On a relaxed Cahn–Hilliard tumour growth model with single-well potential and degenerate mobility
2025
Read On a relaxed Cahn–Hilliard tumour growth model with single-well potential and degenerate mobility (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-2512-15426.txt.
Introduction
Cecilia Cavaterra and co authors prove solutions exist for a relaxed tumor growth model in a 2025 math paper. The model pairs a Cahn Hilliard equation, which tracks how tumor and healthy phases separate, with a nutrient equation. Its mobility degenerates and its potential is singular, which makes the math delicate.
The system also includes terms for cell growth, nutrient use, and chemotaxis, the pull of cells toward chemical signals. The authors add an elliptic regularization set by a parameter delta and prove a weak solution exists for the relaxed system. Letting delta go to zero then gives a weak solution for the original tumor model.
Catalog record
| YEAR | 2025 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 153 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-15426
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On a relaxed Cahn–Hilliard tumour growth model with single-well potential and degenerate mobility Abstract We consider a phase-field system modelling solid tumour growth. This system consists of a Cahn–Hilliard equation coupled with a nutrient equation. The former is characterised by a degenerate mobility and a singular potential. Both equations are subject to suitable reaction terms which model proliferation and nutrient consumption. Chemotactic effects are also taken into account. Adding an elliptic regularisation, depending on a relaxation parameter δ > 0 \delta>0 , in the equation for the chemical potential, we prove the existence of a weak solution to an initial and boundary value problem for the relaxed system. Then, we let δ \delta go to zero, and we recover the existence of a weak solution to the original system. Key words: Tumour growth, phase-field models, Cahn–Hilliard equation, degenerate mobility, singular potential, nutrient equation, chemotaxis, existence of a weak solution. AMS (MOS) Subject Classification: 35Q92, 92C15. In memoriam Federica Bubba (1992-2020) Cecilia Cavaterra 1 e-mail: [email protected] Matteo Fornoni 2 e-mail: [email protected] Maurizio Grasselli 3 e-mail: [email protected] Benoît Perthame 4 e-mail: [email protected] 1 Dipartimento di Matematica “F. Enriques”, Università degli Studi di Milano & IMATI–C.N.R., Pavia, Italy 2 Dipartimento di Matematica “F. Enriques”, Università degli Studi di Milano, Italy 3 Dipartimento di Matematica, Politecnico di Milano, Italy 4 Sorbonne Université, CNRS, Université Paris-Cité, Inria, Laboratoire “Jacques-Louis Lions”, Paris, France 1 Introduction The diffuse interface or phase-field approach has recently found many applications in the description of biological phenomena, among which tumour growth is a prime example [ 26 , 9 ] . This is due to a solid physical foundation on continuum mechanics and the theory of mixtures, and to the mathematical properties of the corresponding systems of PDEs, both from the analytical and numerical point of view. In this context, a basic phase-field model in the mechanics of living tissues is the following initial and boundary value problem for a degenerate Cahn–Hilliard equation \displaystyle\partial_{t}\varphi-\divergence(b(\varphi)\nabla\mu)=0\hskip 10.00002pt in Q_{T} , \displaystyle\mu=-\gamma\Delta\varphi+\Psi^{\prime}(\varphi)\hskip 10.00002pt in Q_{T} , \displaystyle\partial_{\boldsymbol{n}}\varphi=b(\varphi)\partial_{\boldsymbol{n}}\mu=0\hskip 10.00002pt on \Sigma_{T} , \displaystyle\varphi(0)=\varphi_{0}\hskip 10.00002pt in \Omega . Here Q_{T}=\Omega\times(0,T) , where T>0 is a given final time, and \Sigma_{T}=\partial\Omega\times(0,T) , where \Omega\subset\mathbb{R}^{d} , d=2,3 , is a given bounded domain. In the above model, \varphi usually represents the volume fraction of one of the two cell types, while \mu is the corresponding chemical potential, namely the first variation of the Ginzburg–Landau-type free energy \mathcal{E}_{GL}(\varphi)=\int_{\Omega}\frac{\gamma}{2}\lvert\nabla\varphi\rvert^{2}+\Psi(\varphi)\,\mathrm{d}x. In this case, \gamma>0 is a positive parameter related to the width of the diffuse interface, that is, the region of width proportional to \sqrt{\gamma} where \varphi varies continuously between the pure phases \varphi=0 and \varphi=1 . Thus, the first term in the functional penalises steep transitions between the two phases, while the potential \Psi(\varphi) is typically of double-well type with equal minima in \varphi=0 and \varphi=1 . However, in solid tumour growth modelling (see, e.g., [ 1 ] and references therein), a more appropriate assumption for \Psi is that it is concave and degenerate near \varphi=0 (short-range attraction) and convex for \varphi close to 1 (long-range repulsion). Moreover, it is usually taken singular at \varphi=1 to model saturation by one phase. A prototypical potential density satisfying these properties is the single-well logarithmic potential of Lennard–Jones type \Psi(s)=-(1-s^{*})\log(1-s)-\frac{s^{3}}{3}-(1-s^{*})\frac{s^{2}}{2}-(1-s^{*})s\hskip 10.00002pt\hbox{for $s\in[0,1)$}, (1.1) for some s^{*}\in(0,1) , corresponding to the position of the single-well. The mobility function b(\cdot) is usually assumed to degenerate at pure phases, such as b(s)=s(1-s)^{2}, (1.2) thus formally enforcing the property that 0\leq\varphi\leq 1 . The combination of a singular potential and a degenerate mobility makes the above equation very challenging to study analytically. Indeed, only very weak existence results are available in the literature [ 1 , 2 ] (see also [ 15 ] for the classical Cahn–Hilliard equation, and the recent [ 22 ] for an anisotropic version). At the same time, numerical analysis is also highly non-trivial, due to the degeneracy of the fourth-order differential operator. For these reasons, in [ 27 ] , the authors introduced a relaxed degenerate Cahn–Hilliard equation combined with a convex splitting (cf. [ 29 ] for a simpler, but similar, model). The relaxation is obtained by adding an elliptic regularisation term to the definition of the chemical potential \mu , multiplied by a small parameter \delta>0 . In this way, the resulting Cahn–Hilliard equation can be seen as a system of two coupled PDEs, a parabolic and an elliptic one respectively. More precisely, the system proposed in [ 27 ] takes the following form: \displaystyle\partial_{t}\varphi-\divergence(b(\varphi)\nabla(\mu+\Psi^{\prime}_{+}(\varphi)))=0\hskip 10.00002pt in Q_{T} , \displaystyle-\delta\Delta\mu+\mu=-\gamma\Delta\varphi+\Psi^{\prime}_{-}\left(\varphi-\frac{\delta}{\gamma}\mu\right)\hskip 10.00002pt in Q_{T} , \displaystyle\partial_{\boldsymbol{n}}\left(\varphi-\frac{\delta}{\gamma}\mu\right)=b(\varphi)\partial_{\boldsymbol{n}}(\mu+\Psi^{\prime}_{+}(\varphi))=0\hskip 10.00002pt on \Sigma_{T} , \displaystyle\varphi(0)=\varphi_{0}\hskip 10.00002pt in \Omega . The convex splitting \Psi=\Psi_{+}+\Psi_{-} allows us to keep the convex and stable part in the parabolic equation for \varphi and the concave and unstable part \Psi_{-} in the (now elliptic) equation for \mu . In the case of the potential introduced above ( 1.1 ) we have, for some given \kappa\in\mathbb{R} , \Psi_{+}(s)=-(1-s^{*})\log(1-s)-\frac{s^{3}}{3}+\kappa,\hskip 10.00002pt\Psi_{-}(s)=-(1-s^{*})\frac{s^{2}}{2}-(1-s^{*})s, (1.3) where \Psi_{+} is convex if s^{*}\in(0,0.7] . We will tacitly assume this condition on s^{*} in everything that follows. In [ 27 ] the authors established the existence of a weak solution for \delta>0 . Then, they proved that a suitable sequence converges to a weak solution to the original system as \delta\to 0 . This provides a validation of the approximation, which turns to be useful also from the numerical viewpoint. Indeed, in [ 4 ] , the authors derive and analyse a structure-preserving numerical scheme for the relaxed system, showing the validity of the proposed relaxation. In this contribution, we extend this relaxation to a more refined phase-field model for tumour growth which accounts for the presence of a nutrient as well as for chemotaxis. We recall that many solid tumour growth models based on the Cahn–Hilliard equation are available in the literature [ 21 ] , however, degenerate mobility is rarely employed, despite being relevant from a physical viewpoint, due to the above-mentioned technical difficulties. Let us introduce our extended model. The phase variable \varphi represents the volume fraction of tumour cells in a tissue, with \varphi=0 being the healthy phase and \varphi=1 the cancerous one. Then, we model the tumour proliferation through the consumption of a key nutrient \sigma , such as oxygen or glucose, whose evolution is ruled by a reaction-diffusion equation. Additionally, by including also some cross-diffusion terms weighted by a non-negative parameter \chi\geq 0 , we introduce chemotactic effects to model the natural movement of tumour cells towards regions with higher nutrient concentrations. Summing up, we study the following system: \displaystyle\partial_{t}\varphi_{\delta}-\divergence(b(\varphi_{\delta})\nabla(\mu_{\delta}+\Psi^{\prime}_{+}(\varphi_{\delta})))=R_{1}(\varphi_{\delta},\mu_{\delta},\sigma_{\delta}), (1.4) \displaystyle-\delta\Delta\mu_{\delta}+\mu_{\delta}=-\gamma\Delta\varphi_{\delta}+\Psi^{\prime}_{-}\left(\varphi_{\delta}-\frac{\delta}{\gamma}\mu_{\delta}\right)-\chi\sigma_{\delta}, (1.5) \displaystyle\partial_{t}\sigma_{\delta}-\Delta\left(\sigma_{\delta}-\chi\left(1-\left(\varphi_{\delta}-\frac{\delta}{\gamma}\mu_{\delta}\right)\right)\right)=R_{2}(\varphi_{\delta},\mu_{\delta},\sigma_{\delta}), (1.6) in Q_{T} , subject to the boundary and initial conditions \displaystyle\partial_{\boldsymbol{n}}\left(\varphi_{\delta}-\frac{\delta}{\gamma}\mu_{\delta}\right)=b(\varphi_{\delta})\partial_{\boldsymbol{n}}(\mu_{\delta}+\Psi^{\prime}_{+}(\varphi_{\delta}))=\partial_{\boldsymbol{n}}\sigma_{\delta}=0\hskip 10.00002pt on \Sigma_{T} , (1.7) \displaystyle\varphi_{\delta}(0)=\varphi_{0},\hskip 10.00002pt\sigma_{\delta}(0)=\sigma_{0}\hskip 10.00002pt in \Omega . (1.8) This system is characterised by the following free energy \begin{split}\mathcal{E}_{\delta}(\varphi_{\delta},\sigma_{\delta})&=\int_{\Omega}\left(\Psi_{+}(\varphi_{\delta})+\frac{\gamma}{2}\left\lvert\nabla\left(\varphi_{\delta}-\frac{\delta}{\gamma}\mu_{\delta}\right)\right\rvert^{2}+\frac{\delta}{2\gamma}\lvert\mu_{\delta}\rvert^{2}+\Psi_{-}\left(\varphi_{\delta}-\frac{\delta}{\gamma}\mu_{\delta}\right)\right)\,\mathrm{d}x\\ &\quad+\int_{\Omega}\left(\frac{1}{2}\lvert\sigma_{\delta}\rvert^{2}+\chi\sigma_{\delta}\left(1-\left(\varphi_{\delta}-\frac{\delta}{\gamma}\mu_{\delta}\right)\right)\right)\,\mathrm{d}x.\end{split} (1.9) We observe that, even if \mu_{\delta} appears in its expression, for any fixed time t\in(0,T) , the free energy actually depends only on \varphi_{\delta} and \sigma_{\delta} , since \mu_{\delta} can be determined as the unique solution to the stationary elliptic equation ( 1.5 ). Moreover, the system also presents an entropy structure due to the presence of the degenerate mobility. Indeed, one can define an entropy density function \eta:(0,1)\to\mathbb{R} as the solution to