Abstract:The computation of the ground states of special multi-component Bose-Einstein condensates (BECs) can be formulated as an energy functional minimization problem with spherical constraints. It leads to a nonconvex quartic-quadratic optimization problem after suitable discretizations. First, we generalize the Newton-based methods for single-component BECs to the alternating minimization scheme for multi-component BECs. Second, the global convergent alternating Newton-Noda iteration (ANNI) is proposed. In particular, we prove the positivity preserving property of ANNI under mild conditions. Finally, our analysis is applied to a class of more general "multi-block" optimization problems with spherical constraints. Numerical experiments are performed to evaluate the performance of proposed methods for different multi-component BECs, including pseudo spin-1/2, anti-ferromagnetic spin-1 and spin-2 BECs. These results support our theory and demonstrate the efficiency of our algorithms.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is the problem of calculating the ground state of multi - component Bose - Einstein condensates (BECs). Specifically, the authors focus on how to effectively and globally convergently calculate the ground state of specific multi - component BECs, ensuring that the algorithm can maintain the non - negativity of the solution under certain conditions.
### Problem Background
1. **Bose - Einstein Condensates (BECs)**: Since the first realization of BEC in 1995, researchers have carried out a large number of explorations on the theoretical research and numerical methods of single - component BECs. However, with the progress of experimental techniques, multi - component BEC systems have gradually become a research hotspot, especially various spin - condensed states realized by optical confinement, such as pseudo - spin - 1/2, antiferromagnetic spin - 1 and spin - 2 BECs.
2. **Ground State Problem**: In BEC, finding the ground state is a fundamental problem, which is defined as the solution to the minimization problem of the Gross - Pitaevskii (GP) energy functional under certain physical constraints. For multi - component BECs, this problem can be transformed into a non - convex quartic - quadratic optimization problem with spherical constraints.
### Main Contributions of the Paper
1. **Generalization of the Newton Method**: The authors generalize the Newton method used for single - component BECs to an alternating minimization scheme to deal with multi - component BECs.
2. **Proposal of the ANNI Algorithm**: A globally convergent alternating Newton - Noda iteration (ANNI) is proposed, and its non - negativity - maintaining property is proved under mild conditions.
3. **Application and Verification**: The analysis is applied to more general "multi - block" optimization problems, and the performance of the proposed method on different multi - component BECs, including pseudo - spin - 1/2, antiferromagnetic spin - 1 and spin - 2 BECs, is evaluated through numerical experiments.
### Mathematical Model
The mathematical model in the paper mainly involves the following energy functional minimization problem:
\[
\min E(\Phi)=\int_{\mathbb{R}^d}\left[\frac{1}{2}(|\nabla\phi_1|^2 + |\nabla\phi_2|^2)+V(x)(|\phi_1|^2 + |\phi_2|^2)+\frac{1}{2}\beta_{11}|\phi_1|^4+\frac{1}{2}\beta_{22}|\phi_2|^4+\beta_{12}|\phi_1|^2|\phi_2|^2\right]dx
\]
where:
- \(x\in\mathbb{R}^d\) is the spatial coordinate vector,
- \(\Phi = (\phi_1(x),\phi_2(x))^T\) is a two - component vector wave function,
- \(V(x)\) is a real - valued external trapping potential,
- \(\beta_{11},\beta_{22},\beta_{12}\in\mathbb{R}\) are interaction constants.
### Discretized Optimization Problem
After discretization, the problem is transformed into a non - convex optimization problem with spherical constraints:
\[
\min_{u,v\in\mathbb{R}^n}f(u,v)=\frac{\beta_{11}}{2}\sum_{i = 1}^n u_i^4+u^TA_1u+\frac{\beta_{22}}{2}\sum_{i = 1}^n v_i^4+v^TA_2v+\beta_{12}\sum_{i = 1}^n u_i^2v_i^2
\]
\[
\text{s.t.}\quad u^Tu = 1,\quad v^Tv = 1
\]
where \(A_1\) and \(A_2\) are discretization matrices, representing the sum of the negative Laplace operator and the potential \(V(x)\).
### Conclusion
By introducing the alternating Newton - Noda iteration (ANNI), the paper provides an efficient and globally convergent algorithm to calculate the ground state of multi - component BECs, and under certain conditions...