The Green's function formalism is particularly applicable for problems
where: (1) the domain is nearly infinite or, (2) when the linear
response on selected boundary contours is sought given inhomogeneous
Neumann conditions (or inversely).  Typical problems that fall in this
category are those of a plasma in contact with a resistive wall and the
vacuum boundary condition computations in codes such as PEST and DCON. A
code, GRIN, has been developed with similar capability to VACUUM, the
code by Morrell Chance.  Indeed GRIN borrows most of the ideas and
algorithms pioneered by VACUUM. However, contrary to VACUUM, the code 
GRIN has a multilayer structure giving the user full control over the 
choice of Green's function and geometry while at the same time offering 
canned routines for ease-of-use. Accordingly, GRIN could be applied for
problems outside plasma physics, such as electrostatics for instance.

Although an early version of GRIN was written a few years ago, the code
underwent a number of modifications and improvements as part of a summer 
student project. This talk will focus on these recent improvements,  
including a more accurate computation of the toroidal Laplace Green's 
function using the hypergeometric series and a new matrix/vector class
based on the standard template library (STL), a recent addition to the
C++ standard.  The code is now configured to build on various architectures
using  autoconf, a tool that is widely employed in GNU projects.

Comparisons with VACUUM will be shown.