Information Technology Reference
In-Depth Information
reduced effectively with reasonable calculation cost and the existing
experimental result shows that this algorithm is superior to the other similar
algorithms. In addition, some special relations have been realized, such as
equality and inequality reasoning, unit sharing strategy of identical relation and
the combination of inequality graph and interval reasoning. These realization
combines evaluation of constraint expression and symbol relation together and
strengthens symbolic deduction ability of constraint reasoning. In constraint
languages, we designed SCL¾an object-oriented constraint language. SCL
realize an integrated constraint reasoning with default constraint representation
and adopts certain-type control of regular language (e.g. condition structure),
while confines the uncertain part to the data. As a result, constraint propagation
with intelligent backtracking can be used to reduce uncertainty and narrow search
space. At the same time, such regular structure also improves code readability
and makes language studying easily. We have also realized constraint
representation embedded in C++ so that constraint programming design fully
utilizes the abundant computational resources of C++.
An constraint satisfaction problem(abbreviated as CSP) includes a set of
variables and constraints between variables. In general, variables represent field
parameter and each variable has a fixed value domain. One variable's value
domain may be limited, for example, one Boolean domain only contains two
values. The value domain may be dispersed and limitless(e.g. integer domain)
and may be continuous as well(e.g. real number domain). Constraints can be used
to describe field object's property, interrelation, task requirement, goal and so on.
The goal of constrain satisfaction problem to find one or more assignments of all
variables in order to satisfy all constraints.
Constraint representation is easy to understand, code and effective realize.
Constraint representation has the following advantages:
(1) Constraint representation allows to represent domain knowledge in a
declarative way and it has strong expressing ability. The application only
needs to define goal conditions and data interrelation in a problem. Therefore
constraint
representation
has
the
resembling
characteristics
of
logic
representation.
(2) Constraint representation allows variable domain to contain arbitrary values,
while the proposition does only fetch true or false values. So constraint
representation can keep some structural information of problem, such as
variable domain size, variable interrelation etc. in order to offer heuristic
