CPLEX Chap Or2 Thesis
The IBM ILOG CPLEX Optimization Studio is one of the most powerful optimization solvers available nowadays.
From the user's prospective, CPLEX can be seen as a black-box: it provides a solution for a given LP model without showing the underlying computations.
Although CPLEX is a general-purpose solver, it can be customized to efficiently yield optimal solutions to the TSP problem. Indeed, it offers to developers the possibility to create their own callbacks, which can be retrieved at runtime.
The CPLEX solver will always provide one optimal solution if such exists and it can be found before the time limit's
expiration.
\
\begin{algorithm}
\caption{Variable Neighborhood Search}
current\_tour $\gets$ initial\_tour\;
cost $\gets$ initial\_cost\;
\While{remaining\_time > 0}{
current\_tour $\gets$ 2opt(current\_tour) \CommentSty{/*remove all cross*/}\\
cost $\gets$ compute\_cost(current\_tour)\;
\If{cost < best\_cost} {
best\_cost $\gets$ cost\;
best\_tour $\gets$ current\_tour\;
}
\For{$i := 0$ \textbf{to} $4$}{
current\_tour $\gets$ kick(current\_tour)\;
}
cost $\gets$ compute\_cost(current\_tour)\;
}
\end{algorithm}