\section{INTRODUCTION}\label{sect:Introduction}

The computer code DRAGON is a lattice code designed around solution techniques of 
the neutron transport equation.\cite{PIP2009} The DRAGON project results from an effort made at
{\sl \'Ecole Polytechnique de Montr\'eal} to rationalize and unify into a single code
the different models and algorithms used in a lattice code.\cite{Dragon1,Dragon2,Dragon3,Dragon4}
One of the main concerns was to ensure
that the structure of the code was such that the development and implementation
of new calculation techniques would be facilitated. DRAGON is therefore a
lattice cell code which is divided into many calculation modules linked together
around the Ganlib kernel and can be called from CLE-2000.\cite{ganlib5,cle2000} These modules exchange
informations only via well defined data structures.

The two main components of the code DRAGON are its multigroup flux solver and
its one-group collision probability (CP) tracking modules. The CP modules  all
perform the same task but using different levels of approximation.

The SYBIL tracking option emulates the main flux calculation option available in
the APOLLO-1 code,\cite{Apollo,SPH} and includes a new version of the
EURYDICE-2 code which performs reactor assembly calculations in both rectangular
and hexagonal geometries using the interface current method. The option
is activated when the \moc{SYBILT:} module is called.

The EXCELL tracking option is used to generate the collision probability
matrices for the cases having cluster, two-dimensional or three-dimensional
mixed rectangular and cylindrical geometries.\cite{DragonPIJI,Mtl93a} A cyclic
tracking option is also available for treating specular boundary conditions in
two-dimensional rectangular geometry.\cite{DragonPIJS1,Mtl93b} EXCELL
calculations are performed using the \moc{EXCELT:} or \moc{NXT:} module.

The MCCG tracking option activates the long characteristics solution technique.
This implementation uses the same tracking as EXCELL and perform flux
integration using the long characteristics algorithm proposed by Igor
Suslov.\cite{mccg,suslov2,chicago2} The option
is activated when both \moc{EXCELT:} (or \moc{NXT:}) and \moc{MCCGT:} modules are called.

After the collision probability or response matrices associated with a given
cell have been generated, the multigroup solution module can be activated. This
module uses the power iteration method and  requires a number of iteration
types.\cite{PIM} The thermal iterations are carried out by DRAGON so as to
rebalance the flux distribution only in cases where neutrons undergo
up-scattering. The power iterations are performed by DRAGON to solve the fixed
source or eigenvalue problem in the cases where a multiplicative medium is
analyzed. The effective multiplication factor ($K_{\rm eff}$) is obtained during
the power iterations. A search for the critical buckling may be superimposed
upon the power iterations so as to force the multiplication factor to take on a
fixed value.\cite{Buck}

DRAGON can access directly standard microscopic cross-section libraries in
various formats. It has the capability of
exchanging macroscopic cross-section libraries with a code such as TRANSX-CTR or
TRANSX-2 by the use of GOXS format files.\cite{MATXS,TRANSX2} The macroscopic
cross section can also be read in DRAGON via the input data stream.