Convergence of semilinear parabolic flows with general initial data

1. Problem Statement

The paper investigates the long-time asymptotic behavior ($t \to \infty$) of non-negative solutions $u(t, x)$ to the semilinear parabolic equation (Cauchy problem) in the Euclidean space $\mathbb{R}^n$ ($n \geq 3$):
$$\begin{cases} \partial_t u = \Delta u - f(u) & \text{in } (0, \infty) \times \mathbb{R}^n, \\ u(t, x) \to 0 & \text{as } |x| \to \infty, \\ u(0, x) = u_0(x). \end{cases}$$
The equation is viewed as the $L^2$-gradient flow of the energy functional:
$$J(v) = \int_{\mathbb{R}^n} \left( \frac{1}{2}|\nabla v|^2 + F(v) \right) dx, \quad \text{where } F(v) = \int_0^v f(s) ds.$$

The Core Challenge:
In bounded domains, solutions typically converge to a unique stationary state. However, in $\mathbb{R}^n$, the energy functional $J$ is translation-invariant, leading to a lack of compactness. This allows for "bubbling phenomena," where a sequence of solutions might converge to a superposition of multiple stationary solutions (ground states) separated by infinite distances, rather than a single profile.

• Previous Limitations: Existing literature (e.g., [CdPE99, FP97]) established convergence to a single ground state but required the initial data $u_0$ to have compact support.
• Open Question: Does the flow converge to a unique ground state for general initial data (including those with non-compact support), or can bubbling occur?

2. Methodology

The author employs a variational approach combined with quantitative stability analysis of critical points. The strategy relies on two main pillars:

A. Quantitative Stability of Almost Critical Points

The solution $u(t)$ is treated as a "nearly critical point" of $J$ because $\nabla J(u) = \partial_t u$, and $\|\partial_t u\|_{L^2} \to 0$ as $t \to \infty$ (due to energy dissipation).

• The paper establishes a sharp stability estimate (analogous to quantitative functional inequalities) for solutions near a superposition of "bubbles" (ground states).
• It proves that if the energy deficit is small, the solution is close to a superposition of ground states $\sum \alpha_i \tau_{x_i}[\xi]$, where $\xi$ is the unique radial ground state and $\tau$ denotes translation.

B. Control of Interaction and Bubbling

To rule out the formation of multiple bubbles (bubbling), the author analyzes the interaction between separated ground states.

• Energy Deficit Estimate: The paper derives an inequality linking the energy difference $J(u) - M J(\xi)$ to the distance between $u$ and the best-matching multi-bubble configuration.
• Tail Interaction Control: A crucial step involves proving that the interaction energy between widely separated bubbles is energetically unfavorable under specific conditions on $f$. This prevents the flow from splitting into multiple distinct solutions at infinity.
• Dichotomy Argument: The proof utilizes a "stability dichotomy": either the time derivative $\|\partial_t u\|_{L^2}$ is large (driving the system away from a multi-bubble state), or the solution is coercively stable near a single bubble.

C. Key Assumptions on Nonlinearity $f$

The results hold for nonlinearities satisfying:

1. $f \in C^{1,\beta}_{loc}$, $f(0)=0$, $f'(0) > 0$.
2. Uniqueness & Non-degeneracy: The stationary problem $\Delta \xi = f(\xi)$ has a unique positive radial solution (up to translation), and the linearized operator $D^2J(\xi)$ has a kernel generated only by translations $\{\partial_i \xi\}$.
3. Structural Conditions (H1 & H2):
   • (H1) $t f'(0) \geq f(t)$ for $t \in [0, \max \xi]$.
   • (H2) $f$ is concave in a neighborhood of 0.
   • These conditions ensure that the "tail" interactions between separated bubbles do not lower the energy, preventing the splitting of the solution.

3. Key Contributions

1. Removal of Compact Support Assumption: The primary contribution is proving convergence for general initial data $u_0 \in W^{1,2}(\mathbb{R}^n)$, removing the restrictive requirement of compact support found in prior works.
2. Resolution of the Bubbling Question: The paper provides a partial answer to the open problem posed in [Fei97], demonstrating that for the specific class of nonlinearities considered, bubbling does not occur for general initial data. The flow converges to a unique ground state.
3. Exponential Convergence Rates: Unlike previous results that established convergence without rates (or only algebraic rates), this paper proves exponential convergence:
   $$\|u(t, \cdot) - \tau_{x^*}[\xi]\|_{W^{2,2} \cap C^2} \leq C e^{-t/C}.$$
4. Extension to Constrained Flows: The methodology is adapted to volume-constrained gradient flows (e.g., Allen-Cahn with volume constraint), showing that the strategy applies even when the equation becomes non-autonomous due to Lagrange multipliers.
5. New Stability Estimates: The paper develops new quantitative stability estimates for multi-bubble configurations in $\mathbb{R}^n$, refining the concentration-compactness theory by quantifying the error terms in the decomposition of almost-critical points.

4. Main Results

Theorem 1.1 (Main Convergence Result):
Let $n \geq 3$ and $f$ satisfy the structural conditions (including the specific form $f(t) = a_0 t - \sum a_i t^{p_i}$). For any non-negative initial data $u_0 \in W^{1,2}(\mathbb{R}^n)$, the solution $u(t)$ exists uniquely and satisfies one of the following:

1. Vanishing: $\lim_{t \to \infty} J(u(t)) = 0$ and $\|u(t)\| \to 0$.
2. Convergence to Ground State: There exists a unique radial positive solution $\xi$ and a shift $x^* \in \mathbb{R}^n$ such that:
   $$J(u(t)) - J(\xi) \leq C e^{-t/C}, \quad \|u(t) - \tau_{x^*}[\xi]\|_{W^{2,2} \cap C^2} \leq C e^{-t/C}.$$

Theorem 2.6 (Generalized Version):
Extends Theorem 1.1 to a broader class of nonlinearities satisfying the general hypotheses (H1) and (H2), provided the ground state is unique and non-degenerate.

Corollary (Threshold Phenomenon):
The results extend the known trichotomy (vanishing, convergence, blow-up) for compactly supported data to non-compactly supported data.

5. Significance and Impact

• Theoretical Advancement: This work bridges the gap between the theory of gradient flows in bounded domains (where compactness is standard) and unbounded domains (where translation invariance causes non-compactness). It demonstrates that for a wide class of problems, the "bubbling" instability is suppressed by the specific structure of the nonlinearity, even without compact support.
• Methodological Innovation: The technique of using quantitative stability estimates to control the interaction of bubbles is a powerful tool. It moves beyond qualitative "subsequence" convergence arguments to provide explicit rates and global convergence guarantees.
• Applications: The framework is applicable to various physical models involving phase transitions, pattern formation, and geometric flows in unbounded media. The extension to volume-constrained flows suggests potential applications in isoperimetric problems and diffuse interface models.
• Resolution of Open Problems: By answering the question of whether bubbling can occur for general data, the paper settles a long-standing issue in the analysis of semilinear heat equations in $\mathbb{R}^n$.

In summary, Restrepo's paper provides a rigorous, quantitative proof that semilinear parabolic flows with general initial data converge exponentially to a unique ground state, effectively ruling out the formation of multiple bubbles at infinity under specific structural conditions on the reaction term.