Welcome to pyRVEA’s documentation!

README

https://mybinder.org/badge_logo.svgBinder

The python version reference vector guided evolutionary algorithm.

Currently supported: Multi-objective minimization with visualization and interaction support. Preference is accepted as a reference point.

To test the code, open the binder link and read example.ipynb.

Read the documentation here

Requirements:

Installation process:

  • Download and extract the code

  • Create a new virtual environment for the project

  • Run poetry install --no-dev in the activated virtual environment to install the packages necessary to run the code.

  • If you want to take part in the development process, run poetry install instead.

See the details of RVEA in the following paper

R. Cheng, Y. Jin, M. Olhofer and B. Sendhoff, A Reference Vector Guided Evolutionary Algorithm for Many-objective Optimization, IEEE Transactions on Evolutionary Computation, 2016

The source code of pyRVEA is implemented by Bhupinder Saini

If you have any questions about the code, please contact:

Bhupinder Saini: bhupinder.s.saini@jyu.fiProject researcher at University of Jyväskylä.

pyRVEA package

pyRVEA.EAs

pyRVEA.EAs.NSGAIII module

class pyRVEA.EAs.NSGAIII.NSGAIII(population: Population, EA_parameters: dict = None)

Bases: pyRVEA.EAs.baseEA.BaseDecompositionEA

Python Implementation of NSGA-III. Based on the pymoo package.

[description]

select(population: Population)

Describe a selection mechanism. Return indices of selected individuals.

Parameters

population (Population) – Contains the current population and problem information.

Returns

List of indices of individuals to be selected.

Return type

list

set_params(population: Population = None, population_size: int = None, lattice_resolution: int = None, interact: bool = True, a_priori_preference: bool = False, generations_per_iteration: int = 100, iterations: int = 10, plotting: bool = True)

Set up the parameters. Save in self.params

pyRVEA.EAs.RVEA module

class pyRVEA.EAs.RVEA.RVEA(population: Population, EA_parameters: dict = None)

Bases: pyRVEA.EAs.baseEA.BaseDecompositionEA

The python version reference vector guided evolutionary algorithm.

See the details of RVEA in the following paper

R. Cheng, Y. Jin, M. Olhofer and B. Sendhoff, A Reference Vector Guided Evolutionary Algorithm for Many-objective Optimization, IEEE Transactions on Evolutionary Computation, 2016

The source code of pyRVEA is implemented by Bhupinder Saini

If you have any questions about the code, please contact:

Bhupinder Saini: bhupinder.s.saini@jyu.fi

Project researcher at University of Jyväskylä.

select(population: Population)

Describe a selection mechanism. Return indices of selected individuals.

# APD Based selection. # This is different from the paper. # params.genetations != total number of generations. This is a compromise. Also this APD uses an archived ideal point, rather than current, potentially worse ideal point.

Parameters

population (Population) – Population information

Returns

list: Indices of selected individuals.

Return type

list

set_params(population: Population = None, population_size: int = None, lattice_resolution: int = None, interact: bool = False, a_priori_preference: bool = False, generations_per_iteration: int = 100, iterations: int = 10, Alpha: float = 2, plotting: bool = True)

Set up the parameters. Save in RVEA.params. Note, this should be changed to align with the current structure.

Parameters

  • population (Population) – Population object

  • population_size (int) – Population Size

  • lattice_resolution (int) – Lattice resolution

  • interact (bool) – bool to enable or disable interaction. Enabled if True

  • a_priori_preference (bool) – similar to interact

  • generations_per_iteration (int) – Number of generations per iteration.

  • iterations (int) – Total Number of iterations.

  • Alpha (float) – The alpha parameter of APD selection.

  • plotting (bool) – Useless really.

pyRVEA.EAs.baseEA module

class pyRVEA.EAs.baseEA.BaseDecompositionEA(population: Population, EA_parameters: dict = None)

Bases: pyRVEA.EAs.baseEA.BaseEA

This class provides the basic structure for decomposition based Evolutionary algorithms, such as RVEA or NSGA-III.

continue_evolution() → bool

Checks whether the current iteration should be continued or not.

continue_iteration()

Checks whether the current iteration should be continued or not.

select(population) → list

Describe a selection mechanism. Return indices of selected individuals.

Parameters

population (Population) – Contains the current population and problem information.

Returns

List of indices of individuals to be selected.

Return type

list

class pyRVEA.EAs.baseEA.BaseEA

Bases: object

This class provides the basic structure for Evolutionary algorithms.

set_params()

Set up the parameters. Save in self.params

pyRVEA.OtherTools

pyRVEA.OtherTools.IsNotebook module

pyRVEA.OtherTools.IsNotebook.IsNotebook() → bool

Checks if the current environment is a Jupyter Notebook or a console.

Returns

True if notebook. False if console

Return type

bool

pyRVEA.OtherTools.ReferenceVectors module

class pyRVEA.OtherTools.ReferenceVectors.ReferenceVectors(lattice_resolution: int, number_of_objectives: int, creation_type: str = 'Uniform', vector_type: str = 'Spherical', ref_point: list = None)

Bases: object

Class object for reference vectors.

adapt(fitness: numpy.ndarray)

Adapt reference vectors. Then normalize.

Parameters

fitness (np.ndarray) –

add_edge_vectors()

Add edge vectors to the list of reference vectors.

Used to cover the entire orthant when preference information is provided.

iteractive_adapt_1(ref_point, translation_param=0.2)

Adapt reference vectors linearly towards a reference point. Then normalize.

The details can be found in the following paper: Hakanen, Jussi & Chugh, Tinkle & Sindhya, Karthik & Jin, Yaochu & Miettinen, Kaisa. (2016). Connections of Reference Vectors and Different Types of Preference Information in Interactive Multiobjective Evolutionary Algorithms.

Parameters

  • ref_point –

  • translation_param – (Default value = 0.2)

neighbouring_angles() → numpy.ndarray

Calculate neighbouring angles for normalization.

normalize()

Normalize the reference vectors to a unit hypersphere.

slow_interactive_adapt(ref_point)

Basically a wrapper around rotate_toward. Slowly rotate ref vectors toward ref_point. Return a boolean value to tell if the ref_point has been reached.

Parameters

ref_point (list or np.ndarray) – The reference vectors will slowly move towards the ref_point.

Returns

True if ref_point has been reached. False otherwise.

Return type

boolean

pyRVEA.OtherTools.ReferenceVectors.householder(vector)

Return reflection matrix via householder transformation.

pyRVEA.OtherTools.ReferenceVectors.normalize(vectors)

Normalize a set of vectors.

The length of the returned vectors will be unity.

Parameters

vectors (np.ndarray) – Set of vectors of any length, except zero.

pyRVEA.OtherTools.ReferenceVectors.rotate(initial_vector, rotated_vector, other_vectors)

Calculate the rotation matrix that rotates the initial_vector to the rotated_vector. Apply that rotation on other_vectors and return. Uses Householder reflections twice to achieve this.

pyRVEA.OtherTools.ReferenceVectors.rotate_toward(initial_vector, final_vector, other_vectors, degrees: float = 5)

Rotate other_vectors (with the centre at initial_vector) towards final_vector by an angle degrees.

Parameters

  • initial_vector (np.ndarray) – Centre of the vectors to be rotated.

  • final_vector (np.ndarray) – The final position of the center of other_vectors.

  • other_vectors (np.ndarray) – The array of vectors to be rotated

  • degrees (float, optional) – The amount of rotation (the default is 5)

Returns

  • rotated_vectors (np.ndarray) – The rotated vectors

  • reached (bool) – True if final_vector has been reached

pyRVEA.OtherTools.ReferenceVectors.shear(vectors, degrees: float = 5)

Shear a set of vectors lying on the plane z=0 towards the z-axis, such that the resulting vectors ‘degrees’ angle away from the z axis.

z is the last element of the vector, and has to be equal to zero.

Parameters

  • vectors (numpy.ndarray) – The final element of each vector should be zero.

  • degrees (float, optional) – The angle that the resultant vectors make with the z axis. Unit is radians. (the default is 5)

pyRVEA.OtherTools.newRV module

pyRVEA.OtherTools.plotlyanimate module

pyRVEA.OtherTools.plotlyanimate.animate_2d_init_(data: Union[numpy.ndarray, pandas.core.frame.DataFrame, list], filename: str) → dict

Initiate a 2D scatter animation.

Only for 2D data.

Parameters

  • data (Union[np.ndarray, pd.DataFrame, list]) – Objective values

  • filename (str) – Name of the file to which plot is saved

Returns

Plotly Figure Object

Return type

dict

pyRVEA.OtherTools.plotlyanimate.animate_2d_next_(data: Union[numpy.ndarray, pandas.core.frame.DataFrame, list], figure: dict, filename: str, generation: int) → dict

Plot the next set of individuals in a 2D scatter animation.

Parameters

  • data (Union[np.ndarray, pd.DataFrame, list]) – The objective values to be plotted

  • figure (dict) – Plotly figure object compatible dict

  • filename (str) – Name of the file to which the plot is saved

  • generation (int) – Iteration Number

Returns

Plotly Figure Object

Return type

dict

pyRVEA.OtherTools.plotlyanimate.animate_3d_init_(data: Union[numpy.ndarray, pandas.core.frame.DataFrame, list], filename: str) → dict

Plot the first (or zeroth) iteration of a population.

Intended as a frames object. Plots Scatter 3D data.

Parameters

  • data (Union[np.ndarray, pd.DataFrame, list]) – Contains the data to be plotted. Each row is an individual’s objective values.

  • filename (str) – Contains the name of the file to which the plot is saved.

Returns

Plotly figure object

Return type

dict

pyRVEA.OtherTools.plotlyanimate.animate_3d_next_(data: Union[numpy.ndarray, pandas.core.frame.DataFrame, list], figure: dict, filename: str, generation: int) → dict

Plot the next set of individuals in an animation.

Plots scatter for 3D data.

Parameters

  • data (Union[np.ndarray, pd.DataFrame, list]) – The objective values to be plotted

  • figure (dict) – Plotly figure object compatible dict

  • filename (str) – Name of the file to which the plot is saved

  • generation (int) – Iteration Number

Returns

Plotly Figure Object

Return type

dict

pyRVEA.OtherTools.plotlyanimate.animate_init_(data: Union[numpy.ndarray, pandas.core.frame.DataFrame, list], filename: str) → dict

Plot the first (or zeroth) iteration of a population.

Intended as a frames object. Plots Scatter for 2D and 3D data. Plots parallel coordinate plot for higher dimensional data.

Parameters

  • data (Union[np.ndarray, pd.DataFrame, list]) – Contains the data to be plotted. Each row is an individual’s objective values.

  • filename (str) – Contains the name of the file to which the plot is saved.

Returns

Plotly figure object

Return type

dict

pyRVEA.OtherTools.plotlyanimate.animate_next_(data: Union[numpy.ndarray, pandas.core.frame.DataFrame, list], figure: dict, filename: str, generation: int) → dict

Plot the next set of individuals in an animation.

Plots scatter for 2D and 3D data, parallel coordinate plot for 4D and up.

Parameters

  • data (Union[np.ndarray, pd.DataFrame, list]) – The objective values to be plotted

  • figure (dict) – Plotly figure object compatible dict

  • filename (str) – Name of the file to which the plot is saved

  • generation (int) – Iteration Number

Returns

Plotly Figure Object

Return type

dict

pyRVEA.OtherTools.plotlyanimate.animate_parallel_coords_init_(data: Union[numpy.ndarray, pandas.core.frame.DataFrame, list], filename: str) → dict

Plot the first (or zeroth) iteration of a population.

Intended as a frames object. Plots parallel coordinate plot for >3D data.

Parameters

  • data (Union[np.ndarray, pd.DataFrame, list]) – Contains the data to be plotted. Each row is an individual’s objective values.

  • filename (str) – Contains the name of the file to which the plot is saved.

Returns

Plotly figure object

Return type

dict

pyRVEA.OtherTools.plotlyanimate.animate_parallel_coords_next_(data: Union[numpy.ndarray, pandas.core.frame.DataFrame, list], figure: dict, filename: str, generation: int) → dict

Plot the next set of individuals in an animation.

Plots parallel coordinate plot for 4D and up.

Parameters

  • data (Union[np.ndarray, pd.DataFrame, list]) – The objective values to be plotted

  • figure (dict) – Plotly figure object compatible dict

  • filename (str) – Name of the file to which the plot is saved

  • generation (int) – Iteration Number

Returns

Plotly Figure Object

Return type

dict

pyRVEA.OtherTools.plotlyanimate.test()
pyRVEA.OtherTools.plotlyanimate.test2()

pyRVEA.OtherTools.symmetric_vectors module

pyRVEA.Population

pyRVEA.Population.Population module

class pyRVEA.Population.Population.Population(problem: baseProblem, assign_type: str = 'RandomAssign', plotting: bool = True, *args)

Bases: object

Define the population.

add(new_pop: numpy.ndarray)

Evaluate and add individuals to the population. Update ideal and nadir point.

Parameters

new_pop (np.ndarray) – Decision variable values for new population.

append_individual(ind: numpy.ndarray)

Evaluate and add individual to the population.

Parameters

ind (np.ndarray) –

create_new_individuals(design: str = 'LHSDesign', pop_size: int = None, decision_variables=None)

Create, evaluate and add new individuals to the population. Initiate Plots.

The individuals can be created randomly, by LHS design, or can be passed by the user.

Parameters

  • design (str, optional) – Describe the method of creation of new individuals. “RandomDesign” creates individuals randomly. “LHSDesign” creates individuals using Latin hypercube sampling.

  • pop_size (int, optional) – Number of individuals in the population. If none, some default population size based on number of objectives is chosen.

  • decision_variables (numpy array or list, optional) – Pass decision variables to be added to the population.

eval_fitness()

Calculate fitness based on objective values. Fitness = obj if minimized.

evaluate_individual(ind: numpy.ndarray)

Evaluate individual.

Returns objective values, constraint violation, and fitness.

Parameters

ind (np.ndarray) –

evolve(EA: BaseEA = None, EA_parameters: dict = None) → Population

Evolve the population with interruptions.

Evolves the population based on the EA sent by the user.

Parameters

  • EA ("BaseEA") – Should be a derivative of BaseEA (Default value = None)

  • EA_parameters (dict) – Contains the parameters needed by EA (Default value = None)

hypervolume(ref_point)

Calculate hypervolume. Uses package pygmo. Add checks to prevent errors.

Parameters

ref_point –

keep(indices: list)

Remove individuals from population which are not in “indices”.

Parameters

indices (list) – Indices of individuals to keep

mate()

Conduct crossover and mutation over the population.

Conduct simulated binary crossover and bounded polunomial mutation.

non_dominated()

Fix this. check if nd2 and nds mean the same thing

plot_init_()

Initialize animation objects. Return figure

plot_objectives(iteration: int)

Plot the objective values of individuals in notebook. This is a hack.

Parameters

iteration (int) – Iteration count.

update_ideal_and_nadir(new_objective_vals: list = None)

Updates self.ideal and self.nadir in the fitness space.

Uses the entire population if new_objective_vals is none.

Parameters

new_objective_vals (list, optional) – Objective values for a newly added individual (the default is None, which calculated the ideal and nadir for the entire population.)

pyRVEA.Problem package

Submodules

pyRVEA.Problem.baseProblem module

class pyRVEA.Problem.baseProblem.baseProblem(name=None, num_of_variables=None, num_of_objectives=None, num_of_constraints=0, upper_limits=1, lower_limits=0)

Bases: object

Base class for the problems.

constraints(decision_variables)

Accept a sample and/or corresponding objective values.

Parameters

decision_variables –

objectives(decision_variables)

Accept a sample. Return Objective values.

Parameters

decision_variables –

update()

Update the problem based on new information.

pyRVEA.Problem.testProblem module

class pyRVEA.Problem.testProblem.testProblem(name=None, num_of_variables=None, num_of_objectives=None, num_of_constraints=0, upper_limits=1, lower_limits=0)

Bases: pyRVEA.Problem.baseProblem.baseProblem

Defines the problem.

constraints(decision_variables, objective_variables)

Calculate constraint violation.

Parameters

  • decision_variables –

  • objective_variables –

objectives(decision_variables) → list

Use this method to calculate objective functions.

Parameters

decision_variables –

pyRVEA.Recombination package

Module contents

pyRVEA.Selection package

Submodules

pyRVEA.Selection.APD_select module

pyRVEA.Selection.APD_select.APD_select(fitness: list, vectors: ReferenceVectors, penalty_factor: float, ideal: list = None)

Select individuals for mating on basis of Angle penalized distance.

Parameters

  • fitness (list) – Fitness of the current population.

  • vectors (ReferenceVectors) – Class containing reference vectors.

  • penalty_factor (float) – Multiplier of angular deviation from Reference vectors. See RVEA paper for details.

  • ideal (list) – ideal point for the population. Uses the min fitness value if None.

Returns

A list of indices of the selected individuals.

Return type

[type]

pyRVEA.Selection.NSGAIII_select module

pyRVEA.Selection.NSGAIII_select.NSGAIII_select(fitness: list, ref_dirs: ReferenceVectors, ideal_point: list = None, worst_point: list = None, extreme_points: list = None, n_survive: int = None)
pyRVEA.Selection.NSGAIII_select.associate_to_niches(F, ref_dirs, ideal_point, nadir_point, utopian_epsilon=0.0)
pyRVEA.Selection.NSGAIII_select.calc_niche_count(n_niches, niche_of_individuals)
pyRVEA.Selection.NSGAIII_select.calc_perpendicular_distance(N, ref_dirs)
pyRVEA.Selection.NSGAIII_select.get_extreme_points_c(F, ideal_point, extreme_points=None)

Taken from pymoo

pyRVEA.Selection.NSGAIII_select.get_nadir_point(extreme_points, ideal_point, worst_point, worst_of_front, worst_of_population)
pyRVEA.Selection.NSGAIII_select.niching(F, n_remaining, niche_count, niche_of_individuals, dist_to_niche)