Abstract
Several problems, issued from physics, biology or the medical science, lead to parabolic equations set in two subdomains separated by a membrane with selective permeability to specific molecules. The corresponding boundary conditions, describing the flow through the membrane, are compatible with mass conservation and energy dissipation and are called the Kedem–Katchalsky conditions. Additionally, in these models, written as reaction–diffusion systems, the reaction terms have a quadratic behaviour. M. Pierre and his collaborators have developed a complete \(L^1\) theory for reaction–diffusion systems with different diffusions. Here, we adapt this theory to the membrane boundary conditions and prove the existence of weak solutions when the initial data have only \(L^1\) regularity using the truncation method for the nonlinearities. In particular, we establish several estimates as the \(W^{1,1}\) regularity of the solutions. Also, a crucial step is to adapt the fundamental \(L^2\) (space, time) integrability lemma to our situation.
This is a preview of subscription content, access via your institution.
References
 1.
Adams, R. A., and Fournier, J. J. Sobolev spaces. Elsevier, 2003.
 2.
Baras, P., and Pierre, M. Problemes paraboliques semilineaires avec donnees mesures. Applicable Analysis 18, 12 (1984), 111–149.
 3.
Bathory, M., Bulíček, M., and Souček, O. Existence and qualitative theory for nonlinear elliptic systems with a nonlinear interface condition used in electrochemistry. Z. Angew. Math. Phys. 71, 3 (2020), Paper No. 74, 24.
 4.
Bothe, D., and Pierre, M. Quasisteadystate approximation for a reaction–diffusion system with fast intermediate. Journal of Mathematical Analysis and Applications 368, 1 (2010), 120–132.
 5.
Brezis, H. Functional analysis, Sobolev spaces and partial differential equations. Springer Science & Business Media, 2010.
 6.
Brezis, H., Caffarelli, L. A., and Friedman, A. Reinforcement problems for elliptic equations and variational inequalities. Annali di Matematica Pura ed Applicata 123, 1 (1980), 219–246.
 7.
Cañizo, J. A., Desvillettes, L., and Fellner, K. Improved duality estimates and applications to reactiondiffusion equations. Comm. Partial Differential Equations 39, 6 (2014), 1185–1204.
 8.
Calabrò, F., and Zunino, P. Analysis of parabolic problems on partitioned domains with nonlinear conditions at the interface: application to mass transfer through semipermeable membranes. Mathematical models and methods in applied sciences 16, 4 (2006), 479–501.
 9.
Cangiani, A., and Natalini, R. A spatial model of cellular molecular trafficking including active transport along microtubules. Journal of theoretical biology 267, 4 (2010), 614–625.
 10.
Chaplain, M. A., Giverso, C., Lorenzi, T., and Preziosi, L. Derivation and application of effective interface conditions for continuum mechanical models of cell invasion through thin membranes. SIAM Journal on Applied Mathematics 79, 5 (2019), 2011–2031.
 11.
Demengel, F., Demengel, G., and Erné, R. Functional spaces for the theory of elliptic partial differential equations. Springer, 2012.
 12.
Dimitrio, L. Modelling nucleocytoplasmic transport with application to the intracellular dynamics of the tumor suppressor protein p53. PhD thesis, Université Pierre et Marie CurieParis VI and Università degli Studi di Roma La Sapienza, 2012.
 13.
Evans, L. C. Partial differential equations. American Mathematical Soc., 2010.
 14.
Gallinato, O., Colin, T., Saut, O., and Poignard, C. Tumor growth model of ductal carcinoma: from in situ phase to stroma invasion. Journal of Theoretical Biology 429 (2017), 253–266.
 15.
Kedem, O., and Katchalsky, A. A physical interpretation of the phenomenological coefficients of membrane permeability. The Journal of General Physiology 45, 1 (1961), 143–179.
 16.
Laamri, E.H., and Perthame, B. Reactiondiffusion systems with initial data of low regularity. Journal of Differential Equations (in press).
 17.
Laamri, E. H., and Pierre, M. Global existence for reaction–diffusion systems with nonlinear diffusion and control of mass. Annales de l’Institut Henri Poincaré C, Analyse non linéaire 34, 3 (2017), 571–591.
 18.
Li, H., and Wang, X. Effective boundary conditions for the heat equation with interior inclusion. Communications in Mathematical Research 36, 3 (2020), 272–295.
 19.
Li, J., Su, L., Wang, X., and Wang, Y. BulkSurface Coupling: Derivation of Two Models. Preprint (2019).
 20.
Morrey Jr, C. B. Multiple integrals in the calculus of variations. Springer Science & Business Media, 2009.
 21.
Pierre, M. Global existence in reactiondiffusion systems with control of mass: a survey. Milan Journal of Mathematics 78, 2 (2010), 417–455.
 22.
Pierre, M., and Rolland, G. Global existence for a class of quadratic reactiondiffusion system with nonlinear diffusion and \(l^1\) initial data. Nonlinear Analysis 138 (2016), 369–387.
 23.
Quarteroni, A., Veneziani, A., and Zunino, P. Mathematical and numerical modeling of solute dynamics in blood flow and arterial walls. SIAM Journal on Numerical Analysis 39, 5 (2002), 1488–1511.
 24.
Quittner, P., and Souplet, P. Superlinear parabolic problems. Birkhäuser Advanced Texts: Basler Lehrbücher. [Birkhäuser Advanced Texts: Basel Textbooks]. Birkhäuser/Springer, Cham, 2019. Blowup, global existence and steady states, Second edition of [MR2346798].
 25.
Serafini, A. Mathematical models for intracellular transport phenomena. PhD thesis, Université degli Studi di Roma La Sapienza, 2007.
Funding
The authors have received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant Agreement No. 740623). The work of G.C. was also partially supported by GNAMPAINdAM.
Author information
Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no competing interests.
Data availability statement
Not applicable.
Code availability
Not applicable
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Regularity
We now analyse in detail regularity in our problem referring to Lemma 1.2 that we have rewritten here below, whereas in the next Appendix, we discuss about compactness. We extend previous results for reaction–diffusion systems without membrane [2, 4, 16, 17, 21] and we refer to [24] for the general theory of parabolic equations. We also refer to [17] for a regularity lemma.
Lemma A.1
(A priori bounds). We consider w solution of the following problem in dimension \(d\ge 2\)
with \(f\in L^1(Q_T)\) and \(w_0\in L^1(\Omega )\). Then,

\(w \in L^{\beta }\big (0,T;W^{1,\beta }(\Omega )\big ), \; \forall \beta \in \left[ 1,\frac{d}{d1}\right) \) and \((1+w)^\alpha \in L^2\big (0,T;H^1(\Omega )\big )\; \text{ for } \alpha \in \left[ 0,\frac{1}{2}\right) \).

The mapping \((w_0,f)\longmapsto w\) is compact from \(L^1(\Omega )\times L^1(Q_T)\) into \(L^1\big (0,T;L^{\gamma _1} (\Omega )\big )\), for all \(\gamma _1<\frac{d}{d2}\) and \(L^{\gamma _2}(Q_T)\) for all \({\gamma _2} < \frac{2+d}{d}\).

The trace mapping \((w_0,f)\longmapsto Tr_\Gamma (w)\in L^\beta \big (0,T;L^\beta (\Gamma )\big ),\; \beta \in \left[ 1,\frac{d}{d1}\right) \) is also compact.
Notice that we do not use the information \(w\in L^2(Q_T)\) here but \(w\in L^\infty (0,T; L^1(\Omega ))\). That is used in [22] and leads to the exponent \(\beta < \frac{4}{3}\).
Proof
The proof is based on manipulating nonlinear quantities and Sobolev embeddings. We divide it in several steps.
Some \(L^2\) regularity of \(\nabla w\). Multiplying the equation of w in (30) by \(\frac{w}{(1+w^\frac{1}{\mu })^\mu }\) and integrating on \(\Omega \), we obtain three terms which we estimate separately.
We begin with the Laplacian term. Recalling the membrane conditions and applying the Leibniz rule and the divergence theorem, arguing by a regularization and a limit technique, we gain, since \(\frac{w}{(1+w^\frac{1}{\mu })^\mu }\) is an increasing function,
We analyse now the reaction term. We remark that \(0\le \frac{w}{(1+w^\frac{1}{\mu })^\mu }\le 1\) and, using that \(f\in L^1(Q_T)\), we conclude
Next, for the time derivative, we define the antiderivative \(0 \le \psi _\mu (w)= \int _0^w \frac{v\, dv}{(1+v ^\frac{1}{\mu })^{\mu }} \le w\), then
Therefore, combining the previous equality and inequalities, we find
At this point, we can integrate in time and obtain
Since, for all \(\mu > 1\) there is a \(C_\mu \) such that
we conclude that
And thus, there is a constant \(C_\alpha \) which also depends on \(\Vert w_0\Vert _{L^1(\Omega )} + \Vert f\Vert _{L^1(Q_T)}\) such that
Integrability of w. The Sobolev embedding (see Appendix C) gives
which is only useful when \(\alpha 2^* >1\), i.e. \( \frac{d2}{2d} <\alpha \). Then, we can interpolate between \(L^1\) and \(L^{\alpha 2^*}\) and find
We may choose \(\frac{1\theta }{2\alpha }=1\), and, recalling that \(\alpha < \frac{1}{2}\), we find the integrability
We may also choose \(\frac{\gamma (1\theta )}{2\alpha }=1\), \(\alpha < \frac{1}{2}\) and find the integrability
Regularity of \(\nabla w\). On the other hand, Hölder inequality gives
with
We can choose \(\eta p = \gamma _1\) from above, which requires \(\eta \left( \frac{1}{2(1\alpha )} + \frac{1}{\gamma _1} \right) = 1\), \(\beta = \frac{\eta }{1\alpha }= \frac{2 \gamma _1}{\gamma _1+2(1\alpha )}\) and we find, thanks to the estimate (31),
This concludes the proof of the gradient estimate. Moreover, considering that \(\beta <\gamma _2\), thanks to Sobolev embeddings, we can infer that \(w \in L^\beta (0,T;L^\beta (\Omega ))\).
The trace. The regularity of the trace derives from its continuity property [5] (p. 315), i.e.
Compactness
In order to conclude the proof of Lemma A.1, it remains to adapt compactness arguments to the case of the membrane problem. A proof based on a dual approach, see [2, 4], could be used. We rather go to a direct proof.
Compactness in space. It can be obtained using the Rellich–Kondrachov theorem [1], since we know the approximate family is bounded in the spaces \(W^{1,\beta }(\Omega ^\lambda )\), \(\lambda =1, \; 2\) which are compactly embedded in \(L^{\gamma _1}(\Omega ^\lambda )\), with \(\gamma _1<\frac{d}{d2}\).
Compactness in time. We use the Fréchet–Kolmogorov criteria, see [5] for instance. Let \(\varphi (x)\) be a nonnegative, radially symmetric, \(C^\infty _c(\mathbb {R}^d)\) standard mollifier with mass 1. We define the family \((\varphi _\delta )_{\delta >0}\) by
Moreover, we have
and it holds ([13], p. 273) that for any function \(g\in W^{1,p}(\Omega )\),
About the derivative of order k of \(\varphi _\delta \), we know that
Proof
To complete the proof of time compactness, we shall prove that, as \(h \rightarrow 0\),
By comparison with the mollified versions, the triangular equality yields
Here, \(\delta \) depends on h (to be specified later on) and converges to zero. It suffices to prove that each integral converges to zero as \(h\rightarrow 0\).
First term. We analyse the first term in the righthand side. It holds that
thanks to w regularity and to (36), which proves that it converges to zero as \(h\rightarrow 0\).
Second term. For the second integral, we can proceed as for the first one obtaining
Third term. Remembering (30), the last term can be written as
after exchanging derivatives in the convolution. From (35) we deduce
Finally, thanks to (34) and (37), we obtain choosing \(\delta =h^{1/4}\)
and (38) follows combining this estimate with (39) and (40). \(\square \)
Applying the Fréchet–Kolmogorov theorem [5], we conclude that the set of functions \(w \in L^1(Q_T)\) under consideration is compact in \(L^1(Q_T)\). Consequently, we claim compactness in \(L^1\big (0,T; L^{\gamma _1}(\Omega )\big )\) with \({\gamma _1}< \frac{d}{d2}\) and in \(L^{\gamma _2}(Q_T)\) with \({\gamma _2}< \frac{2+d}{d} \). In fact, since we have \(L^1\)convergence of \(L^p\)functions, we deduce convergence in the space \(L^q\), for \(q<p\).
Compactness of traces in \(L^\beta \big (0,T;L^\beta (\Gamma )\big )\). Space compactness can be deduced, in each \(\Omega ^\lambda \), from trace continuity and a compactness result for the boundary ([11], p. 167) such that \(W^{1\frac{1}{\beta },\beta }(\Gamma ) \subset \subset L^{\beta }(\Gamma )\). Time compactness is again achieved through the Fréchet–Kolmogorov theorem. Following the same proof as before and W changing the order of the time integrals, we need to recall Kedem–Katchalsky membrane conditions from which we can infer that \(\partial _tTr_\Gamma (w) \in L^1(0,T;L^1(\Gamma ))\) and so we can conclude the proof. \(\square \)
Sobolev and Poincaré inequalities with membrane
For completeness, we explain why the Sobolev embeddings can be extended to the membrane problem, leading to (31) and (32). More precisely, we explain how to arrive to
There are two difficulties. First, the boundary condition is not Dirichlet everywhere. Second, we are dealing with a singular domain \(\Omega \) and so we cannot use directly the Sobolev or Poincaré inequalities in \(\Omega \), but only some easy generalizations that we explain now.
We are going to prove the
Theorem C.1
(Gagliardo–Nirenberg–Sobolev inequality with membrane). We consider the bounded domain \(\Omega = \Omega ^1\, \cup \,\Omega ^2 \subset \mathbb {R}^d, \; d\ge 2\), with piecewise \(C^1\) subdomains \(\Omega ^1\) and \(\Omega ^2\) and a \(C^1\) membrane \(\Gamma =\partial \Omega ^1\, \cap \,\partial \Omega ^2\) which decomposes \(\Omega \) in the two parts. We take the function \(v=(v^1,v^2) \in {\mathbf{H}^\mathbf{1}}\) (see Definition 1.3), then, for \(\lambda =1,2\),
and consequently
The reason why we want to prove this theorem is that the domain \(\Omega \) described above is not enough regular to use the usual Gagliardo–Nirenberg–Sobolev inequality ([5], p. 284). Consequently, we need to build smoother domains containing each \(\Omega ^i\) in which we can apply known results and then, with a restriction to \(\Omega \), we can find (41) and (42). The construction is made considering an extension of \(\Gamma \) and a domain with the same internal structure as \(\Omega \) such that it contains \(\Omega \) and each extension of the \(\Omega ^i\) is of class \(C^1\).
We first recall the standard Sobolev inequality ([5], p. 284) in a bounded open set.
Theorem C.2
(Sobolev embedding). Let Q be a bounded open subset of class \(C^1\) in \(\mathbb {R}^d\). There is a constant \( C_Q\) such that for all \(v\in H^1(Q),\) we have
Proof
We recall how to prove Theorem C.2 departing from the case of the full space. We use the regularity of the domain which assures us the existence of a linear and continuous extension operator \(T: H^1(Q) \rightarrow H^1(R^d)\), which is also the extension from \(L^2(Q)\) into \(L^2(\mathbb {R}^d)\) ([5], p. 272). So, we obtain that:
Moreover, for construction (see the proof of the extension theorem [5], p. 272), this operator is in \(H^1_0(R^d)\). Consequently, using a corollary of the Sobolev inequality ([13], p. 265), we get that
We proceed with some estimates due to the application of (43), (44), (45). First of all, we deduce
Since \(T(v)\in L^{2^*}(\mathbb {R}^d)\) and \(T(v)=v\) on Q, we get \(v\in L^{2^*}(Q)\) and
The proof of Theorem C.2 is complete. \(\square \)
Since we do not impose Dirichlet conditions on the full boundary, we need the following generalized Poincaré inequality ([20] p. 82).
Theorem C.3
(Poincaré inequality). Suppose Q a bounded and connected open subset of \(\mathbb {R}^d\) of class \(C^1\) and consider a portion of its boundary \(\Sigma _0 \subset \partial Q \) such that \(\Sigma _0>0\). Then, there exists a constant \(C(Q,\Sigma _0)\) such that
Proof
If the statement is not true, we can find a sequence \({v_n}\) such that each \(v_n\in H^1(Q)\) and
On account of the homogeneity (normalizing), we may assume that \(\Vert v_n\Vert _{L^2(Q)}=1\), for each n. So, we infer that
which implies that
Therefore, \(\nabla v_n \rightarrow 0\) in \(L^2(Q)\). Moreover, \(v_n\) is bounded in \(H^1(Q)\), so, up to a subsequence, it converges weakly in \(H^1(Q)\) to some v. So, \(\nabla v_n \rightharpoonup \nabla v\), that means \(\nabla v=0\). This shows that v is a constant (since Q is connected). For the continuity of the trace operator and (47), we deduce
and so \(v=0\).
At the same time, thanks to the Rellich–Kondrachov compactness theorem [1, 5, 13], up to a subsequence, \(v_n\) converges strongly in \(L^2(Q)\) to \(v=0\). Hence, since \(\Vert v_n\Vert _{L^2(Q)}=1\), we arrive to a contradiction. \(\square \)
At this point we are able to give the proof of Theorem C.1.
Proof
We apply Theorems C.2 and C.3. First of all, we consider the extension of \(\Gamma \) into the space \(\mathbb {R}^d\) such that now \(\Gamma \) separates the space into two pieces \(P^\lambda \) with \(\lambda =1,2\). Since we have Dirichlet boundary conditions on \(\Gamma ^\lambda \), we can extend the function to zero in the whole \(P^\lambda \). So now, considering \(Q^\lambda \) a domain of class \(C^1\) such that \(\Omega ^\lambda \subset Q^\lambda \subset P^\lambda \) and for \(\lambda , \sigma =1,2\), \(\;Q^\lambda \cap P^\sigma \) is a portion of \(\Gamma \), we can apply Theorems C.2 and C.3 to
This proves Theorem C.1 in \(Q^\lambda \) and, so, in \(\Omega ^\lambda \). \(\square \)
Rights and permissions
About this article
Cite this article
Ciavolella, G., Perthame, B. Existence of a global weak solution for a reaction–diffusion problem with membrane conditions. J. Evol. Equ. (2020). https://doi.org/10.1007/s00028020006337
Accepted:
Published:
Keywords
 Kedem–Katchalsky conditions
 Membrane boundary conditions
 Reaction–diffusion equations
 Mathematical biology
Mathematics Subject Classification
 35K57
 35D30
 35Q92