Technical Summary: Endogenous Feedback in Size-Structured Transport Equations

1. Problem Formulation

The paper investigates a nonlinear size-structured transport equation where the system's dynamics are governed by an endogenous scalar feedback mechanism. The state variable $x(t, l)$ represents the density of individuals structured by size $l \in \Omega = [l_0, l_m]$. The evolution of this density is described by the transport equation:
$$\partial_t x + \partial_l (g(E(t), l)x) = -(\mu(E(t), l) + u(t, l))x$$
where $g$ is the velocity (growth rate) and $\mu$ is the mortality rate. Crucially, both $g$ and $\mu$ depend on a scalar output $E(t)$, defined as the weighted integral of the population density:
$$E(t) = \int_{l_0}^{l_m} \chi(l) x(t, l) \, dl$$
This creates a closed-loop architecture where the state $x$ determines the output $E$, which in turn feeds back into the principal coefficients of the transport operator. The system is subject to an inflow flux (recruitment) boundary condition $g(E(t), l_0)x(t, l_0) = p(t)$ and an initial condition $x(0, l) = x_0(l)$.

The primary mathematical challenge addressed is that the feedback enters the principal coefficient $g$, rendering the system non-semilinear. Standard perturbation frameworks for semilinear equations are insufficient because the transport geometry itself depends on the state. The authors aim to establish well-posedness, analyze stationary equilibria, determine spectral stability, and characterize the adjoint system for optimal harvesting.

2. Methodology

The authors employ a fixed-point approach combined with Volterra integral equation theory and semigroup methods.

• Frozen-Path Operator: The closed-loop problem is reduced to a scalar fixed-point problem $E = \mathcal{K}(E)$. Here, $\mathcal{K}$ is an operator that takes a prescribed feedback path $E$, solves the resulting linear non-autonomous transport equation (the "frozen" problem), and computes the new output.
• Intrinsic Feedback Interval: Unlike approaches that impose external compactness assumptions, the authors derive an intrinsic feedback interval $[0, M_T]$ directly from mass balance principles. This interval is determined by the initial mass and the cumulative inflow, ensuring the feedback remains bounded without artificial constraints.
• Contraction Mapping: Well-posedness is established by proving that the operator $\mathcal{K}$ is a contraction in a Bielecki norm (a weighted $L^\infty$ norm). This relies on $L^1$ and $BV$ (bounded variation) estimates for the frozen transport solutions.
• Linearization and Renewal Equations: For stability analysis, the system is linearized around a stationary equilibrium. The authors demonstrate that the linearized dynamics reduce to a finite-memory scalar renewal equation (a Volterra convolution equation) with a characteristic function $\mathcal{E}(\lambda)$.
• Adjoint Analysis: The stationary optimal harvesting problem is analyzed using an adjoint system. The authors show that the feedback dependence results in a rank-one perturbation of the frozen adjoint operator, allowing for an explicit Sherman–Morrison-type solution formula.

3. Key Contributions and Results

A. Well-Posedness

The paper proves the existence and uniqueness of a nonnegative weak solution for the closed-loop system on any finite time interval $[0, T]$.

• Theorem 4.4: Under standard regularity assumptions on coefficients and data, the map $\mathcal{K}$ is a contraction in the Bielecki norm, guaranteeing a unique fixed point $E_{cl}$.
• Continuation Criterion: The solution can be extended globally in time provided the feedback remains within the intrinsic bounds and the velocity $g$ stays strictly positive.

B. Stationary Closure and Resonance

For stationary controls and inflow, the system reduces to a scalar closure equation $E = \Phi(E)$.

• Uniqueness: The authors prove that if the derivative of the closure map satisfies $\Phi'(E) < 1$, the equilibrium is unique.
• Fold Bifurcation: The condition $\Phi'(E^*) = 1$ is identified as a nondegenerate fold threshold. At this point, the closure map loses invertibility, leading to a saddle-node bifurcation.
• Decomposition: The derivative $\Phi'(E)$ is decomposed into two competing terms: a residence-time amplification term ($\mathcal{R}$) and a cumulative survival attenuation term ($\mathcal{C}$). This clarifies the algebraic sign of the slope.

C. Spectral Stability

The stability of the equilibrium is determined by the roots of the characteristic equation $\mathcal{E}(\lambda) = 1$, derived from the linearized renewal equation.

• Headline Identity: A central result is the identity $\mathcal{E}(0) = \Phi'(E^*) = B(0)$, where $B(0)$ is the loop gain of the rank-one adjoint system at zero discount.
• Stability Criterion:
  • If $\Phi'(E^*) < 1$, the zero root lies in the left half-plane, and the system is linearly stable (provided no other roots cross the imaginary axis).
  • If $\Phi'(E^*) > 1$, a positive real root exists, implying linear instability.
  • The condition $\Phi'(E^*) = 1$ corresponds to a zero eigenvalue, marking the onset of instability and the fold bifurcation.
• Nonlinear Stability: Under a quadratic remainder estimate (verified under second-order regularity assumptions), the linearized stability implies local exponential asymptotic stability for the nonlinear system.

D. Optimal Harvesting and Adjoint Structure

The paper analyzes the stationary optimal harvesting problem with a discounted yield functional.

• Rank-One Perturbation: The full adjoint equation is shown to be a rank-one perturbation of the frozen adjoint equation. This yields an explicit formula for the adjoint variable $\lambda$:
  $$\lambda = \lambda_{red} + \frac{\langle \sigma, \lambda_{red} \rangle}{1 - B(r)} \psi_r$$
  where $\lambda_{red}$ is the solution to the frozen problem, and the denominator $1 - B(r)$ represents the feedback loop gain.
• Switching Geometry: The optimal control is of "bang-bang" type, determined by the sign of a switching function $S(l) = \pi(l) - \lambda(l)$. The rank-one correction shifts the switching threshold. The authors provide conditions under which the switching structure remains a single threshold and characterize the displacement of this threshold.

4. Significance and Claims

The paper claims to establish a rigorous mathematical framework for size-structured systems with endogenous feedback that enters the principal transport coefficient. Its significance lies in three main areas:

1. Intrinsic Feedback Bounds: The derivation of the feedback interval $[0, M_T]$ from mass conservation eliminates the need for external compactness assumptions, making the well-posedness result more robust and physically grounded.
2. Unified Scalar Margin: The authors identify a single scalar quantity, $1 - \Phi'(E^*)$, that simultaneously governs:
   • The invertibility of the stationary closure map (static bifurcation).
   • The location of the zero characteristic root of the linearized dynamic system (dynamic stability).
   • The solvability of the rank-one adjoint equation (optimal control sensitivity).
     This "closure-resonance" phenomenon links static bifurcation theory directly to dynamic spectral analysis via an explicitly computable scalar.
3. Explicit Adjoint Reduction: By reducing the complex feedback adjoint system to a rank-one perturbation, the paper provides a tractable formula for optimal control problems in structured populations, clarifying how feedback modifies the switching thresholds in harvesting policies.

The authors note that while the paper establishes well-posedness and local stability, the global existence of optimal controls and dynamic synthesis are reserved for future work. The nonlinear stability result is conditional on a quadratic remainder estimate, which is verified under specific second-order regularity assumptions.