Dynamics of threshold solutions for energy critical NLS with inverse square potential

Imagine you are watching a complex dance performance on a stage. The dancers are waves of energy (mathematically called "solutions" to a Nonlinear Schrödinger Equation, or NLS). The stage has a special, tricky feature: a deep, invisible pit in the center (the "inverse square potential") that pulls everything toward the middle, making the dance much harder to predict than on a flat stage.

This paper is about understanding the critical moment in this dance. Specifically, it asks: What happens when the dancers have exactly the same amount of energy as the most stable, perfect formation known as the "Ground State" (let's call it The Master Formation)?

Here is the breakdown of what the authors discovered, using simple analogies:

1. The Setup: The Energy Balance

Think of the "Kinetic Energy" as the dancer's speed and movement, and the "Potential Energy" as the pull of the pit.

The Master Formation (W): This is a perfect, stationary pose. It's the most efficient way to stand still in this pit.
The Threshold: The paper looks at dancers who have exactly the same total energy as this Master Formation.

2. The Slow Dancers (Below the Threshold)

Imagine a dancer who is moving slightly slower than the Master Formation.

The Result: If they are moving too slowly, they have two fates:
1. Scattering: They lose their energy, slow down completely, and drift away into the distance, disappearing into nothingness (mathematically, they "scatter to zero").
2. The Stable/Unstable Manifold: They get caught in a very specific, narrow path.
  - The "Unstable" Path: They start slightly off-center and are pulled toward the Master Formation, slowly settling into that perfect pose as time goes forward.
  - The "Stable" Path: They start in the perfect pose but are nudged slightly off. They slowly drift away from the pose, but only along a very specific, narrow track.
The Analogy: Think of a marble rolling on a hill. If it's below a certain speed, it either rolls down into a valley (scattering) or gets stuck in a tiny groove that leads exactly to the top of a specific bump (the Master Formation).

3. The Fast Dancers (Above the Threshold)

Now, imagine a dancer moving faster than the Master Formation.

The Result: In most cases, this is a disaster. The dancer moves so fast that the pull of the pit and their own speed create a feedback loop. They spiral inward, getting tighter and tighter until they crash (mathematically, they "blow up" in finite time).
The Exception: There are two very special, rare paths (in 5-dimensional space) where a fast dancer doesn't crash. Instead, they are on a "highway" that leads them away from the Master Formation without ever crashing. But if they aren't on this specific highway, they crash.

4. The Tools Used to Solve the Puzzle

How did the authors figure this out? They used three main tools:

Spectral Analysis (The X-Ray): They looked at the "Master Formation" and asked, "If I poke it slightly, how does it vibrate?" They found that the formation is like a saddle: it's stable in some directions but unstable in others. They identified the "safe" directions and the "dangerous" directions.
Invariant Manifold Theory (The Train Tracks): They realized that the "dangerous" directions aren't just random chaos; they form smooth, curved tracks (manifolds). If a dancer is on these tracks, they will either slide smoothly toward the Master Formation or slide smoothly away from it.
Virial Analysis (The Speedometer): This is a mathematical way of measuring how "spread out" the dancer is. The authors used this to prove that if a dancer isn't on one of those special tracks, they must eventually crash or drift away. It's like a speedometer that tells you, "If you aren't on the highway, you're going to hit a wall."

5. The Big Picture

The paper essentially draws a map of the "Energy Surface."

The Center: The Master Formation (The Ground State).
The Tracks: Two narrow paths leading to and from the center (the Stable and Unstable manifolds).
The Rest of the Map:
- If you are slower than the Master Formation, you either drift away or get stuck on the tracks.
- If you are faster, you almost certainly crash, unless you are incredibly lucky and on one of the two special "escape tracks."

Why Does This Matter?

In the world of physics, these equations describe how light behaves in fiber optics or how atoms behave in Bose-Einstein condensates. Understanding these "threshold" behaviors helps scientists predict when a system will stay stable and when it will collapse.

The authors had to overcome a huge hurdle: the "pit" in the center (the inverse square potential) breaks the usual symmetries of the universe (like translation). It's like trying to dance on a stage that is constantly shifting under your feet. They had to invent new ways to measure the dance to prove that even with this tricky stage, the rules of the dance are actually quite orderly and predictable.

In short: They proved that at the exact edge of stability, nature is very picky. You either drift away, get stuck in a perfect pose, or crash—unless you are on one of the very few "magic paths" that allow you to escape the crash.

Here is a detailed technical summary of the paper "Dynamics of Threshold Solutions for Energy Critical NLS with Inverse Square Potential" by Kai Yang, Chongchun Zeng, and Xiaoyi Zhang.

1. Problem Statement

The paper investigates the long-time dynamics of solutions to the focusing energy-critical Nonlinear Schrödinger Equation (NLS) with an inverse square potential in dimension $d=3$ (with remarks on $d=4,5$ ). The equation is given by:
$(i\partial_t - L_a)u + |u|^{\frac{4}{d-2}}u = 0, \quad u(0,x) = u_0$
where $L_a = -\Delta + \frac{a}{|x|^2}$ and $a \in (-\frac{1}{4}, 0)$ . The exponent $\frac{4}{d-2}$ corresponds to the energy-critical case.

The central object of study is the ground state soliton $W$ , which is the unique positive solution (up to symmetries) to the elliptic equation $L_a W = |W|^{\frac{4}{d-2}}W$ . The authors aim to classify all solutions lying on the energy surface $E(u) = E(W)$ , specifically distinguishing between solutions with kinetic energy less than, equal to, or greater than that of the ground state ( $\|u\|_{\dot{H}^1_a} \lessgtr \|W\|_{\dot{H}^1_a}$ ).

2. Methodology

The authors employ a combination of spectral analysis, invariant manifold theory, and global virial analysis. The methodology proceeds in several stages:

A. Spectral Analysis of the Linearized Operator

The authors linearize the NLS equation around the ground state $W$ . Letting $v = u - W$ , the linearized operator is $JL$ , where $J$ is the symplectic structure and $L$ is the Hessian of the energy functional.

Spectral Decomposition: They prove that the operator $iJL$ $i J L$ possesses a specific spectral structure:
- A 1-dimensional unstable subspace ( $E_u$ ) and a 1-dimensional stable subspace ( $E_s$ ).
- A codimension-2 center subspace ( $E_c$ ) containing the kernel of $L$ (generated by scaling and phase symmetries).
- Crucially, they establish the absence of a generalized kernel for $JL$ , which allows them to apply the exponential trichotomy theory (from previous work [21]).
Technical Challenge: The inverse square potential $a/|x|^2$ is non-perturbative (it scales like the Laplacian). This breaks translation symmetry and introduces singularities at the origin. The authors perform a detailed analysis of the asymptotic behavior of the eigenfunctions near $r=0$ and $r=\infty$ to handle the singular potential and prove the spectral gap.

B. Modulation Analysis

To study solutions near the manifold of ground states $\{g_{\theta, \mu}W\}$ , the authors use modulation theory.

They decompose a solution $u$ near $W$ as $u = g_{\theta(t), \mu(t)}(W + \alpha(t)W + v(t))$ , where $v(t)$ is orthogonal to the kernel directions ( $iW, W_1$ ).
They derive differential equations for the modulation parameters $\theta(t)$ (phase) and $\mu(t)$ (scaling), showing that their evolution is controlled by the distance function $d(u) = |\|u\|_{\dot{H}^1_a}^2 - \|W\|_{\dot{H}^1_a}^2|$ .

C. Local Invariant Manifold Construction

Using the spectral trichotomy and Strichartz estimates adapted to the singular potential, they construct local stable and unstable manifolds ( $W^+$ and $W^-$ ) tangent to the unstable and stable eigenspaces at $W$ .

They prove the existence of unique solutions $W^\pm$ that converge exponentially to $W$ as $t \to \mp \infty$ .
$W^-$ corresponds to the branch with kinetic energy less than $W$ , and $W^+$ to the branch with kinetic energy greater than $W$ .

D. Global Virial Analysis

To extend the local dynamics to global behavior, the authors utilize a Virial identity.

They define a truncated Virial functional $V_R(t)$ and analyze its second derivative $\partial_{tt} V_R(t)$ .
The sign of $\partial_{tt} V_R(t)$ $\partial_{tt} V_{R} (t)$ is determined by the sign of $d(u(t))$ $d (u (t))$ .
- If $\|u\| < \|W\|$ , $\partial_{tt} V_R \approx 16 d(u) > 0$ (convexity).
- If $\|u\| > \|W\|$ , $\partial_{tt} V_R \approx -16 d(u) < 0$ (concavity).
This monotonicity is used to rule out "runaway" solutions and prove that non-scattering solutions must stay close to the ground state manifold.

E. Compactness and Precompactness

Sub-critical case ( $\|u\| < \|W\|$ ): In dimensions $d=4,5$ , and the radial case in $d=3$ , the authors utilize the concentration-compactness principle to show that non-scattering solutions are precompact modulo scaling. This allows them to control the error terms in the Virial estimate and prove exponential decay.
Super-critical case ( $\|u\| > \|W\|$ ): Compactness does not hold generally. The authors impose radial symmetry and $L^2$ boundedness to control the solution's behavior and prevent it from escaping to low frequencies, leading to finite-time blow-up.

3. Key Contributions and Results

A. Existence and Uniqueness of Threshold Solutions (Theorem 1.2)

The authors prove the existence of two unique (up to time translation) solutions $W^+$ and $W^-$ on the energy surface $E(W)$ :

$W^-$ : Converges exponentially to $W$ as $t \to \infty$ with $\|W^-\|_{\dot{H}^1_a} < \|W\|_{\dot{H}^1_a}$ .
$W^+$ : Converges exponentially to $W$ as $t \to -\infty$ with $\|W^+\|_{\dot{H}^1_a} > \|W\|_{\dot{H}^1_a}$ .
These solutions form the 1-dimensional stable and unstable manifolds of $W$ .

B. Classification of Sub-Critical Solutions (Theorem 1.5, Part b)

For solutions with $E(u) = E(W)$ and $\|u_0\|_{\dot{H}^1_a} < \|W\|_{\dot{H}^1_a}$ :

If the solution does not scatter (i.e., the scattering norm is infinite), it must belong to the stable/unstable manifold of the ground state.
Specifically, such solutions converge exponentially to $W$ (or $W^-$ ) in the energy space.
Note: In 3D without radial symmetry, this result is conditional on the precompactness assumption of non-scattering solutions (Assumption 1.4). In 4D and 5D, the result is unconditional.

C. Classification of Super-Critical Solutions (Theorem 1.5, Part c)

For solutions with $E(u) = E(W)$ and $\|u_0\|_{\dot{H}^1_a} > \|W\|_{\dot{H}^1_a}$ :

Radial Case: If $u$ is radially symmetric and in $L^2$ , it blows up in finite time in both forward and backward directions, except if it coincides with the unstable manifold solution $W^+$ (up to symmetries).
Dimension 5 Exception: In 5D, $W^+ \in L^2$ , so solutions on the unstable manifold do not blow up. In 3D and 4D, $W^+ \notin L^2$ , so all radial solutions with higher kinetic energy blow up.

4. Significance

Handling Singular Potentials: This work extends the seminal classification of threshold solutions (originally by Merle and Duyckaerts for the standard NLS) to the case with a singular inverse square potential. The potential breaks translation symmetry and introduces singularities, requiring new spectral and analytical tools.
Resolution of the Threshold Dynamics: It provides a complete picture of the dynamics on the energy surface of the ground state, identifying the ground state as a "saddle point" separating scattering solutions from blow-up solutions.
Conditional vs. Unconditional Results: The paper highlights the specific difficulties in 3D non-radial cases (lack of compactness) and provides a rigorous conditional classification, while offering unconditional results in higher dimensions and radial cases.
Technical Innovation: The development of Strichartz estimates for the linearized operator with singular coefficients and the construction of invariant manifolds in the presence of a non-perturbative potential are significant technical advances.

5. Conclusion

The paper successfully characterizes the dynamics of energy-critical NLS with an inverse square potential at the threshold energy. It establishes that the ground state acts as a separator: solutions with lower kinetic energy either scatter or converge to the ground state (via the stable manifold), while solutions with higher kinetic energy generally blow up, with the only exceptions being the unstable manifold solutions. The results rely heavily on a delicate interplay between spectral analysis of the singular operator, modulation theory, and global virial estimates.