[en] A - Description of program or function: MORSE-CGA was developed to add the capability of modeling rectangular lattices for nuclear reactor cores or for multi-partitioned structures. It thus enhances the capability of the MORSE code system. The MORSE code is a multipurpose neutron and gamma-ray transport Monte Carlo code. It has been designed as a tool for solving most shielding problems. Through the use of multigroup cross sections, the solution of neutron, gamma-ray, or coupled neutron-gamma-ray problems may be obtained in either the forward or adjoint mode. Time dependence for both shielding and criticality problems is provided. General three-dimensional geometry may be used with an albedo option available at any material surface. Isotropic or anisotropic scattering up to a P16 expansion of the angular distribution is allowed. B - Method of solution: Monte Carlo methods are used to solve the forward and the adjoint transport equations. Quantities of interest are then obtained by summing the contributions over all collisions, and frequently over most of phase space. Standard multigroup cross sections, such as those used in discrete ordinates codes, may be used as input; either CCC-254/ANISN, CCC-42/DTF-IV, or CCC-89/DOT cross section formats are acceptable. Anisotropic scattering is treated for each group-to-group transfer by utilizing a generalized Gaussian quadrature technique. The Morse code is organised into functional modules with simplified interfaces such that new modules may be incorporated with reasonable ease. The modules are (1) random walk, (2) cross section, (3) geometry, (4) analysis, and (5) diagnostic. The MARS module allows the efficient modeling of complex lattice geometries. Computer memory requirements are minimized because fewer body specifications are needed and nesting and repetition of arrays is allowed. While the basic MORSE code assumes the analysis module is user-written, a general analysis package, SAMBO is included. SAMBO handles some of the drudgery associated with the analysis of random walks and minimizes the amount of user-written coding. An arbitrary number of detectors, energy-dependent response functions, energy bins, time bins, and angle bins are allowed. Analysis is divided for each detector as follows: un-collided and total response, fluence versus energy, time-dependent response, fluence versus time and energy, and fluence versus angle and energy. Each of these quantities is listed as output. The diagnostic module provides an easy means of printing out, in useful form, the information in the various labeled commons and any part of blank COMMON. The module is very useful to debug a problem and to gain further insight into the physics of the random walk. C - Restrictions on the complexity of the problem: Flexible dimensioning techniques require the use of a large container array in blank COMMON