Control of the Schrödinger equation in $\mathbb{R}^3$: The critical case

This paper establishes the local null controllability of the energy-critical nonlinear Schrödinger equation in $\mathbb{R}^3$ at the $H^1$ level by first proving well-posedness via Strichartz estimates, demonstrating linear controllability through the Hilbert uniqueness method, and finally extending the result to the nonlinear case using a perturbation argument.

The Gist

Imagine you are trying to steer a massive, invisible cloud of energy through a vast, empty universe. This cloud isn't just floating randomly; it's governed by complex rules of physics (the Schrödinger equation) that make it ripple, spread out, and interact with itself in tricky ways.

This paper is about learning how to steer that cloud to a complete stop (a state of zero energy) using a "remote control" that only works in a specific part of the universe.

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

1. The Problem: The "Critical" Storm

The authors are studying a specific type of energy cloud called the Nonlinear Schrödinger Equation.

• The "Nonlinear" part: Imagine the cloud is made of water. If the waves are small, they just pass through each other. But if they get big, they crash and create new, unpredictable waves. This is the "nonlinear" part—it's chaotic.
• The "Critical" part: This is the most dangerous scenario. It's like a storm where the wind speed and the size of the waves are perfectly balanced. If the wind gets slightly stronger, the storm explodes (blows up). If it gets weaker, the waves die out. The authors are working exactly on this "knife-edge" balance in 3D space.

The Goal: They want to prove that if you start with a small enough cloud of energy, you can apply a force (a control) to make the entire cloud vanish completely at a specific time $T$.

2. The Challenge: The "Whole Space" Difficulty

Usually, scientists study these clouds inside a box (a bounded room). It's easier to control something when you can push on all the walls.

• This paper's twist: They are studying the cloud in infinite space (the whole universe, $\mathbb{R}^3$). There are no walls. You can't just push on the boundary because there isn't one.
• The Solution: They use a "spotlight" control. Imagine a giant, invisible flashlight that only shines on the outer edges of the universe (far away from the center). They prove that even if you can only push on the edges of the cloud, you can still steer the whole thing to zero.

3. The Three-Step Strategy

The authors used a three-step "recipe" to solve this puzzle:

Step A: Proving the Cloud Doesn't Explode (Well-Posedness)

Before you can steer a car, you need to know it won't fall apart the moment you turn the key.

• The Analogy: They used mathematical tools called Strichartz estimates. Think of these as a "stability check." They proved that for small clouds, the chaotic nonlinear waves won't suddenly explode into infinity. The cloud stays manageable and predictable for a while.

Step B: Steering the Simple Version (Linear Control)

First, they ignored the messy "nonlinear" crashing waves and looked at a simplified, calm version of the cloud (the Linear Schrödinger equation).

• The Analogy: They asked, "If this were just a calm ripple, could we stop it?"
• The Method: They used a technique called Hilbert Uniqueness Method (HUM). Imagine you want to stop a moving car. You look at the car's path backward in time. If you can prove that the car's movement leaves a unique "fingerprint" on the road (observability), you can calculate exactly how much force to apply to stop it.
• The Result: They proved that for the calm version, you can indeed stop the cloud using a control on the edges.

Step C: Tackling the Real Chaos (Perturbation)

Now, they brought the chaos back. They knew the calm version was controllable. They argued that since the "real" cloud (with the crashing waves) is just a small disturbance away from the calm version, the same steering trick should work.

• The Analogy: Imagine you know how to steer a bicycle on a flat, empty road. Now, imagine a very light breeze (the nonlinearity). If the breeze is weak enough, you can still steer the bike using the same techniques you used on the flat road. You just make tiny adjustments.
• The Result: They proved that as long as the initial energy of the cloud is small enough, the "breeze" isn't strong enough to break their steering method.

4. Why This Matters

• First of its kind: This is the first time anyone has proven you can control this specific "critical" type of energy equation in 3D space using internal controls.
• No "Magic" Walls Needed: Previous studies often required the control to be everywhere or on a specific boundary. This paper shows that even a control acting on a "thick" ring far away from the center is enough to stop the whole system.
• Future Steps: The authors admit this is just the beginning. They can stop small clouds. The next big challenge is: "Can we stop huge storms?" and "Can we build a feedback loop that automatically keeps the cloud calm forever?"

Summary

Think of the authors as master pilots. They took a plane that was known to be extremely unstable (the critical equation) and flying in an infinite void (3D space). They proved that if the plane isn't too heavy (small initial data), they can use a remote control that only works on the wings' tips to gently guide the entire plane to a perfect, safe landing (zero state).

Technical Summary

1. Problem Statement

The paper addresses the local null controllability of the energy-critical nonlinear Schrödinger equation (C-NLS) in three spatial dimensions ($\mathbb{R}^3$). The system is given by:
$$\begin{cases} i\partial_t u + \Delta u \pm |u|^4 u = f(x, t), & \text{on } \mathbb{R}^3 \times [0, T], \\ u(x, 0) = u_0(x) \in H^1(\mathbb{R}^3), \end{cases}$$
where the nonlinearity $|u|^4 u$ is "energy-critical" because it preserves the scaling invariance of the associated energy functional.

The Control Problem: Given a time $T > 0$ and a small initial state $u_0 \in H^1(\mathbb{R}^3)$, does there exist a control input $f(x, t)$, supported in a specific region (specifically, outside a large ball $B_R(0)$), such that the solution $u$ satisfies the null condition $u(x, T) = 0$?

The authors focus on the case where the control is of the form $f(x, t) = \phi(x)h(x, t)$, where $\phi$ is a smooth cutoff function that is $0$ inside a ball $B_R$ and $1$ outside $B_{R+1}$.

2. Methodology

The proof strategy relies on a combination of well-posedness theory, duality arguments, and perturbation methods. The methodology is structured in three main stages:

A. Well-Posedness via Strichartz Estimates

To handle the critical nonlinearity in the whole space $\mathbb{R}^3$, the authors establish the local well-posedness of the Cauchy problem in the $H^1$ space.

• Tools: They utilize Strichartz estimates for the linear Schrödinger operator in $\mathbb{R}^3$.
• Admissible Pairs: They define $L^2$-admissible and $H^1$-admissible pairs $(q, r)$ to control the mixed space-time norms (e.g., $L^{10}_t L^{10}_x$).
• Fixed Point Argument: Using a contraction mapping principle (Banach Fixed Point Theorem) on a suitable ball in the space $C([0, T]; H^1) \cap L^{10}([0, T]; L^{10})$, they prove the existence and uniqueness of local solutions for small initial data.
• Dissipative Modification: To prove global existence for the linearized control problem, they introduce a modified damping term involving the operator $\Lambda^{-1}$ (an isomorphism between $H^1$ and $H^{-1}$) to ensure energy decay, preventing finite-time blow-up.

B. Linear Controllability via Hilbert Uniqueness Method (HUM)

The authors prove that the associated linear Schrödinger equation is null controllable.

• Duality: By the Hilbert Uniqueness Method (HUM), null controllability of the forward system is equivalent to an observability inequality for the adjoint (backward) system:
  $$i\partial_t v + \Delta v = 0, \quad v(T) = v_0.$$
• Observability Inequality: They aim to prove:
  $$\|v_0\|_{H^{-1}}^2 \leq C \int_0^T \|\phi v(t)\|_{H^{-1}}^2 dt.$$
• Proof Technique: The proof of this inequality involves a contradiction argument combined with:
  1. Unique Continuation Property: Utilizing a Carleman estimate (from prior work by [21]) to show that if a solution vanishes on the control region (where $\phi=1$), it must vanish everywhere.
  2. Compactness-Uniqueness Argument: They construct a sequence of solutions, extract weak limits, and use the unique continuation property to show the limit is zero, leading to a contradiction if the observability inequality were false.
  3. Smoothing Properties: They leverage local smoothing properties of the Schrödinger equation to handle the regularity in $H^{-1}$.

C. Nonlinear Controllability via Perturbation

Once linear controllability is established, the result is extended to the nonlinear system.

• Perturbation Argument: Following the approach of E. Zuazua, the authors define a mapping $B$ that relates the initial data of the adjoint system to the initial data of the nonlinear system.
• Fixed Point: They show that for sufficiently small initial data $u_0$, the operator $B$ is a contraction on a small ball in $H^{-1}$. By the Banach Fixed Point Theorem, there exists a fixed point, which corresponds to a control $h$ that drives the nonlinear system to zero.

3. Key Contributions

1. First Local Null Controllability in Critical Regime: This is the first result establishing local null controllability for the energy-critical NLS ($|u|^4 u$) in the whole space $\mathbb{R}^3$ at the $H^1$ level. Previous results were largely restricted to subcritical cases or bounded domains.
2. Relaxation of Well-Posedness Framework: While critical NLS theory is often developed in the homogeneous space $\dot{H}^1$, the authors demonstrate that the availability of Strichartz estimates in $\mathbb{R}^3$ allows them to work directly in the inhomogeneous space $H^1(\mathbb{R}^3)$. This is a significant technical improvement for control problems.
3. Handling the Whole Space: The work addresses the difficulties of controlling the equation in unbounded domains ($\mathbb{R}^3$) rather than bounded domains or compact manifolds. They show that the geometric control condition (GCC) is naturally satisfied by the choice of the control support (the exterior of a ball), avoiding the need for complex geometric constraints required in bounded domains.
4. Extension to Critical Nonlinearity: The paper bridges the gap between control theory for subcritical NLS and the critical case, which is notoriously difficult due to the potential for blow-up and lack of compactness.

4. Main Results

• Theorem 1.1 (Local Null Controllability): For any $T > 0$, there exists a $\delta > 0$ such that for any initial data $u_0 \in H^1(\mathbb{R}^3)$ with $\|u_0\|_{H^1} \leq \delta$, there exists a control $h \in C([0, T]; H^1(\mathbb{R}^3))$ supported in $\mathbb{R}^3 \setminus B_R(0)$ such that the solution $u$ of the critical NLS satisfies $u(T) = 0$.
• Theorem 1.2 (Linear Controllability): The linear Schrödinger equation with the same control structure is exactly null controllable in $H^1$ for any initial data.

5. Significance and Future Directions

• Theoretical Impact: The results deepen the understanding of the interplay between dispersion, nonlinearity, and control in critical regimes. It confirms that the "critical" nature of the nonlinearity does not prevent local controllability if the initial data is small.
• Open Problems: The authors outline several future research directions:
  • Stabilization: Can a feedback control law be designed to asymptotically stabilize the system to zero?
  • Global Control: Can the result be extended to large data (global controllability), potentially by combining local control with global stabilization?
  • Geometric Control Conditions: Investigating more general control sets (thick sets) that satisfy the observability inequality in $\mathbb{R}^3$.

In summary, this paper provides a rigorous mathematical foundation for controlling the energy-critical Schrödinger equation in 3D, overcoming significant analytical hurdles related to criticality and unbounded domains through advanced functional analysis and control theory techniques.