the applicability of WG methods from linear to nonlinear spectral problems with rigorous error analysis.