Model is, similarly as in the sequential mesh generator, described by
the boundary representation and consists of the same model
entities -- vertices, curves, surfaces and regions.
This time, however, each model entity (including regions) is based on
free-form representation in terms of tensor product polynomial
entities (again, the rational Bezier entities are employed),
which essentially restricts the model topology. In this approach, each region is bounded by six
surfaces, boundary each of which is formed by four curves, each of
which is in turn given by two vertices. The advantage of this model
representation consists in existence of unique mapping between the
parametric and real spaces of each of the model entities. This
significantly contributes to unified handling of individual model
entities. On the other hand, the restriction on the model topology
results in some reduction of modeling flexibility. To enhance the
modeling capability curves and surfaces (and even regions) are allowed
to be degenerated into entity of lower dimension. Also
a simple entity-to-entity fixation concept is employed.
Each model entity is keeping the list of entities which are bounding
and sharing that entity. No further topological information are required for
description of valid domain of arbitrary complexity.

*Daniel Rypl
2005-12-03*