[Paper Review] Null controllability for a degenerate population model in divergence form via Carleman estimates

Genni Fragnelli|arXiv (Cornell University)|Dec 29, 2018

Stability and Controllability of Differential Equations21 references4 citations

TL;DR

This paper establishes null controllability for a degenerate population model in divergence form with age, time, and space variables, using Carleman estimates to handle diffusion coefficients degenerating at both endpoints of the spatial domain. The key contribution is proving null controllability for arbitrary final time $ T $, even when $ T < A $, under minimal regularity assumptions on the birth rate $ \beta $, overcoming limitations of prior works that required $ T \geq A $ or stronger smoothness conditions.

ABSTRACT

In this paper we consider a degenerate population equation in divergence form depending on time, on age and on space and we prove a related null controllability result via Carleman estimates.

Motivation & Objective

• To establish null controllability for a degenerate population model in divergence form with time, age, and space variables.
• To address the limitation of prior results requiring $ T \geq A $, which is unrealistic for large maximal ages $ A $.
• To extend previous null controllability results to cases where the diffusion coefficient $ k(x) $ degenerates at both endpoints of the spatial domain $ [0,1] $.
• To relax regularity assumptions on the birth rate $ \beta $, requiring only continuity instead of $ C^2 $ smoothness as in earlier works.
• To provide a streamlined proof using Carleman estimates and cutoff functions, avoiding fixed-point arguments and multi-valued theory.

Proposed method

• Derives Carleman estimates for the adjoint problem of the degenerate population equation with $ k(x) $ degenerating at both $ x=0 $ and $ x=1 $.
• Applies the duality method between the controllability of the forward equation and the observability of the adjoint equation.
• Uses a cutoff function technique to localize the control region and handle degeneracy at both boundaries.
• Employs weighted energy estimates and integration by parts to control the degenerate terms in the Carleman inequality.
• Imposes structural conditions on $ k(x) $, such as $ xk'(x) \leq M_1 k(x) $ and $ (x-1)k'(x) \leq M_2 k(x) $, to ensure the validity of the estimates.
• Relies on the observability inequality derived from the Carleman estimate to deduce null controllability.

Experimental results

Research questions

• RQ1 Can null controllability be achieved for a degenerate population model in divergence form when the final time $ T $ is less than the maximal age $ A $?
• RQ2 What conditions on the degenerate diffusion coefficient $ k(x) $ are sufficient to ensure null controllability via Carleman estimates?
• RQ3 How can the proof be simplified and made more robust by avoiding fixed-point arguments and strong regularity assumptions on $ \beta $?
• RQ4 Is it possible to extend the null controllability result to cases where $ k(x) $ degenerates at both endpoints of the spatial domain?
• RQ5 What is the optimal structure of the weight function in the Carleman estimate to handle degeneracy at both boundaries?

Key findings

• Null controllability is established for the degenerate population model in divergence form even when $ T < A $, removing a restrictive assumption from earlier works.
• The result holds under weak regularity assumptions: $ \beta $ is only continuous, not $ C^2 $, as required in previous studies.
• The proof avoids the Leray-Schauder fixed point theorem and multi-valued analysis, relying solely on Carleman estimates and cutoff functions.
• A new weighted energy estimate is derived, showing $ \int_0^1 \frac{k(x)}{(1-x)^2} w^2(x) dx \leq C \int_0^1 k(x) |w'(x)|^2 dx $, which is optimal for $ \beta = \frac{\theta+1}{2} $.
• The method generalizes to the non-divergence form case, extending [13, Theorem 4.8] with improved precision in the calculation.
• The observability inequality is proven via a refined analysis of the Carleman estimate, with precise control over degenerate terms using the monotonicity of $ k(x)/(1-x)^\theta $.