Optimization dialog

The following boxes are described below:

Ø Choose optimizer

Ø DE: Differential Evolution parameters

Ø HDE: Hybridized Differential Evolution

Ø VBDE: Variable Birthrate Differential Evolution

Ø IADE: Immune Accelerated Differential Evolution

Choose optimizer

- Optimizer: Select optimizer from the drop-down list. Only Differential Evolution is used in this thesis.

- N = 1/Tol: Optimization is stopped when the standard deviation of cost in the population is lower than Tol, the reciprocal of N, the number entered in this box.

- Minimum function value: Optimization is stopped if the cost becomes lower than this value.

DE: Differential Evolution parameters

- The parameters shown are those used for DE with binary crossover throughout the thesis. For details, see chapter 1.

- Strategy: Select the DE strategy from the drop-down list.

- perturbed: which vector is mutated

- # difference: the number of differentials used in mutation

- crossover: the type of crossover – binary or exponential

- Number of parameters D: calculated automatically by FanOpt

- Population size: should be roughly 10D for best results

- Parameter bounds: not used currently

- weighting factor F: controls strength of mutation

- Crossover factor, CR: controls amount of crossover

- Random number seed: initial seed for the (pseudo)-random number generator

- Maximum # of generations: Optimization is stopped after this generation

- Refresh: determines how often the screen is refreshed with the latest data

HDE: Hybridized Differential Evolution

- Hybridization with Downhill Simplex (DS) is activated by selecting the box

- Simplex choice: Select the initial simplex (D+1 vectors) from the drop-down list.

- D+1 best: least costly vectors in the population

- 1 best, D random: single vector with least cost, plus D chosen at random

- D+1 random: initial simplex is chosen completely at random

- mutate on best: single vector with least cost, plus D vectors, each of which adds to one variable of the best the standard deviation of that variable in the population

- Replacement choice: Select the manner in which vectors improved by DS replace those in the population

- D+1 best: the new simplex replaces the best vectors in the population

- D+1 worst: the new simplex replaces the best vectors in the population

- D+1 random: the new simplex replaces D+1 vectors chosen at random

- 1 best: the least costly vector in the new simplex replaces the vector that was previously best in the population

- 1 random: the least costly vector in the new simplex replaces a population vector chosen at random

- none: no population vector is replaced, but update the best vector in the population (used by DE/best/… and DE/rand-to-best/…)

- Frequency: determines how often DS is run

- # simplex iterations: determines the number of DS iterations used each time it is run

VBDE: Variable Birthrate Differential Evolution

- Select a birthrate cut-off value, BR, from [0,1].

- Use high BR values to keep most vectors from reproducing

- Use BR=0 to de-activate VBDE.

IADE: Immune Accelerated Differential Evolution

- IADE is activated by selecting the box.

- IADE conditions the antibodies to have characteristics of the antigens.

- In v. 3.3, the immune conditioning occurs after every generation

- In v. 3.5, the immune conditioning occurs only immediately following a DS run (in HDE)

- PGA: The best PGA% of the population is chosen as antigens.

- PBA: The worst PBA% of the population is chosen as antibodies.

- PEA: At any given time, PEA% of the antibodies are exposed to a given antigen.