Sign-changing solutions to the Yamabe problem on manifolds with boundary

1. Problem Statement and Context

The Yamabe Problem with Boundary:
The paper addresses the existence of metrics conformally equivalent to a given metric $g$ on a compact Riemannian manifold $(M, g)$ with boundary $\partial M$, such that the new metric has:

1. Constant scalar curvature on $M$.
2. Constant mean curvature on $\partial M$.

Analytically, this reduces to finding a positive solution $u$ to the nonlinear boundary value problem:
$$\begin{cases} -\Delta_g u + c_n R_g u = a_n |u|^{\frac{4}{n-2}} u & \text{on } M, \\ \frac{\partial u}{\partial \nu} + d_n h_g u = b |u|^{\frac{2}{n-2}} u & \text{on } \partial M, \end{cases}$$
where $R_g$ is the scalar curvature, $h_g$ is the mean curvature, $\nu$ is the outward unit normal, and $a_n, c_n, d_n$ are dimensional constants.

The Gap:
While the existence of positive solutions to this problem is well-established (by Cherrier, Escobar, Han, and Li), the existence of sign-changing (nodal) solutions remains largely open. Nodal solutions correspond to "generalized metrics" (where the conformal factor changes sign) and are of significant interest in geometric analysis and spectral theory.

The Specific Goal:
The authors aim to prove the existence of least-energy sign-changing solutions (nodal solutions with minimal energy among all sign-changing solutions) under specific geometric conditions.

---

2. Methodology

The approach is variational, relying on the analysis of energy functionals, conformal invariants, and the concentration-compactness principle.

A. Variational Setting

The problem is formulated as finding critical points of the energy functional $J: H^1(M) \to \mathbb{R}$:
$$J(u) = \frac{1}{2} Q(u) - \frac{a_n}{p} \int_M |u|^p \, dv - \frac{b}{q} \int_{\partial M} |u|^q \, d\sigma$$
where $Q(u)$ is the quadratic form associated with the conformal Laplacian and boundary operator, and $p = \frac{2n}{n-2}, q = \frac{2(n-1)}{n-2}$ are critical exponents.

• Nehari Manifold ($\mathcal{N}$): The set of non-trivial functions where the derivative of $J$ vanishes along the direction of the function itself ($J'(u)u = 0$).
• Nodal Set ($\mathcal{E}$): The subset of $\mathcal{N}$ containing functions that change sign, such that both their positive part $u^+$ and negative part $u^-$ belong to $\mathcal{N}$.

B. Conformal Invariants

The authors define two key conformal invariants:

1. $\mu_1(M) = \inf_{u \in \mathcal{N}} J(u)$: The least energy for positive solutions.
2. $\mu_2(M) = \inf_{u \in \mathcal{E}} J(u)$: The least energy for sign-changing solutions.

The core strategy is to show that $\mu_2(M)$ is attained. This requires proving that a minimizing sequence for $\mu_2(M)$ does not lose compactness (i.e., does not "bubble off" to infinity or split into separate positive and negative components).

C. Key Technical Tools

1. Strict Positive Invariance: For $b \ge 0$, the authors prove that neighborhoods of the cones of positive ($P_+$) and negative ($P_-$) functions are strictly positively invariant under the negative gradient flow of $J$. This allows them to construct a Palais-Smale sequence that stays at a positive distance from both cones (ensuring the limit is sign-changing).
2. Minimax Characterization: They characterize $\mu_2(M)$ using a specific class of continuous maps $\Sigma$ connecting $P_+$ and $P_-$, utilizing the Poincaré-Miranda theorem to ensure the existence of a point in $\mathcal{E}$ along any such path.
3. Energy Estimates and Blow-up Analysis: To prove compactness, they must show that the energy level $\mu_2(M)$ is strictly below a critical threshold where compactness fails. This threshold is related to the energy of solutions on the spherical cap $B_b$ (conformally equivalent to the half-space $\mathbb{R}^n_+$).
4. Test Functions: They construct a specific test function $W_\epsilon$ based on the "bubble" solution $U$ on the half-space, modified by a cut-off function and a correction term $\phi$ to account for the non-umbilic boundary point.

---

3. Key Contributions and Results

Main Theorem (Theorem 1.1)

Assumptions:

• $n \ge 7$ (Dimension constraint).
• $b \ge 0$ (Non-negative boundary parameter).
• $M$ is a "positive" manifold (the first eigenvalue of the conformal Laplacian with boundary conditions is positive).
• $\partial M$ contains at least one non-umbilic point (a point where the second fundamental form is not a multiple of the identity).

Conclusion:
The problem (1.1) admits a least-energy sign-changing solution.

Key Intermediate Results

• Theorem 2.8: Establishes that if $\mu_2(M) < \mu_1(M) + \mu_1(B_b)$, then a least-energy nodal solution exists. This reduces the geometric problem to an energy inequality.
• Corollary 2.9: Provides a practical criterion to verify the energy inequality using a linear combination of a positive solution $u_0$ and a test function $W_\epsilon$.
• Energy Estimate (Section 3): The authors prove that for $n \ge 7$ and a non-umbilic boundary point, the energy of the constructed test function satisfies:
  $$J(su_0 - tW_\epsilon) < \mu_1(M) + \mu_1(B_b)$$
  for small $\epsilon$. This strict inequality is crucial. The term $-C_0 \epsilon^2$ arising from the non-umbilic geometry ensures the energy drops below the threshold required for compactness.

Why $n \ge 7$?

The dimension constraint arises from the asymptotic analysis of the error terms in the energy expansion. Specifically, the interaction terms between the positive solution $u_0$ and the bubble $W_\epsilon$ decay at a rate of $O(\epsilon^{(n-2)/2})$. For the negative quadratic term (from the non-umbilic correction) to dominate the error terms and ensure the strict inequality, one requires $(n-2)/2 > 2$, which implies $n > 6$.

---

4. Significance and Impact

1. Resolution of an Open Problem: This work provides the first existence result for least-energy nodal solutions to the Yamabe problem on manifolds with boundary under general geometric conditions (non-umbilic boundary). Previous results were limited to the case $b=0$ (minimal boundary) or required specific symmetries.
2. Extension of Han-Li Results: It extends the celebrated work of Han and Li (who solved the positive case for $b \in \mathbb{R}$) to the nodal case, specifically for $b \ge 0$.
3. Methodological Innovation: The paper introduces a robust variational framework using strictly invariant cones under the gradient flow. This technique effectively isolates sign-changing solutions from positive/negative ones, a method that could be applied to other nonlinear elliptic problems with critical growth.
4. Geometric Insight: The result highlights the role of the non-umbilic boundary points as the mechanism for breaking symmetry and lowering the energy of nodal configurations, analogous to how non-umbilic points or Weyl tensor non-vanishing drive existence in the closed manifold case.
5. Limitations and Future Work: The authors note that the case $b < 0$ remains an open and challenging problem because the invariance of the positive/negative cones under the gradient flow is not guaranteed when the boundary term has a negative coefficient.

In summary, the paper successfully bridges the gap between the existence of positive and sign-changing solutions for the Yamabe problem with boundary, utilizing sophisticated variational techniques and precise energy estimates to overcome the lack of compactness inherent in critical exponent problems.