10.2.2.3. TEgoSolver

There are 4 available solvers combining two distinct features:

• dynamic or static optimization: in static, the search restart with a new random population; in dynamic the search restart from the previous population which should be rich enough to find a point that was put aside. In dynamic, the first search is longer than the following one.

• genetic or HJMA algorithm;

This leads to the following classes: TEgoDynSolver, TEgoStdSolver, TEgoHjDynSolver and TEgoHjStdSolver

some methods are provided:

• significantEI with a double parameter is used to define the EI threshold to stop the search loop

• setManyNewItem with an int is used to define the maximum number of new items used to feed the empty resources (unused with TEgoStdSolver).

• for the class using HJMA, setSize with two int parameters defines the number of global search performed by the optimization in the preliminary search and in the following ones.

• for the class using Vizir algorithm, setSolver with a pointer on a TVizirSolver defines the solver to use.

• for the TEgoDynSolver, the first and longer search uses the maximum evaluation number defined in the solver. The following search are shorter and is defined using setStepSize and its int argument.