Genetic Decryption

Java implementation of a simple genetic algorithm used to decrypt keys based on English language frequency of characters. This was used to investigate the effect of each genetic parameter on overall fitness.

Compilation and Execution

Compile with:

$ javac runGeneticDecipher.java

Execute with:

$ java runGeneticDecipher <args>

Where <args> are as follows:

Argument	Meaning	Type	Range
1	Population Size	Integer	[1, n]
2	Number of Genetation	Integer	[1, n]
3	Chromosome Size	Integer	[1, n]
4	Elite Population Size	Integer	[2, m]
5	Tournament Size	Integer	[2, 5]
6	Crossover Type	Integer	[1, 4]
7	Mutation Type	Integer	[1, 2]
8	Crossover Rate	Double	[0.00, 1.00]
9	Mutation Rate	Double	[0.00, 1.00]
10	Random Seed	Integer	[1, n]
11	Encrypted Text	String	N/A

Where n is any maximum and m is equal to half of the population size. The final argument is to a local text file.

For crossover and mutation types, the options are below:

Value	Crossover	Mutation
1	1-Point	Uniform
2	2-Point	Non-uniform
3	Uniform	Non-uniform*
4	Random	N/A

Where non-uniform* means the mutation rate increases at a later time than typical non-uniform mutation (here defined as through 33% of the total generations). For random crossover, it randomizes which crossover operator to use for each generation.

Dependencies

• Java installed
• GNU/Linux

Heuristic Evaluation

The fitness is decided by how well the decrypted text follows the frequency distribution of English letters. More closely resembling English text means a better fitness.