{
  "id": "1711.03949",
  "version": "v1",
  "published": "2017-11-10T18:24:41.000Z",
  "updated": "2017-11-10T18:24:41.000Z",
  "title": "Positivity preserving DG schemes for a Boltzmann - Poisson model of electrons in semiconductors in curvilinear momentum coordinates",
  "authors": [
    "José A. Morales Escalante",
    "Irene M. Gamba",
    "Eirik Endeve",
    "Cory Hauck"
  ],
  "comment": "Preprint",
  "categories": [
    "math.NA"
  ],
  "abstract": "The work presented in this paper is related to the development of positivity preserving Discontinuous Galerkin (DG) methods for Boltzmann - Poisson (BP) computational models of electronic transport in semiconductors. We pose the Boltzmann Equation for electron transport in curvilinear coordinates for the momentum. We consider the 1D diode problem with azimuthal symmetry, which is a 3D plus time problem. We choose for this problem the spherical coordinate system $\\vec{p}(|\\vec{p}|,\\mu=cos\\theta,\\varphi)$, slightly different to the choice in previous DG solvers for BP, because its DG formulation gives simpler integrals involving just piecewise polynomial functions for both transport and collision terms. Applying the strategy of Zhang \\& Shu, \\cite{ZhangShu1}, \\cite{ZhangShu2}, Cheng, Gamba, Proft, \\cite{CGP}, and Endeve et al. \\cite{EECHXM-JCP}, we treat the collision operator as a source term, and find convex combinations of the transport and collision terms which guarantee the positivity of the cell average of our numerical probability density function at the next time step. The positivity of the numerical solution to the pdf in the whole domain is guaranteed by applying the limiters in \\cite{ZhangShu1}, \\cite{ZhangShu2} that preserve the cell average but modify the slope of the piecewise linear solutions in order to make the function non - negative. In addition of the proofs of positivity preservation in the DG scheme, we prove the stability of the semi-discrete DG scheme under an entropy norm, using the dissipative properties of our collisional operator given by its entropy inequalities. The entropy inequality we use depends on an exponential of the Hamiltonian rather than the Maxwellian associated just to the kinetic energy.",
  "revisions": [
    {
      "version": "v1",
      "updated": "2017-11-10T18:24:41.000Z"
    }
  ],
  "analyses": {
    "keywords": [
      "positivity preserving dg schemes",
      "curvilinear momentum coordinates",
      "poisson model",
      "semiconductors",
      "cell average"
    ],
    "note": {
      "typesetting": "TeX",
      "pages": 0,
      "language": "en",
      "license": "arXiv",
      "status": "editable"
    }
  }
}