A Genetic Algorithm of Flowers
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
data
pics
.gitattributes
.gitignore
Bloom.pde
FitnessGraph.pde
Flower.pde
GenePool.pde
IAAColor.pde
Leaf.pde
Mendel.pde
README.md
Sector.pde
Stem.pde
blomster.pde

README.md

blomster (flowers) - A Genetic Algorithm of Flowers

Clay Heaton - 2014

pic

This Processing sketch is an examination of a genetic algorithm, expressed phenotypically through colorful flowers.

There are two modes to the sketch. To change modes, you must change the mode variable, which is the first in the main blomster.pde file.

  • Set mode to 0 for random mode. This will allow you to see the variation in the flowers. Simply click your mouse to generate a new set of random flowers. The following image is from random mode.

random

  • Set mode to 1 for genetic mode. The sketch will start by trying to converge on a target chromosome.
    • If you don't like the target, simply click with the mouse to get a new random target.
    • When the algorithm converges, you will see selections from the generations that led to the convergence on the canvas.
    • There are several variables at the top of blomster.pde that you can manipulate to change the genetic algorithm.
    • The following images are from different iterations of genetic mode.

pic

pic

pic

pic

Special thanks to Daniel Shiffman and his book, The Nature of Code, for providing such clear and concise guides to complicated concepts.

Here are the notes from the top of blomster.pde:

// blomster (flowers) v. 1.0
// A Genetic Algorithm of Flowers
// Clay Heaton - 2014

// The best way to experience this sketch is to first
// put it into random mode (first parameter below) and 
// launch it. The flowers you see are randomly generated.
// Click the mouse to regenerate them and get a sense
// of the variety.

// You then can put it back into genetic mode and 
// launch the sketch to see the population of random
// flowers evolve towards the target. The most fit
// flower in each generation is shown to the left 
// and the target is shown to the right. When the
// genetic algorithm converges, the screen will 
// show a selection of the most fit flowers in the
// generations of the evolution. The earliest 
// generations are represented in the upper-left
// and the later generations are represented in 
// the lower-right.

// Press p to capture a .pdf of the screen,
// which will be saved in your sketch's file.

import java.util.Collections;
import processing.pdf.*;

/* ************************************* */
/* TWEAK TO AFFECT THE GENETIC ALGORITHM */
/* ************************************* */
// Set to 0 for random, 1 for genetic
int mode               = 1;

// For starting the genetic algorithm
// Higher numbers converge more quickly
int populationSize     = 21;   

// Stop after this many and display as if converged
int numGenerations     = 50000;

// Consider converged when this fitness is reached
float convergenceValue = 0.95;

// The percentage chance that a gene will mutate following crossover
float mutationRate     = 0.015;

// You can seed this with a VALID chromosome
// Or leave as "" to start with a random chromosome.
String targetChromosome = "BLMEBNEDCAPACBCF";

// If this is set to true, then the target chromosome
// above always should evolve in the same manner because
// the random number generator always should return the
// same sequence of numbers. 
boolean seedRandomNumberGenerator = false;
/* ************************************* */
/* ************************************* */

Copyright 2014 Clay Heaton