Application Program Interfaces

Application Program Interfaces

APIs can be used to connect with existing code implementing math-ematical model and objective function. This approach is often preferred in academic and industrial applications where the high level modeling system language is not sufficient or the computation time can be reduced.
DEEP method can be used as the static or dynamic shared object and embedded in the other software. As a C library DEEP provides the following main features:
1. optimization of real parameters,
2. optimization of integer parameters,
3. conversion of real to integer parameters,
4. transformation of constraints in the form of inequalities in order to convert the constrained problem
in the unconstrained one.

To communicate with the DEEP method the client programme should implement the library structures
described below.
DpOpt structure
The control parameters of DEEP method such as population size, offspring generation strategy, the number of the oldest individuals to be substituted in the next generation Ψ and an adaptation constant γ are stored in DpOpt. The definition of the stopping criterion and the current values of internal counters are also packed here.
DpTarget structure
The objective function to be minimized is called target and is defined in the corresponding section of the configuration file. This definition maps to the DpTarget structure. The evaluation of the objective function may involve up to three steps: optional conversion of parameters, obligatory calculation of main objective and optional penalty functions. The function pointers for each step together with supporting information are packed in this structure.
XmModel structure
The mathematical model under consideration is defined by the vectors of parameter values, the lower and upper bounds and the name of the programme that is called to measure the discrepancy between the model and experimental data. This structure also lists fixed and constrained parameters.