Library for the use of various variants of the Gene-pool Optimal Mixing Evolutionary Algorith (GOMEA).
Project description
README
The Gene-pool Optimal Mixing Evolutionary Algorithm (GOMEA) is a state-of-the-art Model-Based Evolutionary Algorithm (MBEA) with an ongoing line of research into various domains of (evolutionary) optimization. The initial release of the GOMEA library (v1.0.0) is described in the publication:
The current release of the GOMEA library supports optimization in the following domains:
- Discrete single-objective optimization
- Real-valued single-objective optimization
Support for multi-objective optimization in both of these domains is planned for a future release.
Relevant most recent publications are the following:
- Dushatskiy, A., Virgolin, M., Bouter, A., Thierens, D., & Bosman, P.A.N. (2021). Parameterless Gene-pool Optimal Mixing Evolutionary Algorithms. arXiv preprint arXiv:2109.05259.
- Bouter, A., Alderliesten, T., & Bosman, P.A.N. (2021). Achieving highly scalable evolutionary real-valued optimization by exploiting partial evaluations. Evolutionary computation, 29(1), 129-155.
INSTALL
The most straightforward installation option is through pip, as follows.
pip install gomea
To build from source, execute the following in the root GOMEA folder.
make install
RUN
Running a simple benchmark problem, e.g., the Rosenbrock function, can be done as follows:
import gomea
frv = gomea.fitness.RosenbrockFunction(10,value_to_reach=1e-10)
lm = gomea.linkage.Univariate()
rvgom = gomea.RealValuedGOMEA(fitness=frv, linkage_model=lm, lower_init_range=-115, upper_init_range=-100)
result = rvgom.run()
Plotting a simple convergence plot can then be done using:
import matplotlib.pyplot as plt
plt.grid()
plt.yscale('log')
plt.xscale('log')
plt.xlabel('Number of evaluations')
plt.ylabel('Best objective value')
plt.plot(result['evaluations'],result['best_obj_val'])
Rather than using a predefined fitness function, it is also possible to define your own custom fitness function for Black-Box Optimization (BBO) or Gray-Box Optimization (GBO). The definition of a custom (real-valued) BBO function can be done as follows.
class CustomRosenbrockFunction(gomea.fitness.BBOFitnessFunctionRealValued):
def objective_function( self, objective_index, variables ):
f = 0
for i in range(len(variables)-1):
x = variables[i]
y = variables[i+1]
f += 100*(y-x*x)*(y-x*x) + (1.0-x)*(1.0-x)
return f
dim = 10
f_rosenbrock = CustomRosenbrockFunction(dim,value_to_reach=1e-10)
The definition of a GBO function requires defining the input and output of each subfunction. Thorough definitions of this are given in the EvoSOFT workshop paper. An example of the Rosenbrock function defined in a GBO setting is as follows:
class CustomRosenbrockFunction(gomea.fitness.GBOFitnessFunctionRealValued):
def number_of_subfunctions( self ):
return self.number_of_variables-1
def inputs_to_subfunction( self, subfunction_index ):
return [subfunction_index, subfunction_index+1]
def subfunction(self, subfunction_index, variables):
x = variables[subfunction_index]
y = variables[subfunction_index+1]
return 100*(y-x*x)*(y-x*x) + (1.0-x)*(1.0-x)
dim = 10
f_rosenbrock = CustomRosenbrockFunction(dim,value_to_reach=1e-10)
Further examples are given in the demos directory.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distributions
Built Distributions
Hashes for gomea-1.0.4-cp311-cp311-win_amd64.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 018125134b7e9481ed6c3d6985cb378b61f1ba4b765372af3fe6e90441941be7 |
|
MD5 | 7d30e5ce0a265faa858c73858b408bce |
|
BLAKE2b-256 | d0504be68db86cb5f2807901bcbea9e59ac8f41946ededa78d4aeb0334ee8022 |
Hashes for gomea-1.0.4-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 714f70f8177c5e34e575ad727c61f6c3cc3d29b12f3fdf85d12074494492a9d4 |
|
MD5 | 291e5cce47eef45e629713c0be3bf407 |
|
BLAKE2b-256 | ffba7592aa9b13d497569db3b9fec232587cdf7b092378094d53144068e65e21 |
Hashes for gomea-1.0.4-cp311-cp311-macosx_10_15_universal2.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | c957addc9fb42c7183f2a5b5c74a998db11b14897e7655448f201510557c3549 |
|
MD5 | c3c5025ad970e113ddc55259d3713568 |
|
BLAKE2b-256 | 5a2baece4b7c3e20a25afc5dc71336a2120182777aae924213327d4b92c50e9e |
Hashes for gomea-1.0.4-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 89ff6664ab9fd10d6186de99afe87e7e79fa0383f9eeb8ccb7a6a13815f01121 |
|
MD5 | 30caa483c576408b5b23e5ac8c17500c |
|
BLAKE2b-256 | 442c3a9fdf347ec44b585cafbf36e11439e1e31ea198a92bad156322aed1cb6a |