Entire solutions to a strongly competitive nonlinear Schrödinger system

1. Problem Statement

The authors investigate the existence of non-radial positive entire solutions to a strongly competitive system of nonlinear Schrödinger equations in $\mathbb{R}^N$ ($N \ge 2$):

$$\begin{cases} -\Delta u_1 + u_1 = u_1^p - \Lambda u_1^{a_1} u_2^{a_2} & \text{in } \mathbb{R}^N \\ -\Delta u_2 + u_2 = u_2^p - \Lambda u_1^{b_1} u_2^{b_2} & \text{in } \mathbb{R}^N \end{cases}$$

Key Parameters and Context:

• Competition Regime: The parameter $\Lambda \to +\infty$. In this regime, the components $u_1$ and $u_2$ tend to segregate (phase separation), meaning their supports become asymptotically disjoint.
• Non-Variational Nature: Unlike the specific Bose-Einstein case ($p=3, a_1=b_2=1, a_2=b_1-1=1$), the general system (1.1) with arbitrary exponents $a_i, b_i$ generally lacks a variational structure. This makes standard minimization techniques inapplicable.
• The Challenge: While radial solutions and sign-changing solutions for the scalar limit equation $-\Delta W + W = |W|^{p-1}W$ are well-understood, constructing non-radial positive solutions for the competitive system in the whole space $\mathbb{R}^N$ has been an open and difficult problem. Previous results were mostly restricted to bounded domains or weakly competitive regimes.

2. Methodology

The authors employ a Lyapunov-Schmidt reduction procedure combined with a finite-dimensional reduction strategy. The core steps are:

A. Construction of the Ansatz

The solution is constructed as a perturbation of an approximate solution $(U_1^*, U_2^*)$, which is a superposition of copies of the unique positive radial solution $U$ to the scalar equation $-\Delta U + U = U^p$.

• Geometric Configuration: The peaks (centers of the bubbles $U$) are arranged in a specific non-radial pattern involving two concentric regular $k$-gons.
  • Component 1 ($u_1$): Peaks are located along the edges of an outer polygon and along rays connecting the vertices of the inner and outer polygons.
  • Component 2 ($u_2$): Peaks are located alternately with $u_1$ on the edges of the outer polygon.
• Symmetry: The configuration is invariant under rotations of angle $2\pi/k$ and reflections, ensuring the resulting solution is non-radial but symmetric.

B. Balancing Conditions

To ensure the approximate solution is close to a true solution, the distances between peaks must satisfy specific balancing conditions derived from the interaction forces:

1. Geometric Condition: Relates the radii and side lengths of the polygons: $2m\ell \sin(\pi/k) + \bar{\ell}' = 2n\bar{\ell}$.
2. Force Equilibrium:
   • Inner Balance: Forces between peaks of the same component must balance (similar to previous works on scalar equations).
   • Outer Balance: Forces between peaks of different components (governed by $\Lambda$) must balance. This is the novel and critical part. The authors derive conditions where the repulsive force from the competing component balances the attractive/repulsive forces within the component.

C. Technical Innovations

• Non-Linear Interaction Expansion: In standard superposition principles, interactions are linear. Here, because $a_2, b_1 > 1$, the interaction terms $u_1^{a_1}u_2^{a_2}$ are non-linear. The authors perform delicate Taylor expansions to control these interactions, showing that despite the lack of linear superposition, the leading-order interaction forces behave consistently with a formal superposition principle.
• Parameter Constraints: They identify specific constraints on the exponents $a_i, b_i$ (Assumption 1.7) and the geometric parameters $(m, n, k)$ to ensure the error terms are controllable. Specifically, they require $c > 1$ and $\max\{a, b\} < c+1$, where $c = \min\{a_1, a_2, b_1, b_2\}$.

D. Reduction to Finite Dimensions

The infinite-dimensional problem is reduced to finding the parameters (positions of the peaks) that make the projection of the error onto the kernel of the linearized operator vanish. This leads to a system of algebraic equations for the translation parameters $(\alpha, \beta, \gamma)$.

3. Key Contributions

1. First Non-Radial Entire Solutions: This paper provides the first example of non-radial positive solutions for strongly competitive Schrödinger systems in the whole space $\mathbb{R}^N$.
2. General Non-Variational Framework: The method works for a broad class of non-variational systems, not just the specific Bose-Einstein case.
3. Handling Strong Competition: The authors successfully navigate the technical difficulties arising from the strong competition limit ($\Lambda \to \infty$) and non-linear coupling terms, proving that the "segregation" phenomenon can occur in complex, non-radial geometric patterns.
4. Infinitely Many Solutions: By varying the integer parameters $(m, n, k)$ (the number of peaks and the polygon symmetry), they prove the existence of infinitely many distinct non-radial solutions.

4. Main Results

Theorem 1.1: Under the sub-critical growth condition on $p$ and the exponent constraints (1.7), for any admissible triplet $(m, n, k)$ (where $k \ge 7$ is sufficiently large), there exists a threshold $\Lambda_0$ such that for all $\Lambda > \Lambda_0$, the system (1.1) admits a non-radial positive solution.

• Structure of the Solution: As $\Lambda \to \infty$, the solution $(u_1, u_2)$ converges to a configuration where $u_1$ and $u_2$ are sums of bubbles centered at the vertices and edges of the described polygonal configuration. The supports of $u_1$ and $u_2$ become disjoint.
• Admissibility: The paper proves (Lemma 2.2) that there are infinitely many admissible triplets $(m, n, k)$, ensuring the existence of infinitely many such solutions.

5. Significance and Limitations

• Significance:

  • It resolves Question (Q2) posed in the introduction regarding the existence of positive solutions resembling the positive/negative parts of non-radial sign-changing solutions of the scalar limit equation.
  • It contrasts with the cooperative case ($\Lambda < 0$), where Busca and Sirakov proved all positive solutions are radially symmetric. This highlights the richness of the competitive regime.
  • It opens the door to studying mixed cooperative/competitive systems with more components.

• Limitations & Remarks:

  • Bose-Einstein System: The result does not directly apply to the standard Bose-Einstein condensate system ($p=3, a_1=b_2=1, a_2=b_1-1=1$) because the exponents fall on the boundary of the required inequalities (critical case). The authors note that the approximate solution is not "good enough" for this specific case without further refinement.
  • Lotka-Volterra: The result does not cover the standard Lotka-Volterra case ($a_1=b_1=a_2=b_2=1$) due to the same criticality issues.
  • Dimension: The specific lower bound $p \ge 2$ simplifies the statement but is not strictly necessary; however, the case $1 < p \le 3/2$ cannot be treated with this method.

In conclusion, this work represents a major advancement in the theory of elliptic systems, demonstrating that strong competition can induce complex, non-radial spatial patterns in the entire space, even in the absence of a variational structure.