tour3.txt

使用遗传算法计算电力市场投标程序

!001
                Genetic Options

   This tutorial covers the options controlling
how the simulator permits the construction of
new bid strategies.  It covers:

   (1) Mutation.
   (2) Crossover.
   (3) Selection.
   (4) Generations.

A basic familiarity with genetic algorithms
(available in the documentation),is assumed.
!002
   Getting the Genetic Parameter Controls

   From the Genetics menu selecting All will
allow the user to edit mutation rate, crossover
type, generations of improvement, and the
selection method used by the genetic algorithm.
The Reset Params selection off of the Genetic
menu returns all the genetic controls to
their default states.
!003
                Mutation

   Whenever a new strategy is produced from
existing strategies the gene describing it
has a probability, independent for each bit
in the gene, that the bit will flip.  The
mutation slider permits the user to select
this per-bit rate.
!004
                Crossover

   Crossover is the blending of two existing
genes to produce a gene.  A radio button
group allows the user to select one point,
two point, uniform, or no crossover.
!005
                 Selection

   Selection is the process of choosing which
genes are chosen to be blended (crossed over)and
mutated to replace inferior genes during the
course of the simulation.  Selection is
probabilistic, proportional to some measure
of quality.  (Continued)
!006
   The user may use a radio button group to chose
selection in proportion to the genes ranking by
profit (Rank), in proportion to the profit
directly (Roulette), or in proportion to an
affine function of profit (Proportional). The
parameters of the affine function are controlled
by the sliders at the bottom of the selection
dialog.
!007
                   Generations

   There is a slider to control the number of
generation the genetic algorithm continues to
search for new bidding strategies.  It is at
the bottom of the genetic controls.
!008