The treatment of Dirichlet boundary conditions for conservation laws in the context of ADER schemes is presented. The present strategy allows to carry out the reconstruction procedure for cells near to the extremes of the computational domain, by using ghost cells. Additionally, it provides vector states at boundaries, which allow to build Riemann problem, as usual, to incorporate correctly the influence of boundaries.

In general, it is assumed that information at boundary is known in terms of prescribed functions. However, if that information is not prescribed or it is just partially available, as for outflows. Then, it is presented a strategy to provide a such information.

Ghost cells are computed as the solution of auxiliary problem, which are defined to evolve in space rather than in time, due to that, these are called here, reverse problems. These are constructed from the conservation laws and the information at boundaries given in terms of prescribed functions. This procedure can be seen as the numerical version of the inverse Lax–Wendroff procedure. But, no Taylor expansions and Cauchy–Kowalewskaya procedure, are required.

Reverse problems are defined at the extremes of the computational domain and they are solved with a low order scheme, which is independent of the numerical scheme used to solve the conservation laws. We have find a criterion to choose the discretization of reverse problem in order to guarantee that both scheme for conservation laws and reverse problems are stable.

Numerical evidence shows that, a second order method for reverse problems allows to achieve up to fifth order of accuracy in both space and time for solving conservation laws. The CPU time of the global scheme is comparable with those using the inverse Lax–Wendroff procedure to deal with boundary conditions.