ga-optimize-sampling
TODO: Update with most recent implementation
A Genetic Algorithm implementation in python that optimizes sampling in 3D space.
In load_cflux_masked.py: A variable is selected from a netcdf file which is saved as a numpy array. A mask can be applied and the indices chosen.
In make_gene_map.py: The numpy array is used to create a gene map for the genetic algorithm. Each data point in the array is assigned a binary string. The binary string is known as the gene string. A dictionary is created where each binary string is the key. The values of these keys are also dictionaries with the keys "location" and "value" and their associated values.
eg.{'001101': {'location': (0, 10, 5), 'value': 1.2},
'001111': {'location': (0, 19, 6), 'value': 3.0}}
If the last binary string is has zeros in it eg '101111' then more binary strings are added to the dictionary until all the possible permuations of 0, 1 are used. These value and loaction for these binary strings are 1e09 and (999, 999, 999).
In make_population.py: The initial population is created. This is done by randomly selecting a zero or a one and appending it to a chromosome string. The length of the chromosome string is the product of the length of a gene string and the number of locations you would like to select to use as a sample. This process is then repeated N times with N being the size you would like the population to be.
In ga_class.py: This is where you will find the attributes and methods that you can use for the genetic algorithm such as selection, crossover, mutation, and the fitness function.
In ga.py: This file runs the genetic algorithms N times. Once the optimal sampling solution has been determined by the ga the results will be plotted.