Forward and inverse problems of a semilinear transport equation
Kui Ren, Yimin Zhong · 2025
Read Forward and inverse problems of a semilinear transport equation by Kui Ren, Yimin Zhong (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-06183.txt.
Introduction
Light traveling through tissue obeys a transport equation whose absorption can depend on the light itself, and Kui Ren and Yimin Zhong study that semilinear model in a 2025 paper. Their forward results give well-posedness for general boundary data, improving earlier theories that needed small data. The setting is motivated by photoacoustic imaging of multiphoton absorption in mixed media.
The inverse problem asks to rebuild the nonlinear absorption coefficient from internal measurements. The authors prove stability for the reconstruction and unify the theory across diffusion and transport regimes with a weighted norm that discounts the boundary region. Together the results firm up the mathematics behind imaging methods that infer tissue properties from light.
Catalog record
| CREATOR | Kui Ren, Yimin Zhong |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 61 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-06183
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Forward and inverse problems of a semilinear transport equation Kui Ren, Yimin Zhong Abstract We study forward and inverse problems for a semilinear radiative transport model where the absorption coefficient depends on the angular average of the transport solution. Our first result is the well-posedness theory for the transport model with general boundary data, which significantly improves previous theories for small boundary data. For the inverse problem of reconstructing the nonlinear absorption coefficient from internal data, we develop stability results for the reconstructions and unify an L 1 L^{1} stability theory for both the diffusion and transport regimes by introducing a weighted norm that penalizes the contribution from the boundary region. The problems studied here are motivated by applications such as photoacoustic imaging of multi-photon absorption of heterogeneous media. {psinputs} Key words. semilinear radiative transport, inverse problems, diffusion approximation, fixed-point theorem, spectral analysis, photoacoustic imaging, multi-photon absorption 1 Introduction The objective of this work is to study forward and inverse problems for a radiative transport model with a nonlinear absorption mechanism, for applications in imaging modalities such as quantitative photoacoustic tomography. To introduce the mathematical model, let \Omega\subset\mathbb{R}^{d} ( d\geq 2 ) be a convex domain with smooth boundary \partial\Omega , and denote \mathbb{S}^{d-1} the unit sphere in \mathbb{R}^{d} , X:=\Omega\times\mathbb{S}^{d-1} , and \Gamma_{\pm}=\{(x,v)\in\partial\Omega\times\mathbb{S}^{d-1}\mid\pm n(x)\cdot v>0\} , where n(x) is the unit outer normal vector at x\in\partial\Omega . We consider the following transport equation for the nonlinear absorption process [ 18 , 35 , 39 ] : \displaystyle v\cdot\nabla u(x,v)+(\Sigma_{a}(|\langle{u}\rangle|)+\Sigma_{s}(x))u(x,v) \displaystyle=\Sigma_{s}(x)\mathcal{K}u(x,v)\quad \displaystyle\text{ in }X, (1) \displaystyle u(x,v) \displaystyle=f_{-}(x,v)\quad \displaystyle\text{ on }\Gamma_{-}, where \Sigma_{a}:L^{\infty}_{+}(\Omega)\mapsto L^{\infty}_{+}(\Omega) is a nonlinear functional and \Sigma_{s}(x)\in L^{\infty}_{+}(\Omega) denote the nonlinear absorption operator and the scattering coefficients, respectively. \langle{u}\rangle is the averaged density of u(x,v) over the angular variable v\in\mathbb{S}^{d-1} , \langle{u}\rangle:=\int_{\mathbb{S}^{d-1}}u(x,v)dv with dv being the normalized surface measure on \mathbb{S}^{d-1} . The absolute value in |\langle{u}\rangle| is to ensure well-posedness. f_{-}(x,v)\geq 0 is the incoming illumination source function on \Gamma_{-} . The linear operator \mathcal{K} is defined by \mathcal{K}u(x,v)=\int_{\mathbb{S}^{d-1}}p(x,v,v^{\prime})u(x,v^{\prime})dv^{\prime}, where p(x,v^{\prime},v)\geq 0 is the scattering phase function at point x such that \int_{\mathbb{S}^{d-1}}p(x,v,v^{\prime})dv^{\prime}=\int_{\mathbb{S}^{d-1}}p(x,v,v^{\prime})dv=1. Radiative transport equations similar to ( 1 ) are useful in modeling the propagation of particles in heterogeneous media where the absorption strength of the media depends on the local density of the particles. Such a situation appears in the propagation of near-infrared photons in biological tissues with multiphoton absorption [ 3 , 28 , 36 , 47 ] . The first goal of this work is to improve existing well-posedness theory for the radiative transport model ( 1 ). Existing theory in [ 35 , 39 ] assumes that the boundary datum f_{-} is sufficiently small. This assumption significantly limits the applicability of the theory, as nonlinear absorption mostly happens when the density of the particles is very high. We will remove this smallness assumption in our theory in Section 2 . In applications such as quantitative photoacoustic imaging of two-photon absorption [ 9 , 30 , 44 , 45 , 46 ] , one is interested in reconstructing the absorption property of ( 1 ) from data related to its solutions. The second goal of this work is to analyze the problem of reconstructing a generic multi-photon absorption model where \Sigma_{a} is a homogeneous polynomial of the local density \langle{u}\rangle from the total absorbed energy data. We will develop some stability theories for the inverse problem in Section 3 , and then unify the stability theories of the diffusion regime and the transport regime in Section 4 . 2 The forward problem We start with the well-posedness of the transport equation ( 1 ). To distinguish from the existing results in [ 35 , 39 ] , we emphasize that our well-posedness theory does not make further assumptions about the smallness of the illumination source f_{-} . This is a significant improvement over existing results. Let us denote X=\Omega\times\mathbb{S}^{d-1} . We introduce the space \mathcal{H}^{p}(X) \mathcal{H}^{p}(X)=\{f(x,v)\mid f\in L^{p}(X),\text{and }v\cdot\nabla f\in L^{p}(X)\}, with L^{p}(X) the standard L^{p} space of functions on X . We define L^{p}(\Gamma_{-}) to be the trace space of \mathcal{H}^{p} on \Gamma_{-} that is equipped with the norm \|f\|_{L^{p}(\Gamma_{-})} : \|f\|_{L^{p}(\Gamma_{-})}:=\left(\int_{\partial\Omega}\int_{v\cdot n(x)<0}|n(x)\cdot v||f(x,v)|^{p}dvds\right)^{1/p}. In the rest of the work, we assume the coefficients and the source function f_{-} satisfy the following conditions: ( \mathcal{A} ). (Boundedness) There exist constants \underline{\Sigma}_{a} , \overline{\Sigma}_{s} , \underline{f} , \overline{f} and a non-decreasing function g\in L^{\infty}(\mathbb{R}_{+}) that \displaystyle 0<\underline{\Sigma}_{a}\leq\Sigma_{a}(m)\leq g(\|m\|_{\infty}),\quad \displaystyle\forall(x,m)\in\Omega\times L^{\infty}_{+}(\Omega), \displaystyle 0\leq\Sigma_{s}(x)\leq\overline{\Sigma}_{s},\quad \displaystyle\forall x\in\Omega, and 0<\underline{f}\leq f_{-}(x,v)\leq\overline{f} on \Gamma_{-} . ( \mathcal{B} ). (Continuity) \Sigma_{a}(\cdot) is continuous under L^{2} metric, that is, \|m_{1}-m_{2}\|_{L^{2}(\Omega)}\to 0 implies \|\Sigma_{a}(m_{1})-\Sigma_{a}(m_{2})\|_{L^{2}(\Omega)}\to 0 . ( \mathcal{C} ). (Positive-definiteness) The nonlinear absorption coefficient \Sigma_{a}(\cdot) is continuously Fréchet differentiable with the Fréchet derivative, denoted by \Sigma_{a}^{\prime} , satisfying \int_{\Omega}\Sigma^{\prime}_{a}[f](x)\cdot f(x)dx\geq 0. The equal sign holds if and only if f\equiv 0 . Let us mention that, in the multi-photon absorption model we will study in Section 3 , the absorption coefficient \Sigma_{a}(|\langle{u}\rangle|) is given in a polynomial form, such as in ( 7 ) and ( 8 ). It is straightforward to verify that such absorption models satisfy the conditions (\mathcal{A}) - (\mathcal{C}) . 2.1 Existence The existence of a solution in \mathcal{H}^{\infty}(X) to ( 1 ) can be proved through a similar idea from [ 35 ] . The following simple estimate is based on linear transport theory [ 8 ] . Lemma 2.1 . Suppose the condition (\mathcal{A}) is satisfied, and let m\in L^{\infty}_{+}(\Omega) be such that 0\leq m\leq\|f_{-}\|_{L^{\infty}(\Gamma_{-})} , then the solution u\in\mathcal{H}^{\infty}(X) to \displaystyle v\cdot\nabla u(x,v)+(\Sigma_{a}(m)+\Sigma_{s}(x))u(x,v) \displaystyle=\Sigma_{s}(x)\mathcal{K}u(x,v)\quad \displaystyle\text{ in }X, (2) \displaystyle u(x,v) \displaystyle=f_{-}(x,v)\quad \displaystyle\text{ on }\Gamma_{-}. satisfies 0<\mathsf{c}\leq u\leq\mathsf{C}, where the constants \mathsf{C}=\|f_{-}\|_{L^{\infty}(\Gamma_{-})} and \mathsf{c}=\underline{f}e^{-\operatorname{diam}(\Omega)g(\mathsf{C})} . Proof. Suppose u\in\mathcal{H}^{\infty}(X) is the solution to ( 2 ), then \mathsf{C}\geq u\geq 0 from the classical linear transport theory [ 8 ] . Let \widetilde{u} be the solution to the following linear transport equation \displaystyle v\cdot\nabla\widetilde{u}(x,v)+\Sigma_{a}(m)\widetilde{u} \displaystyle=0\quad \displaystyle\text{ in }X, \displaystyle\widetilde{u}(x,v) \displaystyle=f_{-}(x,v)\quad \displaystyle\text{ on }\Gamma_{-}\,. Then, from the analytic expression of \widetilde{u} , we can verify that \widetilde{u}\geq\underline{f}e^{-\operatorname{diam}(\Omega)g(\|m\|_{\infty})}\geq\mathsf{c} . We also verify that \phi=u-\widetilde{u} satisfies \displaystyle v\cdot\nabla\phi(x,v)+(\Sigma_{a}(m)+\Sigma_{s}(x))\phi(x,v) \displaystyle=\Sigma_{s}(x)\mathcal{K}u\quad \displaystyle\text{ in }X, \displaystyle\phi(x,v) \displaystyle=0\quad \displaystyle\text{ on }\Gamma_{-}. Thus \phi\geq 0 , which implies u\geq\widetilde{u}\geq\mathsf{c} . ∎ Theorem 2.2 . Suppose the conditions (\mathcal{A}) and (\mathcal{B}) are satisfied, then there exists a solution u\in\mathcal{H}^{\infty} to ( 1 ). Proof. We define the mapping \mathcal{S}:L^{\infty}_{+}(\Omega)\mapsto L^{\infty}_{+}(\Omega) through the relation \mathcal{S}(m)=\langle{u}\rangle, (3) where u\in\mathcal{H}^{\infty}(X) denotes the unique solution to \displaystyle v\cdot\nabla u(x,v)+(\Sigma_{a}(m)+\Sigma_{s}(x))u \displaystyle=\Sigma_{s}(x)\mathcal{K}u(x,v)\quad \displaystyle\text{ in }X, (4) \displaystyle u(x,v) \displaystyle=f_{-}(x,v)\quad \displaystyle\text{ on }\Gamma_{-}. When m\in L^{\infty}_{+}(\Omega) , from the condition (\mathcal{A}) we have that \Sigma_{a}(m)\geq\underline{\Sigma}_{a}>0 . Therefore, 0\leq u\leq\|f_{-}\|_{\infty} . Let \mathcal{M} be the set of bounded functions with L^{2}(\Omega) topology: \mathcal{M}=\{m\in L_{+}^{\infty}(\Omega)\mid 0\leq m\leq\|f_{-}\|_{\infty}\} Therefore, the mapping \mathcal{S}(\mathcal{M})\subset\mathcal{M} . It is straightforward to see that \mathcal{M} is convex and closed. We then verify that the mapping \mathcal{S} is continuous on \mathcal{M} . Let m_{1},m_{2}\in\mathcal{M} and denote by u_{i} the solution to ( 4 ) with the absorption coefficient \Sigma_{a}(m_{i}) , i=1,2 . Then, \phi=u_{1}-u_{2} solves \displaystyle v\cdot\nabla\phi+(\Sigma_{a}(m_{1})+\Sigma_{s})\phi \displaystyle=\Sigma_{s}\mathcal{K}\phi(x,v)-(\Sigma_{a}(m_{1})-\Sigma_{a}(m_{2}))u_{2}\quad \displaystyle\text{ in }X, \displaystyle\phi(x,v) \displaystyle=0\quad \displaystyle\text{ on }\Gamma_{-}. Multiply the above equation by \phi and integrate over X , we find that \displaystyle\int_{X}\Sigma_{a}(m_{1})|\phi|^{2}dxdv \displaystyle\leq\int_{\Gamma_{+}}\frac{1}{2}v\cdot n|\phi|^{2}dvdS+\int_{X}\Sigma_{a}(m_{1})|\phi|^{2}dxdv+\int_{X}\Sigma_{s}(x)(|\phi|^{2}-\phi\mathcal{K}\phi)dxdv \displaystyle=-\int_{X}(\Sigma_{a}(m_{1})-\Sigma_{a}(m_{2}))u_{2}\phi dxdv \displaystyle\leq\|\Sigma_{a}(m_{1})-\Sigma_{a}(m_{2})\|_{L^{2}(X)}\|u_{2}\|_{L^{\infty}(X)}\|\phi\|_{L^{2}(X)}. Using condition (\mathcal{B}) ,