Weak solutions for a hyperbolic system with unilateral constraint and mass loss. (Q1412657): Difference between revisions

The model considered in this paper reads as the Euler equations with source term \[ \partial_t\rho+ \partial_x(\rho u)= 0,\qquad \partial_t\rho+ \partial_x(\rho u^2+ \kappa\rho^\gamma)= Qu, \] in which \(\kappa\) and \(\gamma\) are parameters such that \(1<\gamma\leq 3\), \(\kappa\geq 0\), coupled with the constraints \[ 0\leq\rho\leq 1,\qquad Q\leq0,\qquad(1-\rho)Q= 0. \] A very similar model was studied by the first author in [Math. Models Methods Appl. Sci. 12, No. 2, 249--272 (2002; Zbl 1027.35079)] in the case \(\kappa=0\) (pressureless gases). The case \(\kappa>0\) and \(\gamma=2\) corresponds, in particular, to the Saint-Venant equations for shallow water flow, and then the term \(Qu\) on the right-hand side is easily interpreted as a mass loss due to overflow. One of the main difficulties with the model above is that the unknown \(Q\) should be a measure (supported by the level set \(\{\rho=1\}\)), so that the product \(Qu\) is not well defined a priori (for poorly regular \(u\)). To overcome the problem, the authors propose what they call a weak entropy formulation, which consists in: 1) introducing a new unknown \(v\), to be sought for bounded \(Q\) almost everwhere, replacing \(u\) in the product \(Qu\), and 2) adding a family of entropy inequalities \[ \partial_t\eta(\rho,u)+\partial_xG(\rho,u)\leq Q\eta'(1,v)\cdot (1,v). \] They prove the stability of this weak entropy formulation in the sense that uniformly bounded (in suitable spaces) sequences \((\rho_n,u_n,Q_n,v_n)\) of solutions converge weakly to another solution. And they finally prove the actual existence of a solution \((\rho,u,Q,v)\) with \(\rho\in L^\infty(dt;dx)\cap L^\infty(dt;L^1(dx))\), \(u\in L^\infty(dt\,dx)\), \(Q\in {\mathcal M}([0,+\infty)\times {\mathbb R})\) and \(v\in L^\infty(Q)\) for initial data \(\rho^0\in L^1(dx)\), \(0\leq\rho^0\leq 1\), \(u^0\in L^\infty(dx)\). The proofs use different techniques in the case \(\kappa\neq 0\) than in the pressureless case, and more precisely: in the former case, a kinetic (BGK) formulation (inspired form earlier work by the authors [Asymptotic Anal. 31, No. 2, 153--176 (2002; Zbl 1032.76064); Ann. Fac. Sci. Toulouse, VI. Sér., Math. 9, No. 4, 605--630 (2000; Zbl 1006.82023)]) and compensated compactness (for \(\gamma<3\)) or averaging lemma (for \(\gamma=3\)); in the latter case, sticky blocks dynamics (in close relation with the paper already quoted above [Math. Models Methods Appl. Sci. 12, No. 2, 249--272 (2002; Zbl 1027.35079)]).