Wavelet formulation of the polarizable continuum model. II. Use of piecewise bilinear boundary elements
Permanent lenke
https://hdl.handle.net/10037/8556Dato
2015-07-27Type
Journal articleTidsskriftartikkel
Peer reviewed
Forfatter
Bugeanu, Monica; Di Remigio, Roberto; Mozgawa, Krzysztof; Reine, Simen Sommerfelt; Harbrecht, Helmut; Frediani, LucaSammendrag
The simplicity of dielectric continuum models has made them a standard tool in almost any Quantum
Chemistry (QC) package. Despite being intuitive from a physical point of view, the actual electrostatic
problem at the cavity boundary is challenging: the underlying boundary integral equations depend on
singular, long-range operators. The parametrization of the cavity boundary should be molecular-shaped,
smooth and differentiable. Even the most advanced implementations, based on the integral equation
formulation (IEF) of the polarizable continuum model (PCM), generally lead to working equations which
do not guarantee convergence to the exact solution and/or might become numerically unstable in the
limit of large refinement of the molecular cavity (small tesserae). This is because they generally make
use of a surface parametrization with cusps (interlocking spheres) and employ collocation methods for
the discretization (point charges). Wavelets on a smooth cavity are an attractive alternative to consider:
for the operators involved, they lead to highly sparse matrices and precise error control. Moreover, by
making use of a bilinear basis for the representation of operators and functions on the cavity boundary,
all equations can be differentiated to enable the computation of geometrical derivatives. In this
contribution, we present our implementation of the IEFPCM with bilinear wavelets on a smooth cavity
boundary. The implementation has been carried out in our module PCMSolver and interfaced with
LSDalton, demonstrating the accuracy of the method both for the electrostatic solvation energy and for
linear response properties. In addition, the implementation in a module makes our framework readily
available to any QC software with minimal effort.