Summary of DE

Thursday, December 14, 2023 3:58 PM

• Differential Evolution (DE) was first proposed between 1994 – 1996

by Storn and Price at Berkeley as a new evolutionary algorithm.
• Unlike the conventional evolutionary algorithms which depend on
predefined probability distribution function for mutation process,
DE uses the differences of randomly sampled pairs of objective
vectors for its mutation process.
• DE uses a non-uniform crossover which makes use of child vector
parameters to guide through the minimization process.
• The mutation operation with DE is performed by arithmetical
combinations of individuals rather than perturbing the genes in
individuals with small probability.
• Another main characteristic of DE is its ability to search with
floating point representation instead of binary representation as used
in many basic EAs such as GAs.
DE GA
Initialization Initialization
Evaluation Evaluation
Repeat Repeat
Mutation Reproduction
Recombination Crossover
Evaluation Mutation
Selection Evaluation
Until (termination criteria are Until (termination criteria are
met) met)

Differential Evolution Page 1

Pseudocode of DE
Thursday, December 14, 2023 4:07 PM

-----------------------------------------------------------------------------------------------------------------
Generate and evaluate initial population;
while termination criteria not satisfied
for i = 1 to N
for j = 1 to d
Generate three random integers ;
Generate random integer ;
if rand (0, 1) < CR or i = irand

else

end if
end for
evaluate ;
if

else

end if
end for
end while
-----------------------------------------------------------------------------------------------------------------

Differential Evolution Page 2