[en] 1 - Description of problem or function: TRIDENT solves the two- dimensional-multigroup-transport equations in rectangular (x-y) and cylindrical (r-z) geometries using a regular triangular mesh. Regular and adjoint, inhomogeneous and homogeneous (K-eff and eigenvalue searches) problems subject to vacuum, reflective, white, or source boundary conditions are solved. General anisotropic scattering is allowed and anisotropic-distributed sources are permitted. 2 - Method of solution: The discrete-ordinates approximation is used for the neutron directional variables. An option is included to append a fictitious source to the discrete-ordinates equations that is defined such that spherical-harmonics solutions (in x-y geometry) or spherical-harmonics-like solutions (in r-z geometry) are obtained. This option is useful for cases in which ray-effect distortions are severe. A spatial-finite-element method is used in which the angular flux is expressed as a linear polynomial in each triangle that is discontinuous at triangle boundaries. Both inner (within-group) and outer iteration cycles are accelerated by either whole-system or fine-mesh re-balance. Provision is made for creation of standard interface output files for Sn constants, angle-integrated (scalar) fluxes, and angular fluxes. Standard interface input files for Sn constants, inhomogeneous sources, cross sections, and the scalar flux may be read. Subroutines DRED and DRIT perform information transfers between LCM and random disk. Data transfer between large and small core are performed by CRED and CRIT. Sequential binary operations are performed by SEEK, REED, and RITE. Flexible edit options as well as a dump and restart capability are provided. 3 - Restrictions on the complexity of the problem: Variable dimensioning is used so that any combination of problem parameters leading to a container array less than MAXLEN can be accommodated. On CDC Machines MAXLEN can be about 40 000 words of Small Core Memory (SCM) and Large Core Memory (LCM) is used for most group- dependent data. On IBM machines, TRIDENT executes in double precision (8 bytes per Word) so that MAXLEN can be on the order of 100 000 or more. Although most problems can be fast core contained on the IBM machine and SCM and LCM contained on the CDC machine, an option exists to automatically overflow the storage of selected large arrays to random disk if necessary