├── .gitignore
├── DNE4py
├── __init__.py
├── mpi_extensions.py
├── optimizers
│ ├── __init__.py
│ ├── deepga
│ │ ├── __init__.py
│ │ ├── base.py
│ │ ├── deepga.py
│ │ ├── member.py
│ │ ├── mutation.py
│ │ └── selection.py
│ └── optimizer.py
└── version_v2
│ ├── new_optimizers
│ ├── cmaes
│ │ ├── __init__.py
│ │ └── cmaes.py
│ ├── deepga2
│ │ ├── __init__.py
│ │ ├── base.py
│ │ ├── deepga.py
│ │ ├── member.py
│ │ ├── mutation.py
│ │ ├── ranking.py
│ │ └── selection.py
│ └── random
│ │ ├── __init__.py
│ │ └── batch.py
│ ├── new_tutorials
│ └── tutorials2
│ │ ├── README.md
│ │ ├── pp_run.py
│ │ ├── render.py
│ │ └── run.py
│ ├── postprocessing
│ ├── __init__.py
│ ├── test.py
│ └── utils.py
│ └── sliceops.py
├── LICENSE
├── README.md
├── setup.py
└── tutorials
├── 1_optimizing_user_defined_function
├── TruncatedRealMutatorGA.yaml
└── run.py
├── 2_postprocessing_the_results
├── gif
│ ├── cmaes.gif
│ ├── deepga_truncatedrealmutatorga.gif
│ └── randomsearch.gif
├── pp_run.py
└── render.py
└── README.md
/.gitignore:
--------------------------------------------------------------------------------
1 | # pyenv
2 | .python-version
3 |
4 | # Byte-compiled / optimized / DLL files
5 | __pycache__/
6 | *.py[cod]
7 | *$py.class
8 |
9 | # C extensions
10 | *.so
11 |
12 | # Distribution / packaging
13 | .Python
14 | build/
15 | develop-eggs/
16 | dist/
17 | downloads/
18 | eggs/
19 | .eggs/
20 | lib/
21 | lib64/
22 | parts/
23 | sdist/
24 | var/
25 | wheels/
26 | share/python-wheels/
27 | *.egg-info/
28 | .installed.cfg
29 | *.egg
30 | MANIFEST
31 |
32 | # PyInstaller
33 | # Usually these files are written by a python script from a template
34 | # before PyInstaller builds the exe, so as to inject date/other infos into it.
35 | *.manifest
36 | *.spec
37 |
38 | # Installer logs
39 | pip-log.txt
40 | pip-delete-this-directory.txt
41 |
42 | # Unit test / coverage reports
43 | htmlcov/
44 | .tox/
45 | .nox/
46 | .coverage
47 | .coverage.*
48 | .cache
49 | nosetests.xml
50 | coverage.xml
51 | *.cover
52 | *.py,cover
53 | .hypothesis/
54 | .pytest_cache/
55 | cover/
56 |
57 | # Translations
58 | *.mo
59 | *.pot
60 |
61 | # Django stuff:
62 | *.log
63 | local_settings.py
64 | db.sqlite3
65 | db.sqlite3-journal
66 |
67 | # Flask stuff:
68 | instance/
69 | .webassets-cache
70 |
71 | # Scrapy stuff:
72 | .scrapy
73 |
74 | # Sphinx documentation
75 | docs/_build/
76 |
77 | # PyBuilder
78 | .pybuilder/
79 | target/
80 |
81 | # Jupyter Notebook
82 | .ipynb_checkpoints
83 |
84 | # IPython
85 | profile_default/
86 | ipython_config.py
87 |
88 | # pyenv
89 | # For a library or package, you might want to ignore these files since the code is
90 | # intended to run in multiple environments; otherwise, check them in:
91 | # .python-version
92 |
93 | # pipenv
94 | # According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
95 | # However, in case of collaboration, if having platform-specific dependencies or dependencies
96 | # having no cross-platform support, pipenv may install dependencies that don't work, or not
97 | # install all needed dependencies.
98 | #Pipfile.lock
99 |
100 | # PEP 582; used by e.g. github.com/David-OConnor/pyflow
101 | __pypackages__/
102 |
103 | # Celery stuff
104 | celerybeat-schedule
105 | celerybeat.pid
106 |
107 | # SageMath parsed files
108 | *.sage.py
109 |
110 | # Environments
111 | .env
112 | .venv
113 | env/
114 | venv/
115 | ENV/
116 | env.bak/
117 | venv.bak/
118 |
119 | # Spyder project settings
120 | .spyderproject
121 | .spyproject
122 |
123 | # Rope project settings
124 | .ropeproject
125 |
126 | # mkdocs documentation
127 | /site
128 |
129 | # mypy
130 | .mypy_cache/
131 | .dmypy.json
132 | dmypy.json
133 |
134 | # Pyre type checker
135 | .pyre/
136 |
137 | # pytype static type analyzer
138 | .pytype/
139 |
140 | # Cython debug symbols
141 | cython_debug/
142 |
--------------------------------------------------------------------------------
/DNE4py/__init__.py:
--------------------------------------------------------------------------------
1 | def load_optimizer(config):
2 |
3 | from DNE4py.optimizers.deepga import TruncatedRealMutatorGA
4 |
5 | optimizer_classes = {"TruncatedRealMutatorGA": TruncatedRealMutatorGA}
6 |
7 | optimizer = optimizer_classes[config['id']](config)
8 | return optimizer
9 |
10 | def load_mpidata(name, folder_path):
11 |
12 | import json
13 | import glob
14 | import numpy as np
15 |
16 | # Internals:
17 | nb_files = len(glob.glob1(f'{folder_path}', f'{name}*'))
18 | with open(f'{folder_path}/info.json', 'rb') as f:
19 | info = json.load(f)
20 | nb_generations = info['nb_generations']
21 |
22 | # Get data:
23 | if name in ['costs', 'genotypes']:
24 | data = [[] for i in range(nb_files)]
25 | for i in range(nb_files):
26 | generation_data = []
27 | with open(f'{folder_path}/{name}_w{i}.npy', 'rb') as f:
28 | for g in range(nb_generations):
29 | generation_data.append(np.load(f, allow_pickle=True).tolist())
30 | data[i] = generation_data
31 | data = np.array(data, object)
32 | data = np.transpose(data, (1, 0, 2))
33 | data = data.reshape((data.shape[0], data.shape[1] * data.shape[2]))
34 | return data
35 | elif name in ['initial_guess']:
36 | with open(f'{folder_path}/initial_guess.npy', 'rb') as f:
37 | return np.load(f, allow_pickle=True)
38 | else:
39 | print(f'load_mpidata failed, name "{name}" not found!')
40 | exit()
41 |
42 | def get_best_phenotype(folder_path):
43 |
44 | import json
45 | import numpy as np
46 | from DNE4py.optimizers.deepga.mutation import Member
47 |
48 | # Internals:
49 | # with open(f'{folder_path}/info.json', 'rb') as f:
50 | # info = json.load(f)
51 | # sigma = info['sigma']
52 |
53 | # Read Input:
54 | costs = load_mpidata("costs", f"{folder_path}")
55 | genotypes = load_mpidata("genotypes", f"{folder_path}")
56 | initial_guess = load_mpidata("initial_guess", f"{folder_path}")
57 |
58 | # Select Best Idx:
59 | best_idx = np.unravel_index(costs.argmin(), costs.shape)
60 |
61 | # Create member and get phonetype:
62 | phenotype = Member(initial_guess, genotypes[best_idx]).phenotype
63 | return phenotype
64 |
65 | def get_best_phenotype_generator(folder_path):
66 |
67 | import json
68 | import numpy as np
69 | from DNE4py.optimizers.deepga.mutation import Member
70 |
71 | # Internals:
72 | with open(f'{folder_path}/info.json', 'rb') as f:
73 | info = json.load(f)
74 | nb_generations = info['nb_generations']
75 |
76 | # Read Input:
77 | costs = load_mpidata("costs", f"{folder_path}")
78 | genotypes = load_mpidata("genotypes", f"{folder_path}")
79 | initial_guess = load_mpidata("initial_guess", f"{folder_path}")
80 |
81 | # Select Best Idxs:
82 | min_idxs = np.argmin(costs, axis=1)
83 | for i in range(nb_generations):
84 | genotype = genotypes[i, min_idxs[i]]
85 | yield Member(initial_guess, genotype).phenotype
86 |
--------------------------------------------------------------------------------
/DNE4py/mpi_extensions.py:
--------------------------------------------------------------------------------
1 | import os
2 | import pickle
3 | import logging
4 |
5 | import numpy as np
6 |
7 | class MPISaver:
8 |
9 | def __init__(self, file_path):
10 |
11 | self.file_path = file_path
12 | self.folder_path, self.filename = os.path.split(self.file_path)
13 |
14 | try:
15 | os.makedirs(self.folder_path)
16 | except:
17 | pass
18 |
19 | assert os.path.isfile(self.file_path) == False, f'You should delete all files inside the folder: {self.folder_path} | {self.file_path}'
20 |
21 | def write(self, data):
22 | with open(self.file_path, 'ab') as f:
23 | np.save(f, data)
24 |
25 | class MPILogger:
26 |
27 | def __init__(self, file_path):
28 |
29 | self.file_path = file_path
30 | self.folder_path, self.filename = os.path.split(file_path)
31 |
32 | try:
33 | os.makedirs(self.folder_path)
34 | except:
35 | pass
36 |
37 | assert os.path.isfile(self.file_path) == False, f'You should delete all files inside the folder: {self.folder_path} | {self.file_path}'
38 |
39 | logging.basicConfig(filename=self.file_path,
40 | level=logging.DEBUG,
41 | format='%(message)s')
42 |
43 | def debug(self, msg):
44 | logging.debug(msg)
45 |
--------------------------------------------------------------------------------
/DNE4py/optimizers/__init__.py:
--------------------------------------------------------------------------------
1 | #from .deepga import TruncatedRealMutatorGA
2 |
--------------------------------------------------------------------------------
/DNE4py/optimizers/deepga/__init__.py:
--------------------------------------------------------------------------------
1 | from .deepga import TruncatedRealMutatorGA
2 |
3 | __all__ = [
4 | "TruncatedRealMutatorGA",
5 | ]
6 |
--------------------------------------------------------------------------------
/DNE4py/optimizers/deepga/base.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 | import json
3 | import random
4 |
5 | from abc import ABC, abstractmethod
6 |
7 | from DNE4py.mpi_extensions import MPISaver, MPILogger
8 |
9 | #from DNE4py.optimizers.optimizer import Optimizer
10 |
11 | class BaseGA:
12 |
13 | r'''
14 | Example config:
15 |
16 | id: TruncatedRealMutatorGA
17 | initial_guess = np.array([-0.3, 0.7])
18 | workers_per_rank: 10
19 | num_elite: 3
20 | num_parents: 5
21 | sigma: 0.05
22 | global_seed: 42
23 | output_folder: 'results/DeepGA/TruncatedRealMutatorGA'
24 | save_steps: 1
25 | verbose: 0
26 |
27 | Description:
28 |
29 | * output_folder (int)
30 | => path for the raw_data that will be processed with
31 | postprocessing module
32 |
33 | * save_steps (int)
34 | => save per each save_steps generations
35 |
36 | * verbose (int)
37 | => 0 no printing
38 | => 1 print number of generations with rank 0
39 | => 2 save debug file
40 |
41 | '''
42 |
43 | def __init__(self, config):
44 |
45 | # config.get('env_options').get('max_iteration')
46 |
47 | # super().__init__(**config)
48 |
49 | # Initiate MPI
50 | from mpi4py import MPI
51 | self._MPI = MPI
52 | self._comm = MPI.COMM_WORLD
53 | self._size = self._comm.Get_size()
54 | self._rank = self._comm.Get_rank()
55 |
56 | # Configuration:
57 | self.id = config.get('id')
58 | self.workers_per_rank = config.get('workers_per_rank')
59 | self.num_elite = config.get('num_elite')
60 | self.num_parents = config.get('num_parents')
61 | self.sigma_initial = config.get('sigma_initial')
62 | self.sigma_min = config.get('sigma_min')
63 | self.sigma_decay = config.get('sigma_decay')
64 |
65 | self.global_seed = config.get('global_seed')
66 |
67 | self.output_folder = config.get('output_folder')
68 | self.save_steps = config.get('save_steps')
69 | self.verbose = config.get('verbose')
70 |
71 | self.initial_guess = config.get('initial_guess')
72 |
73 | # Internal:
74 | self.sigma = self.sigma_initial
75 | self.generation = 0
76 | self.population_size = self._size * self.workers_per_rank
77 |
78 | # Initialize Mutator and Selection
79 | self.mutator_initialize()
80 | self.selection_initialize()
81 |
82 |
83 | # Logger and DataCollector for MPI:
84 |
85 | if self.output_folder != None and self.verbose == 2:
86 | self.logger = MPILogger(f'{self.output_folder}/info_w{self._rank}.log')
87 |
88 | if self.output_folder != None:
89 | self.saver_genotypes = MPISaver(f'{self.output_folder}/genotypes_w{self._rank}.npy')
90 | self.saver_costs = MPISaver(f'{self.output_folder}/costs_w{self._rank}.npy')
91 |
92 | if self._rank == 0:
93 | self.saver_initialguess = MPISaver(f'{self.output_folder}/initial_guess.npy')
94 |
95 | #@abstractmethod
96 | #def apply_selection(self, ranks_by_performance):
97 | # pass
98 |
99 | #@abstractmethod
100 | #def apply_mutation(self, ranks_by_performance):
101 | # pass
102 |
103 | def run(self, objective_function, steps):
104 |
105 | self.objective_function = objective_function
106 |
107 | if self._rank == 0:
108 | self.saver_initialguess.write(self.initial_guess)
109 |
110 | for _ in range(steps):
111 |
112 | # =================== LOGGING =====================================
113 | if self.verbose == 1:
114 | if self._rank == 0:
115 | print(f"Generation: {self.generation}/{steps}")
116 | elif self.verbose == 2:
117 | self.logger.debug(f"\nGeneration: {self.generation}/{steps}")
118 | # =================== LOGGING =====================================
119 |
120 | self.step()
121 |
122 | # CLOSING:
123 | if self._rank == 0:
124 | if self.output_folder != None:
125 | info = {'nb_generations': steps}
126 | with open(f'{self.output_folder}/info.json', "w") as f:
127 | json.dump(info, f)
128 |
129 | def step(self):
130 |
131 | # Evaluate member:
132 | self.cost_list = np.zeros(self.workers_per_rank)
133 | for i in range(len(self.cost_list)):
134 | self.cost_list[i] = self.objective_function(self.members[i].phenotype)
135 |
136 | # ======================= LOGGING =====================================
137 | if self.verbose == 2:
138 | self.logger.debug(f"\nPopulation Members (Initial):")
139 | self.logger.debug(f"| index | seed | x0 | x1 | y |")
140 | for index in range(len(self.members)):
141 | seed = self.members[index].genotype
142 | x0 = self.members[index].phenotype[0].round(10)
143 | x1 = self.members[index].phenotype[1].round(10)
144 | y = self.cost_list[index].round(10)
145 | self.logger.debug(f"| {index} | {seed} | {x0} | {x1} | {y} |")
146 | # ===================== END LOGGING ===================================
147 |
148 | # ======================= SAVING =====================================
149 | if (self.generation % self.save_steps == 0):
150 | self.saver_genotypes.write(self.genotypes)
151 | self.saver_costs.write(self.costs)
152 | else:
153 | self.saver_genotypes.write(np.array(np.nan))
154 | self.saver_costs.write(np.array(np.nan))
155 | # ===================== END SAVING ===================================
156 |
157 | # Broadcast fitness:
158 | cost_matrix = np.empty((self._size, self.workers_per_rank))
159 | self._comm.Allgather([self.cost_list, self._MPI.FLOAT],
160 | [cost_matrix, self._MPI.FLOAT])
161 |
162 | # Truncate Selection (broadcast genotypes and update members):
163 | order = np.argsort(cost_matrix.flatten())
164 | rank = np.argsort(order)
165 | ranks_and_members_by_performance = rank.reshape(cost_matrix.shape)
166 |
167 | # Apply selection:
168 | self.apply_selection(ranks_and_members_by_performance)
169 |
170 | # ======================= LOGGING =====================================
171 | if self.verbose == 2:
172 | self.logger.debug(f"\nPopulation Members (After Selection):")
173 | self.logger.debug(f"| index | seed | x0 | x1 |")
174 | for index in range(len(self.members)):
175 | seed = self.members[index].genotype
176 | x0 = self.members[index].phenotype[0].round(10)
177 | x1 = self.members[index].phenotype[1].round(10)
178 | self.logger.debug(f"| {index} | {seed} | {x0} | {x1} |")
179 | # ===================== END LOGGING ===================================
180 |
181 | # Apply mutations:
182 | self.update_sigma()
183 | self.apply_mutation(ranks_and_members_by_performance)
184 |
185 | # ======================= LOGGING =====================================
186 | if self.verbose == 2:
187 | self.logger.debug(f"\nPopulation Members (After Mutation):")
188 | self.logger.debug(f"| index | seed | x0 | x1 |")
189 | for index in range(len(self.members)):
190 | seed = self.members[index].genotype
191 | x0 = self.members[index].phenotype[0].round(10)
192 | x1 = self.members[index].phenotype[1].round(10)
193 | self.logger.debug(f"| {index} | {seed} | {x0} | {x1} |")
194 | # ===================== END LOGGING ===================================
195 |
196 | # Next generation:
197 | self.generation += 1
198 |
199 | @ property
200 | def genotypes(self):
201 | genotypes = []
202 | for member in self.members:
203 | genotypes.append(member.genotype)
204 | return np.array(genotypes, dtype=object)
205 |
206 | @ property
207 | def costs(self):
208 | return self.cost_list
209 |
210 | def update_sigma(self):
211 | self.sigma *= self.sigma_decay
212 | if self.sigma < self.sigma_min:
213 | self.sigma = self.sigma_min
--------------------------------------------------------------------------------
/DNE4py/optimizers/deepga/deepga.py:
--------------------------------------------------------------------------------
1 |
2 | from .base import BaseGA
3 | from .mutation import RealMutator
4 | from .selection import TruncatedSelection
5 |
6 |
7 | class TruncatedRealMutatorGA(TruncatedSelection, RealMutator, BaseGA):
8 | pass
9 |
--------------------------------------------------------------------------------
/DNE4py/optimizers/deepga/member.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 |
3 |
4 | class Member:
5 | r'''
6 | Initialization:
7 | initial_phenotype: initial list of parameters
8 | initial_genotype: initial seed and initial sigma
9 |
10 | Internal attributes:
11 | genotype: list of seeds
12 | phenotype: list of parameters
13 |
14 | Methods:
15 | recreate(new_genotype):
16 | update genotype and phenotype from new_genotype
17 | mutate(rng_genes):
18 | create a new gene and update genotype and phenotype
19 | '''
20 |
21 | def __init__(self, initial_phenotype, genotype):
22 |
23 | # Define initial phenotype:
24 | self.initial_phenotype = initial_phenotype
25 | self.phenotype = self.initial_phenotype.copy()
26 |
27 | # Internal attributes:
28 | self.rng = np.random.RandomState()
29 | self.size = len(self.initial_phenotype)
30 |
31 | # Define genotype and phenotype:
32 | self.recreate(genotype)
33 |
34 | def mutate(self, rng_genes, sigma):
35 | r'''create a new gene and update genotype and phenotype'''
36 |
37 | # Increase genotype:
38 | seed = rng_genes.randint(0, 2 ** 32 - 1)
39 | self.genotype.append([seed, sigma])
40 |
41 | # Mutate phenotype:
42 | self.rng.seed(seed)
43 | self.phenotype += self.rng.randn(self.size) * sigma
44 |
45 |
46 | def recreate(self, new_genotype):
47 | r'''update genotype and phenotype from new_genotype'''
48 |
49 | # Set genotype:
50 | self.genotype = new_genotype[:]
51 |
52 | # Set phenotype:
53 | self.phenotype[:] = self.initial_phenotype[:]
54 | for seed, sigma in self.genotype:
55 | self.rng.seed(seed)
56 | self.phenotype += self.rng.randn(self.size) * sigma
--------------------------------------------------------------------------------
/DNE4py/optimizers/deepga/mutation.py:
--------------------------------------------------------------------------------
1 | import random
2 |
3 | import numpy as np
4 |
5 | from .base import BaseGA
6 | from .member import Member
7 |
8 | class RealMutator(BaseGA):
9 |
10 | def mutator_initialize(self):
11 |
12 | # Global Random Generator:
13 | self.rgn_global = random.Random(self.global_seed)
14 |
15 | # Global seedmatrix:
16 | self.seedmatrix = np.zeros((self._size, self.workers_per_rank), dtype=np.int64)
17 | for i in range(self._size):
18 | row = self.rgn_global.sample(range(2**32 - 1), self.workers_per_rank)
19 | self.seedmatrix[i] = row
20 |
21 | # Generate Genes Random Generator:
22 | self.rng_genes_list = [np.random.RandomState(self.seedmatrix[self._rank, i]) for i in range(self.workers_per_rank)]
23 |
24 | # Generate Initial genes
25 | self.initial_genotypes = []
26 | for i in range(self.workers_per_rank):
27 | initial_genotype = self.rng_genes_list[i].randint(2**32 - 1)
28 | self.initial_genotypes.append([[initial_genotype, self.sigma]])
29 |
30 | self.members = [Member(initial_phenotype=self.initial_guess,
31 | genotype=self.initial_genotypes[i])
32 | for i in range(self.workers_per_rank)]
33 |
34 | def apply_mutation(self, ranks_and_members_by_performance):
35 | """
36 | Preserve the top num_elite members, mutate the rest
37 | """
38 | no_elite_mask = ranks_and_members_by_performance >= self.num_elite
39 |
40 | row, column = np.where(no_elite_mask)
41 | no_elite_tuples = tuple(zip(row, column))
42 |
43 | # ======================= LOGGING =====================================
44 | if self.verbose == 2:
45 | self.logger.debug(f"\nSelected Elite:")
46 | self.logger.debug(f"| rank | member_id |")
47 | elite_mask = np.invert(no_elite_mask)
48 | row, column = np.where(elite_mask)
49 | elite_indexes = tuple(zip(row, column))
50 | for rank, member_id in elite_indexes:
51 | self.logger.debug(f"| {rank} | {member_id} |")
52 | # ===================== END LOGGING ===================================
53 |
54 | for rank, member_id in no_elite_tuples:
55 | if rank == self._rank:
56 | rng_genes = self.rng_genes_list[member_id]
57 | self.members[member_id].mutate(rng_genes, self.sigma)
58 |
--------------------------------------------------------------------------------
/DNE4py/optimizers/deepga/selection.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 |
3 | from .base import BaseGA
4 |
5 | class TruncatedSelection(BaseGA):
6 |
7 | def selection_initialize(self):
8 |
9 | # Init:
10 | self.rgn_selection = np.random.RandomState(self.global_seed)
11 |
12 | if self.num_elite > self.num_parents:
13 | raise AssertionError("Number of elite has to be less than the"
14 | " number of parents")
15 |
16 | def apply_selection(self, ranks_and_members_by_performance):
17 |
18 | # 1 - Define parents and no parents:
19 | parents_mask = ranks_and_members_by_performance < self.num_parents
20 | no_parents_mask = np.invert(parents_mask)
21 |
22 | row, column = np.where(parents_mask)
23 | parents_indexes = tuple(zip(row, column))
24 | parent_options = range(len(parents_indexes))
25 |
26 | row, column = np.where(no_parents_mask)
27 | no_parents_indexes = tuple(zip(row, column))
28 |
29 | # ======================= LOGGING =====================================
30 | if self.verbose == 2:
31 | self.logger.debug(f"\nSelected Parents:")
32 | self.logger.debug(f"| rank | member_id |")
33 | for rank, member_id in parents_indexes:
34 | self.logger.debug(f"| {rank} | {member_id} |")
35 | # ===================== END LOGGING ===================================
36 |
37 | # 2 - Update member for each rank:
38 | # Build message_list:
39 | messenger_list = []
40 | for dest_rank, dest_member_id in no_parents_indexes:
41 | choice = self.rgn_selection.choice(parent_options)
42 | src_rank, src_member_id = parents_indexes[choice]
43 | messenger_list.append((src_rank,
44 | src_member_id,
45 | dest_rank,
46 | dest_member_id))
47 |
48 | # ======================= LOGGING =====================================
49 | if self.verbose == 2:
50 | self.logger.debug(f"\nSelection Messenger List")
51 | self.logger.debug(f"| src_rank | src_member_id | dest_rank | dest_member_id |")
52 | for (src_rank,
53 | src_member_id,
54 | dest_rank,
55 | dest_member_id) in messenger_list:
56 | self.logger.debug(f"| {src_rank} | {src_member_id} | {dest_rank} | {dest_member_id} |")
57 | # ===================== END LOGGING ===================================
58 |
59 | # Dispatch message_list:
60 | for (src_rank,
61 | src_member_id,
62 | dest_rank,
63 | dest_member_id) in messenger_list:
64 | if src_rank != dest_rank: # Need send to MPI
65 | if src_rank == self._rank:
66 | message = (self.members[src_member_id].genotype,
67 | dest_member_id)
68 | self._comm.send(message, dest=dest_rank)
69 | elif dest_rank == self._rank:
70 | new_genotype, member_id = self._comm.recv(source=src_rank)
71 | self.members[member_id].recreate(new_genotype)
72 | else:
73 | if (self._rank == src_rank):
74 | new_genotype = self.members[src_member_id].genotype
75 | member_id = dest_member_id
76 | self.members[member_id].recreate(new_genotype)
77 |
--------------------------------------------------------------------------------
/DNE4py/optimizers/optimizer.py:
--------------------------------------------------------------------------------
1 | r"""
2 | Abstract base module for all DNE4py optimizers.
3 | """
4 |
5 | from abc import ABC, abstractmethod
6 |
7 |
8 | class Optimizer(ABC):
9 | r"""Abstract base class for optimizers."""
10 |
11 | @abstractmethod
12 | def run(self, objective_function, steps):
13 | r"""Optimize the objective function.
14 |
15 | Args:
16 | objective_function: function to be optimized
17 | steps: Number of iterations
18 | """
19 | pass
20 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/cmaes/__init__.py:
--------------------------------------------------------------------------------
1 | from .cmaes import CMAES
2 |
3 | __all__ = [
4 | "CMAES",
5 | ]
6 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/cmaes/cmaes.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 |
3 | from ..optimizer import Optimizer
4 | from DNE4py.utils.mpi_extensions import MPISaver, MPILogger
5 | import cma
6 |
7 |
8 | class CMAES(Optimizer):
9 |
10 | def __init__(self, objective_function, config):
11 |
12 | super().__init__(objective_function, config)
13 |
14 | # self.objective_function
15 | # self.initial_guess
16 | # self.workers_per_rank
17 | # self.sigma
18 | # self.global_seed
19 | # self.save
20 | # self.verbose
21 | # self.output_folder
22 |
23 | # Initiate MPI
24 | from mpi4py import MPI
25 | self._MPI = MPI
26 | self._comm = MPI.COMM_WORLD
27 | self._size = self._comm.Get_size()
28 | self._rank = self._comm.Get_rank()
29 |
30 | # Input:
31 | #self.objective_function = kwargs.get('objective_function')
32 | #self.initial_guess = kwargs.get('initial_guess')
33 | #self.sigma = kwargs.get('sigma')
34 | #self.global_seed = kwargs.get('seed')
35 |
36 | #self.save = kwargs.get('save', 0)
37 | #self.verbose = kwargs.get('verbose', 0)
38 | #self.output_folder = kwargs.get('output_folder', 'DNE4py_result')
39 |
40 | #self.workers_per_rank = kwargs.get('workers_per_rank')
41 |
42 | # Internal:
43 | self.population_size = self._size * self.workers_per_rank
44 | self.optimizer = cma.CMAEvolutionStrategy(self.initial_guess, self.sigma, {'verb_disp': 0, 'seed': self.global_seed, 'popsize': self.population_size})
45 | self.my_ids = np.array_split(range(self.population_size), self._size)[self._rank]
46 | self.generation = 0
47 |
48 | # Logger and DataCollector for MPI:
49 | if (self.verbose == 2) or (self.save > 0):
50 |
51 | if self.verbose == 2:
52 | self.logger = MPILogger(self.output_folder, 'debug_logger', self._rank)
53 | if self.save > 0:
54 | self.mpidata_genotypes = MPIData(self.output_folder,
55 | 'genotypes',
56 | self._rank)
57 | self.mpidata_costs = MPIData(self.output_folder,
58 | 'costs',
59 | self._rank)
60 | self.mpidata_initialguess = MPIData(self.output_folder,
61 | 'initial_guess',
62 | 0)
63 |
64 | def run(self, steps):
65 |
66 | for _ in range(steps):
67 |
68 | # =================== LOGGING =====================================
69 | if self.verbose > 0 and self._rank == 0:
70 | print(f"Generation: {self.generation}/{steps}")
71 | elif self.verbose == 2:
72 | self.logger.debug(f"\nGeneration: {self.generation}/{steps}")
73 | # =================== LOGGING =====================================
74 |
75 | self.step()
76 |
77 | def step(self):
78 |
79 | # Ask solutions:
80 | self.solutions = np.array(self.optimizer.ask())
81 |
82 | # Evaluate only your own solutions:
83 | self.my_solutions = self.solutions[self.my_ids]
84 | self.cost_list = np.zeros(self.workers_per_rank)
85 | for i in range(len(self.cost_list)):
86 | self.cost_list[i] = self.objective_function(self.my_solutions[i])
87 |
88 | # Broadcast fitness:
89 | self.all_costs = np.empty(self.population_size)
90 | self._comm.Allgather([self.cost_list, self._MPI.FLOAT],
91 | [self.all_costs, self._MPI.FLOAT])
92 |
93 | # Tell solutions and evaluations:
94 | self.optimizer.tell(self.solutions, self.all_costs)
95 |
96 | # Save:
97 | if (self.save > 0) and (self.generation % self.save == 0):
98 | self.mpi_save(self.generation)
99 |
100 | # ======================= LOGGING =====================================
101 | if self.verbose == 2:
102 | self.logger.debug(f"\n| index | x0 | x1 | y |")
103 | for index, sol in enumerate(self.solutions):
104 |
105 | x0 = sol[0].round(10)
106 | x1 = sol[1].round(10)
107 | y = self.all_costs[index].round(10)
108 | if index in self.my_ids:
109 | self.logger.debug(f"| {index} | {x0} | {x1} | {y} | X |")
110 | else:
111 | self.logger.debug(f"| {index} | {x0} | {x1} | {y} |")
112 | # ===================== END LOGGING ===================================
113 |
114 | # Next generation:
115 | self.generation += 1
116 |
117 | # Evaluate member:
118 | # self.cost_list = np.zeros(self.workers_per_rank)
119 | # for i in range(len(self.cost_list)):
120 | # self.cost_list[i] = self.objective_function(self.members[i].phenotype)
121 |
122 |
123 | # # Save:
124 | # if (self.save > 0) and (self.generation % self.save == 0):
125 | # self.mpi_save(self.generation)
126 |
127 | # # Broadcast fitness:
128 | # cost_matrix = np.empty((self._size, self.workers_per_rank))
129 | # self._comm.Allgather([self.cost_list, self._MPI.FLOAT],
130 | # [cost_matrix, self._MPI.FLOAT])
131 |
132 | # # Truncate Selection (broadcast genotypes and update members):
133 | # order = np.argsort(cost_matrix.flatten())
134 | # rank = np.argsort(order)
135 | # ranks_and_members_by_performance = rank.reshape(cost_matrix.shape)
136 |
137 | # Apply selection:
138 | # self.apply_selection(ranks_and_members_by_performance)
139 |
140 | # # ======================= LOGGING =====================================
141 | # if self.verbose == 2:
142 | # self.logger.debug(f"\nPopulation Members (After Selection):")
143 | # self.logger.debug(f"| index | seed | x0 | x1 |")
144 | # for index in range(len(self.members)):
145 | # seed = self.members[index].genotype
146 | # x0 = self.members[index].phenotype[0].round(10)
147 | # x1 = self.members[index].phenotype[1].round(10)
148 | # self.logger.debug(f"| {index} | {seed} | {x0} | {x1} |")
149 | # ===================== END LOGGING ===================================
150 |
151 | # Apply mutations:
152 | # self.apply_mutation(ranks_and_members_by_performance)
153 |
154 | # # ======================= LOGGING =====================================
155 | # if self.verbose == 2:
156 | # self.logger.debug(f"\nPopulation Members (After Mutation):")
157 | # self.logger.debug(f"| index | seed | x0 | x1 |")
158 | # for index in range(len(self.members)):
159 | # seed = self.members[index].genotype
160 | # x0 = self.members[index].phenotype[0].round(10)
161 | # x1 = self.members[index].phenotype[1].round(10)
162 | # self.logger.debug(f"| {index} | {seed} | {x0} | {x1} |")
163 | # # ===================== END LOGGING ===================================
164 |
165 | # # Next generation:
166 | # self.generation += 1
167 |
168 | def mpi_save(self, s):
169 |
170 | self.mpidata_genotypes.write(self.my_solutions)
171 | self.mpidata_costs.write(self.cost_list)
172 | if s == 0:
173 | self.mpidata_initialguess.write(self.initial_guess)
174 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/deepga2/__init__.py:
--------------------------------------------------------------------------------
1 | from .deepga import TruncatedRealMutatorCompactGA, TruncatedRealMutatorCompositeGA
2 |
3 | __all__ = [
4 | "TruncatedRealMutatorCompactGA",
5 | "TruncatedRealMutatorCompositeGA"
6 | ]
7 |
8 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/deepga2/base.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 | import pickle
3 | import random
4 |
5 | from abc import ABC, abstractmethod
6 |
7 | from DNE4py.utils import MPIData, MPILogger
8 |
9 | from DNE4py.optimizers.optimizer import Optimizer
10 | #from .mutation import RealMutator
11 | #from .selection import TruncatedSelection
12 |
13 | class BaseGA(Optimizer):
14 |
15 | def __init__(self, objective_function, config):
16 |
17 | super().__init__(objective_function, config)
18 |
19 | # self.initial_guess
20 | # workers_per_rank
21 | # num_elite
22 | # num_parents
23 | # sigma
24 | # seed
25 | # save
26 | # verbose
27 | # output_folder
28 |
29 | # Initiate MPI
30 | from mpi4py import MPI
31 | self._MPI = MPI
32 | self._comm = MPI.COMM_WORLD
33 | self._size = self._comm.Get_size()
34 | self._rank = self._comm.Get_rank()
35 |
36 | # mutator and selection
37 | self.mutator_initialize()
38 | self.selection_initialize()
39 |
40 | # Internal:
41 | self.generation = 0
42 | self.population_size = self._size * self.workers_per_rank
43 |
44 | # Logger and DataCollector for MPI:
45 | if (self.verbose == 2) or (self.save > 0):
46 |
47 | if self.verbose == 2:
48 | self.logger = MPILogger(self.output_folder, 'debug_logger', self._rank)
49 | if self.save > 0:
50 | self.mpidata_genotypes = MPIData(self.output_folder,
51 | 'genotypes',
52 | self._rank)
53 | self.mpidata_costs = MPIData(self.output_folder,
54 | 'costs',
55 | self._rank)
56 | self.mpidata_initialguess = MPIData(self.output_folder,
57 | 'initial_guess',
58 | 0)
59 | self.mpidata_rankings = MPIData(self.output_folder,
60 | 'rankings',
61 | self._rank)
62 | #@abstractmethod
63 | def apply_ranking(self, ranks_by_performance):
64 | pass
65 |
66 | #@abstractmethod
67 | def apply_selection(self, ranks_by_performance):
68 | pass
69 |
70 | #@abstractmethod
71 | def apply_mutation(self, ranks_by_performance):
72 | pass
73 |
74 | def run(self, steps):
75 |
76 | for _ in range(steps):
77 |
78 | # =================== LOGGING =====================================
79 | if self.verbose > 0 and self._rank == 0:
80 | print(f"Generation: {self.generation}/{steps}")
81 | elif self.verbose == 2:
82 | self.logger.debug(f"\nGeneration: {self.generation}/{steps}")
83 | # =================== LOGGING =====================================
84 |
85 | self.step()
86 |
87 | def step(self):
88 | # Apply ranking :
89 | ranks_and_members_by_performance = self.apply_ranking()
90 |
91 | # Apply selection:
92 | self.apply_selection(ranks_and_members_by_performance)
93 |
94 | # ======================= LOGGING =====================================
95 | if self.verbose == 2:
96 | self.logger.debug(f"\nPopulation Members (After Selection):")
97 | self.logger.debug(f"| index | seed | x0 | x1 |")
98 | for index in range(len(self.members)):
99 | seed = self.members[index].genotype
100 | x0 = self.members[index].phenotype[0].round(10)
101 | x1 = self.members[index].phenotype[1].round(10)
102 | self.logger.debug(f"| {index} | {seed} | {x0} | {x1} |")
103 | # ===================== END LOGGING ===================================
104 |
105 | # Apply mutations:
106 | self.apply_mutation(ranks_and_members_by_performance)
107 |
108 | # ======================= LOGGING =====================================
109 | if self.verbose == 2:
110 | self.logger.debug(f"\nPopulation Members (After Mutation):")
111 | self.logger.debug(f"| index | seed | x0 | x1 |")
112 | for index in range(len(self.members)):
113 | seed = self.members[index].genotype
114 | x0 = self.members[index].phenotype[0].round(10)
115 | x1 = self.members[index].phenotype[1].round(10)
116 | self.logger.debug(f"| {index} | {seed} | {x0} | {x1} |")
117 | # ===================== END LOGGING ===================================
118 |
119 | # Next generation:
120 | self.generation += 1
121 |
122 | def mpi_save(self, s):
123 |
124 | self.mpidata_genotypes.write(self.genotypes)
125 | self.mpidata_costs.write(self.costs)
126 | if s == 0:
127 | self.mpidata_initialguess.write(self.initial_guess)
128 |
129 | @ property
130 | def genotypes(self):
131 | genotypes = []
132 | for member in self.members:
133 | genotypes.append(member.genotype)
134 | return np.array(genotypes)
135 |
136 | @ property
137 | def costs(self):
138 | return self.cost_list
139 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/deepga2/deepga.py:
--------------------------------------------------------------------------------
1 |
2 | from .base import BaseGA
3 | from .mutation import RealMutator
4 | from .selection import TruncatedSelection
5 | from .ranking import CompactRanking, CompositeRanking
6 |
7 |
8 | class TruncatedRealMutatorCompactGA(TruncatedSelection, RealMutator, CompactRanking, BaseGA):
9 | pass
10 |
11 | class TruncatedRealMutatorCompositeGA(TruncatedSelection, RealMutator, CompositeRanking, BaseGA):
12 | pass
13 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/deepga2/member.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 |
3 |
4 | class Member:
5 | r'''
6 | Initialization:
7 | initial_phenotype: initial list of parameters
8 | initial_gene: initial seed
9 | sigma: parameter to apply on mutations
10 |
11 | Internal attributes:
12 | genotype: list of seeds
13 | phenotype: list of parameters
14 |
15 | Methods:
16 | recreate(new_genotype):
17 | update genotype and phenotype from new_genotype
18 | mutate(rng_genes):
19 | create a new gene and update genotype and phenotype
20 | '''
21 |
22 | def __init__(self,
23 | initial_phenotype,
24 | initial_genotype,
25 | sigma):
26 |
27 | # Define initial phenotype:
28 | self.initial_phenotype = initial_phenotype
29 | self.phenotype = self.initial_phenotype.copy()
30 |
31 | #self.initial_genotype = initial_genotype.copy()
32 |
33 | # Internal attributes:
34 | self.rng = np.random.RandomState()
35 | self.size, self.sigma = len(self.phenotype), sigma
36 |
37 | # Define genotype and phenotype:
38 | self.recreate(initial_genotype)
39 |
40 | def mutate(self, rng_genes):
41 | r'''create a new gene and update genotype and phenotype'''
42 |
43 | # Increase genotype:
44 | seed = rng_genes.randint(0, 2 ** 32 - 1)
45 | self.genotype.append(seed)
46 |
47 | # Mutate phenotype:
48 | self.rng.seed(seed)
49 | self.phenotype += self.rng.randn(self.size) * self.sigma
50 |
51 | def recreate(self, new_genotype):
52 | r'''update genotype and phenotype from new_genotype'''
53 |
54 | # Set genotype:
55 | self.genotype = new_genotype[:]
56 |
57 | # Set phenotype:
58 | self.phenotype[:] = self.initial_phenotype[:]
59 |
60 | for seed in self.genotype:
61 | self.rng.seed(seed)
62 | self.phenotype += self.rng.randn(self.size) * self.sigma
63 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/deepga2/mutation.py:
--------------------------------------------------------------------------------
1 | import random
2 |
3 | import numpy as np
4 |
5 | from .base import BaseGA
6 | from .member import Member
7 |
8 | class RealMutator(BaseGA):
9 |
10 | def mutator_initialize(self):
11 |
12 | # Global Random Generator:
13 | self.rgn_global = random.Random(self.global_seed)
14 |
15 | # Global seedmatrix:
16 | self.seedmatrix = np.zeros((self._size, self.workers_per_rank), dtype=np.int64)
17 | for i in range(self._size):
18 | row = self.rgn_global.sample(range(2**32 - 1), self.workers_per_rank)
19 | self.seedmatrix[i] = row
20 |
21 | # Generate Genes Random Generator:
22 | self.rng_genes_list = [np.random.RandomState(self.seedmatrix[self._rank, i]) for i in range(self.workers_per_rank)]
23 |
24 | # Generate Initial genes
25 | self.initial_genotypes = []
26 | for i in range(self.workers_per_rank):
27 | initial_genotype = self.rng_genes_list[i].randint(2**32 - 1)
28 | self.initial_genotypes.append([initial_genotype])
29 |
30 | self.members = [Member(initial_phenotype=self.initial_guess,
31 | initial_genotype=self.initial_genotypes[i],
32 | sigma=self.sigma)
33 | for i in range(self.workers_per_rank)]
34 |
35 | def apply_mutation(self, ranks_and_members_by_performance):
36 | """
37 | Preserve the top num_elite members, mutate the rest
38 | """
39 | no_elite_mask = ranks_and_members_by_performance >= self.num_elite
40 |
41 | row, column = np.where(no_elite_mask)
42 | no_elite_tuples = tuple(zip(row, column))
43 |
44 | # ======================= LOGGING =====================================
45 | if self.verbose == 2:
46 | self.logger.debug(f"\nSelected Elite:")
47 | self.logger.debug(f"| rank | member_id |")
48 | elite_mask = np.invert(no_elite_mask)
49 | row, column = np.where(elite_mask)
50 | elite_indexes = tuple(zip(row, column))
51 | for rank, member_id in elite_indexes:
52 | self.logger.debug(f"| {rank} | {member_id} |")
53 | # ===================== END LOGGING ===================================
54 |
55 | for rank, member_id in no_elite_tuples:
56 | if rank == self._rank:
57 | rng_genes = self.rng_genes_list[member_id]
58 | self.members[member_id].mutate(rng_genes)
59 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/deepga2/ranking.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 |
3 | from .base import BaseGA
4 | from DNE4py.utils import MPIData
5 | # class Ranking(BaseGA):
6 | # def ranking_initialize(self):
7 | # self.n_tasks = len(self.objective_function(self.members[0].phenotype))
8 |
9 | # def apply_ranking(self):
10 |
11 | # # Evaluate member:
12 | # self.cost_list = np.zeros((self.workers_per_rank, self.n_tasks)) # (W,T) vs (W)
13 | # for i in range(len(self.cost_list)):
14 | # self.cost_list[i] = self.objective_function(self.members[i].phenotype)
15 |
16 |
17 | # # ======================= LOGGING =====================================
18 | # if self.verbose == 2:
19 | # self.logger.debug(f"\nPopulation Members (Initial):")
20 | # self.logger.debug(f"| index | seed | x0 | x1 | y |")
21 | # for index in range(len(self.members)):
22 | # seed = self.members[index].genotype
23 | # x0 = self.members[index].phenotype[0].round(10)
24 | # x1 = self.members[index].phenotype[1].round(10)
25 | # y = self.cost_list[index].round(10)
26 | # self.logger.debug(f"| {index} | {seed} | {x0} | {x1} | {y} |")
27 | # # ===================== END LOGGING ===================================
28 | # # Save:
29 | # if (self.save > 0) and (self.generation % self.save == 0):
30 | # self.mpi_save(self.generation)
31 |
32 | # # Broadcast fitness:
33 | # cost_matrix = np.empty((self._size, self.workers_per_rank)) # (S,W,T) vs (S,W)
34 | # self._comm.Allgather([self.cost_list, self._MPI.FLOAT],
35 | # [cost_matrix, self._MPI.FLOAT])
36 |
37 | # order = self.ordering(cost_matrix) # (S*W,T) vs (S*W)
38 | # rank = np.argsort(order)
39 | # ranks_and_members_by_performance = rank.reshape(self._size, self.workers_per_rank) #--#
40 |
41 | # return ranks_and_members_by_performance
42 |
43 |
44 | # def ordering(cost_matrix):
45 | # pass
46 |
47 |
48 |
49 | class CompactRanking(BaseGA):
50 | def apply_ranking(self):
51 |
52 | # Evaluate member:
53 | self.cost_list = np.zeros(self.workers_per_rank)
54 | for i in range(len(self.cost_list)):
55 | self.cost_list[i] = self.objective_function(self.members[i].phenotype)
56 |
57 |
58 | # ======================= LOGGING =====================================
59 | if self.verbose == 2:
60 | self.logger.debug(f"\nPopulation Members (Initial):")
61 | self.logger.debug(f"| index | seed | x0 | x1 | y |")
62 | for index in range(len(self.members)):
63 | seed = self.members[index].genotype
64 | x0 = self.members[index].phenotype[0].round(10)
65 | x1 = self.members[index].phenotype[1].round(10)
66 | y = self.cost_list[index].round(10)
67 | self.logger.debug(f"| {index} | {seed} | {x0} | {x1} | {y} |")
68 | # ===================== END LOGGING ===================================
69 | # Save:
70 | if (self.save > 0) and (self.generation % self.save == 0):
71 | self.mpi_save(self.generation)
72 |
73 | # Broadcast fitness:
74 | cost_matrix = np.empty((self._size, self.workers_per_rank))
75 | self._comm.Allgather([self.cost_list, self._MPI.FLOAT],
76 | [cost_matrix, self._MPI.FLOAT])
77 |
78 | # Truncate Selection (broadcast genotypes and update members):
79 | order = np.argsort(cost_matrix.flatten())
80 | rank = np.argsort(order)
81 | ranks_and_members_by_performance = rank.reshape(cost_matrix.shape)
82 | return ranks_and_members_by_performance
83 |
84 |
85 | class CompositeRanking(BaseGA):
86 |
87 | def apply_ranking(self):
88 |
89 | # Evaluate member:
90 | self.cost_list = [] #--#
91 | for i in range(self.workers_per_rank):
92 | self.cost_list.append(self.objective_function(self.members[i].phenotype))
93 | self.cost_list = np.array(self.cost_list)
94 | self.n_tasks = self.cost_list.shape[1]
95 |
96 | # ======================= LOGGING =====================================
97 | if self.verbose == 2:
98 | self.logger.debug(f"\nPopulation Members (Initial):")
99 | self.logger.debug(f"| index | seed | x0 | x1 | y |")
100 | for index in range(len(self.members)):
101 | seed = self.members[index].genotype
102 | x0 = self.members[index].phenotype[0].round(10)
103 | x1 = self.members[index].phenotype[1].round(10)
104 | y = self.cost_list[index].round(10)
105 | self.logger.debug(f"| {index} | {seed} | {x0} | {x1} | {y} |")
106 | # ===================== END LOGGING ===================================
107 |
108 |
109 | if (self.save > 0) and (self.generation % self.save == 0):
110 | self.mpi_save(self.generation)
111 | # Broadcast fitness:
112 | cost_matrix = np.empty((self._size, self.workers_per_rank, self.n_tasks)) #--#
113 | self._comm.Allgather([self.cost_list, self._MPI.FLOAT],
114 | [cost_matrix, self._MPI.FLOAT])
115 |
116 | # Compute "on instances" average rankings
117 | average_ranking = self.ranking(cost_matrix)
118 |
119 | # Save:
120 |
121 | if (self.save > 0) and (self.generation % self.save == 0):
122 | # add the specific ranking data
123 | self.mpidata_rankings.write(average_ranking)
124 |
125 | # Truncate Selection (broadcast genotypes and update members):
126 | order = np.argsort(average_ranking) #--#
127 | rank = np.argsort(order)
128 | ranks_and_members_by_performance = rank.reshape(self._size, self.workers_per_rank) #--#
129 | return ranks_and_members_by_performance
130 |
131 |
132 |
133 | def ranking(self, cost_matrix):
134 | """
135 | @TODO : gèrer les ex-aequo
136 | """
137 | data = cost_matrix.reshape(self._size * self.workers_per_rank, self.n_tasks)
138 |
139 | for inst in range(data.shape[1]):
140 | data[:,inst] = rankdata(data[:,inst])
141 | return np.mean(data, axis=1)
142 |
143 |
144 | # Move somewhere else
145 | # https://github.com/numbbo/coco/blob/master/code-postprocessing/cocopp/toolsstats.py
146 | def rankdata(a):
147 | """Ranks the data in a, dealing with ties appropriately.
148 | Equal values are assigned a rank that is the average of the ranks that
149 | would have been otherwise assigned to all of the values within that set.
150 | Ranks begin at 1, not 0.
151 | Example:
152 | In [15]: stats.rankdata([0, 2, 2, 3])
153 | Out[15]: array([ 1. , 2.5, 2.5, 4. ])
154 | Parameters:
155 | - *a* : array
156 | This array is first flattened.
157 | Returns:
158 | An array of length equal to the size of a, containing rank scores.
159 | """
160 | a = np.ravel(a)
161 | n = len(a)
162 | svec, ivec = fastsort(a)
163 | sumranks = 0
164 | dupcount = 0
165 | newarray = np.zeros(n, float)
166 | for i in range(n):
167 | sumranks += i
168 | dupcount += 1
169 | if i == n - 1 or svec[i] != svec[i + 1]:
170 | averank = sumranks / float(dupcount) + 1
171 | for j in range(i - dupcount + 1, i + 1):
172 | newarray[ivec[j]] = averank
173 | sumranks = 0
174 | dupcount = 0
175 | return newarray
176 | # Dimensions
177 | # Calcul des ranks (non trivial dans le cas composite)
178 |
179 | def fastsort(a):
180 | it = np.argsort(a)
181 | as_ = a[it]
182 | return as_, it
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/deepga2/selection.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 |
3 | from .base import BaseGA
4 |
5 | class TruncatedSelection(BaseGA):
6 |
7 | def selection_initialize(self):
8 |
9 | # Init:
10 | self.rgn_selection = np.random.RandomState(self.global_seed)
11 |
12 | if self.num_elite > self.num_parents:
13 | raise AssertionError("Number of elite has to be less than the"
14 | " number of parents")
15 |
16 | def apply_selection(self, ranks_and_members_by_performance):
17 |
18 | # 1 - Define parents and no parents:
19 | parents_mask = ranks_and_members_by_performance < self.num_parents
20 | no_parents_mask = np.invert(parents_mask)
21 |
22 | row, column = np.where(parents_mask)
23 | parents_indexes = tuple(zip(row, column))
24 | parent_options = range(len(parents_indexes))
25 |
26 | row, column = np.where(no_parents_mask)
27 | no_parents_indexes = tuple(zip(row, column))
28 |
29 | # ======================= LOGGING =====================================
30 | if self.verbose == 2:
31 | self.logger.debug(f"\nSelected Parents:")
32 | self.logger.debug(f"| rank | member_id |")
33 | for rank, member_id in parents_indexes:
34 | self.logger.debug(f"| {rank} | {member_id} |")
35 | # ===================== END LOGGING ===================================
36 |
37 | # 2 - Update member for each rank:
38 | # Build message_list:
39 | messenger_list = []
40 | for dest_rank, dest_member_id in no_parents_indexes:
41 | choice = self.rgn_selection.choice(parent_options)
42 | src_rank, src_member_id = parents_indexes[choice]
43 | messenger_list.append((src_rank,
44 | src_member_id,
45 | dest_rank,
46 | dest_member_id))
47 |
48 | # ======================= LOGGING =====================================
49 | if self.verbose == 2:
50 | self.logger.debug(f"\nSelection Messenger List")
51 | self.logger.debug(f"| src_rank | src_member_id | dest_rank | dest_member_id |")
52 | for (src_rank,
53 | src_member_id,
54 | dest_rank,
55 | dest_member_id) in messenger_list:
56 | self.logger.debug(f"| {src_rank} | {src_member_id} | {dest_rank} | {dest_member_id} |")
57 | # ===================== END LOGGING ===================================
58 |
59 | # Dispatch message_list:
60 | for (src_rank,
61 | src_member_id,
62 | dest_rank,
63 | dest_member_id) in messenger_list:
64 | if src_rank != dest_rank: # Need send to MPI
65 | if src_rank == self._rank:
66 | message = (self.members[src_member_id].genotype,
67 | dest_member_id)
68 | self._comm.send(message, dest=dest_rank)
69 | elif dest_rank == self._rank:
70 | new_genotype, member_id = self._comm.recv(source=src_rank)
71 | self.members[member_id].recreate(new_genotype)
72 | else:
73 | if (self._rank == src_rank):
74 | new_genotype = self.members[src_member_id].genotype
75 | member_id = dest_member_id
76 | self.members[member_id].recreate(new_genotype)
77 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/random/__init__.py:
--------------------------------------------------------------------------------
1 | from .batch import BatchRandomSearch
2 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_optimizers/random/batch.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 | #import pickle
3 | #import random
4 |
5 | #from abc import ABC, abstractmethod
6 |
7 | from ..optimizer import Optimizer
8 | from DNE4py.utils.mpi_extensions import MPISaver, MPILogger
9 |
10 | class BatchRandomSearch(Optimizer):
11 |
12 | def __init__(self, objective_function, config):
13 |
14 | super().__init__(objective_function, config)
15 |
16 | # self.objective_function
17 | # self.dim
18 | # self.workers_per_rank
19 | # self.bounds
20 | # self.global_seed
21 | # self.save
22 | # self.verbose
23 | # self.output_folder
24 |
25 | # Initiate MPI
26 | from mpi4py import MPI
27 | self._MPI = MPI
28 | self._comm = MPI.COMM_WORLD
29 | self._size = self._comm.Get_size()
30 | self._rank = self._comm.Get_rank()
31 |
32 | # Input:
33 | #self.objective_function = kwargs.get('objective_function')
34 | #self.initial_guess = kwargs.get('initial_guess')
35 | #self.bounds = kwargs.get('bounds')
36 | #self.global_seed = kwargs.get('seed')
37 |
38 | #self.save = kwargs.get('save', 0)
39 | #self.verbose = kwargs.get('verbose', 0)
40 | #self.output_folder = kwargs.get('output_folder', 'DNE4py_result')
41 |
42 | #self.workers_per_rank = kwargs.get('workers_per_rank')
43 |
44 | # Internal:
45 | self.population_size = self._size * self.workers_per_rank
46 | self.generator = np.random.RandomState(self.global_seed)
47 | self.my_ids = np.array_split(range(self.population_size), self._size)[self._rank]
48 | self.generation = 0
49 |
50 | # Logger and DataCollector for MPI:
51 | if (self.verbose == 2) or (self.save > 0):
52 |
53 | if self.verbose == 2:
54 | self.logger = MPILogger(self.output_folder, 'debug_logger', self._rank)
55 | if self.save > 0:
56 | self.mpidata_genotypes = MPIData(self.output_folder,
57 | 'genotypes',
58 | self._rank)
59 | self.mpidata_costs = MPIData(self.output_folder,
60 | 'costs',
61 | self._rank)
62 |
63 | def run(self, steps):
64 |
65 | for _ in range(steps):
66 |
67 | # =================== LOGGING =====================================
68 | if self.verbose > 0 and self._rank == 0:
69 | print(f"Generation: {self.generation}/{steps}")
70 | elif self.verbose == 2:
71 | self.logger.debug(f"\nGeneration: {self.generation}/{steps}")
72 | # =================== LOGGING =====================================
73 |
74 | self.step()
75 |
76 | def step(self):
77 |
78 | # Get solutions:
79 | self.solutions = self.generator.uniform(self.bounds[0], self.bounds[1], (self.population_size, self.dim))
80 |
81 | # Evaluate only your own solutions:
82 | self.my_solutions = self.solutions[self.my_ids]
83 | self.cost_list = np.zeros(self.workers_per_rank)
84 | for i in range(len(self.cost_list)):
85 | self.cost_list[i] = self.objective_function(self.my_solutions[i])
86 |
87 | # Save:
88 | if (self.save > 0) and (self.generation % self.save == 0):
89 | self.mpi_save(self.generation)
90 |
91 | # ======================= LOGGING =====================================
92 | if self.verbose == 2:
93 | self.logger.debug(f"\n| index | x0 | x1 | y |")
94 | for index, sol in enumerate(self.solutions):
95 |
96 | x0 = sol[0].round(10)
97 | x1 = sol[1].round(10)
98 | y = self.all_costs[index].round(10)
99 | if index in self.my_ids:
100 | self.logger.debug(f"| {index} | {x0} | {x1} | {y} | X |")
101 | else:
102 | self.logger.debug(f"| {index} | {x0} | {x1} | {y} |")
103 | # ===================== END LOGGING ===================================
104 |
105 | # Next generation:
106 | self.generation += 1
107 |
108 | def mpi_save(self, s):
109 |
110 | self.mpidata_genotypes.write(self.my_solutions)
111 | self.mpidata_costs.write(self.cost_list)
112 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_tutorials/tutorials2/README.md:
--------------------------------------------------------------------------------
1 | ## `run.py`
2 | * It changes results folder
3 | ```console
4 | foo@bar:~$ mpiexec -n 2 python3 run.py 3
5 | ```
6 |
7 | ## `pp_run.py`
8 | * It changes pp_results folder
9 | * To Optimize: (Use Optimize Transparency 10%) https://ezgif.com/optimize/
10 |
11 | ```console
12 | foo@bar:~$ python3 pp_run.py 3
13 | foo@bar:~$ cd pp_results/RandomSearch/
14 | foo@bar:~$ convert -layers OptimizeTransparency -delay 20 -loop 0 `ls -v` randomsearch.gif
15 | ```
16 |
17 | ## `render.py`
18 | * It defines the behaviour to render the optimization procedure
19 |
20 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_tutorials/tutorials2/pp_run.py:
--------------------------------------------------------------------------------
1 | import sys
2 | import numpy as np
3 |
4 | from render import deepga_render, composite_deepga_render, cmaes_render, randomsearch_render
5 |
6 |
7 | def objective(x):
8 | result = np.sum(x**2)
9 | return result
10 |
11 | def render_Compact_DeepGA():
12 | sigma = 0.05
13 | num_parents = 3
14 | num_elite = 1
15 | deepga_render("results/DeepGA/TruncatedRealMutatorCompactGA", 20, objective, sigma, num_parents, num_elite)
16 |
17 | def render_Composite_DeepGA():
18 |
19 | n_tasks = 4
20 | sigma = 0.05
21 | num_parents = 3
22 | num_elite = 1
23 | composite_deepga_render("results/DeepGA/TruncatedRealMutatorCompositeGA", 20, objective, sigma, num_parents, num_elite)
24 |
25 | def render_CMAES():
26 | sigma = 0.05
27 | cmaes_render("results/CMAES", 20, objective, sigma)
28 |
29 | def render_RandomSearch():
30 | randomsearch_render("results/RandomSearch", 20, objective)
31 |
32 | # Composite example task
33 | # def multi_objective(x, n_tasks = 4):
34 | # """
35 | # The composite objective is composed of n_tasks functions of the form f:(x,y) -> (x^2 + y^2) * i
36 | # The sum of the component individual objectives is used to evaluate the optimizer
37 | # (we expect it to optimize the average)
38 | # """
39 | # result = np.array([np.sum(x**2 + i*x ) * i for i in range(n_tasks)])
40 | # return sum(result)
41 |
42 |
43 | if "__main__" == __name__:
44 |
45 | # Options:
46 | switcher = {
47 | 0: render_Compact_DeepGA,
48 | 1: render_Composite_DeepGA,
49 | 2: render_CMAES,
50 | 3: render_RandomSearch,
51 | }
52 | renderize = switcher.get(int(sys.argv[1]), lambda: "Invalid index for optimizer")
53 | renderize()
54 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_tutorials/tutorials2/render.py:
--------------------------------------------------------------------------------
1 | # After running these methods, you can generate a compressed .gif in the following command:
2 | # $: convert -layers OptimizeTransparency -delay 20 -loop 0 `ls -v` myimage.gif
3 | # To Optimizer: (Use Optimize Transparency 10%) https://ezgif.com/optimize/
4 |
5 |
6 | import pickle
7 |
8 | from scipy.stats import multivariate_normal, find_repeats
9 | import numpy as np
10 |
11 | from DNE4py.postprocessing.utils import load_mpidata
12 | from DNE4py.optimizers.cmaes import CMAES
13 |
14 |
15 | def randomsearch_render(folder_path, nb_generations, objective):
16 |
17 | import numpy as np
18 | import matplotlib.pyplot as plt
19 |
20 | def start_image():
21 | fig, ax = plt.subplots()
22 | ax.set_xlim([-1, 1])
23 | ax.set_ylim([-1, 1])
24 | ax.set_xticks(np.arange(-1, 1, 0.2))
25 | ax.set_yticks(np.arange(-1, 1, 0.2))
26 | return fig, ax
27 |
28 | # Read Input:
29 | costs = load_mpidata(f"{folder_path}", "costs", nb_generations)
30 | genotypes = load_mpidata(f"{folder_path}", "genotypes", nb_generations)
31 | initial_guess = load_mpidata(f"{folder_path}", "initial_guess", 1)[0]
32 |
33 | # Start figure:
34 | fig, ax = start_image()
35 |
36 | # Plot function:
37 | resolution = 100
38 | x1, x2 = np.meshgrid(np.linspace(-1, 1, resolution), np.linspace(-1, 1, resolution))
39 | X = np.array([x1.flatten(), x2.flatten()]).T
40 | Y = np.array([objective(x) for x in X]).reshape(resolution, resolution)
41 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
42 |
43 | for g in range(nb_generations - 1):
44 |
45 | # Image 1:
46 |
47 | # Plot function:
48 | fig, ax = start_image()
49 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
50 |
51 | # Plot points:
52 | ax.scatter(genotypes[g][:, 0], genotypes[g][:, 1], s=10, c='black')
53 | plt.savefig(f"pp_{folder_path}/{g+1}_1.jpeg")
54 |
55 | def cmaes_render(folder_path, nb_generations, objective, sigma):
56 |
57 | import numpy as np
58 | import matplotlib.pyplot as plt
59 |
60 | def start_image():
61 | fig, ax = plt.subplots()
62 | ax.set_xlim([-1, 1])
63 | ax.set_ylim([-1, 1])
64 | ax.set_xticks(np.arange(-1, 1, 0.2))
65 | ax.set_yticks(np.arange(-1, 1, 0.2))
66 | return fig, ax
67 |
68 | # Read Input:
69 | costs = load_mpidata(f"{folder_path}", "costs", nb_generations)
70 | genotypes = load_mpidata(f"{folder_path}", "genotypes", nb_generations)
71 | initial_guess = load_mpidata(f"{folder_path}", "initial_guess", 1)[0]
72 |
73 | # Start figure:
74 | fig, ax = start_image()
75 |
76 | # Plot function:
77 | resolution = 100
78 | x1, x2 = np.meshgrid(np.linspace(-1, 1, resolution), np.linspace(-1, 1, resolution))
79 | X = np.array([x1.flatten(), x2.flatten()]).T
80 | Y = np.array([objective(x) for x in X]).reshape(resolution, resolution)
81 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
82 |
83 | # Show Initial Guess
84 | ax.scatter(initial_guess[0], initial_guess[1], c='black', s=20)
85 | plt.savefig(f"pp_{folder_path}/0.jpeg")
86 |
87 | cmaes = CMAES(objective=objective,
88 | initial_guess=initial_guess,
89 | workers_per_rank=10,
90 | sigma=sigma,
91 | seed=100,
92 | save=0,
93 | verbose=0,
94 | output_folder='DNE4py_result')
95 | optimizer = cmaes.optimizer
96 |
97 | # Loop:
98 | contour_x0, contour_x1 = np.mgrid[-1:1:.01, -1:1:.01]
99 | pos = np.dstack((contour_x0, contour_x1))
100 | for g in range(nb_generations - 1):
101 |
102 | # Image 1:
103 |
104 | # Plot function:
105 | fig, ax = start_image()
106 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
107 |
108 | # Plot current distribution (black):
109 | mu_x0, mu_x1 = optimizer.mean
110 | variance = optimizer.sigma
111 | rv = multivariate_normal([mu_x0, mu_x1], [[variance, 0], [0, variance]])
112 | ax.contour(contour_x0, contour_x1, rv.pdf(pos), colors='black', alpha=0.3)
113 | # plt.savefig(f"{folder_path}/pp/{g+1}_1.jpeg")
114 |
115 | # Plot current points (black):
116 | ax.scatter(genotypes[g][:, 0], genotypes[g][:, 1], s=10, c='black')
117 | plt.savefig(f"pp_{folder_path}/{g+1}_1.jpeg")
118 |
119 | # Update CMAES
120 | solutions = np.array(optimizer.ask())
121 | optimizer.tell(genotypes[g], costs[g])
122 |
123 | # Plot next distribution (red):
124 | mu_x0, mu_x1 = optimizer.mean
125 | variance = optimizer.sigma
126 | rv = multivariate_normal([mu_x0, mu_x1], [[variance, 0], [0, variance]])
127 | ax.contour(contour_x0, contour_x1, rv.pdf(pos), colors='red', alpha=0.3)
128 | plt.savefig(f"pp_{folder_path}/{g+1}_2.jpeg")
129 |
130 | # Plot current points (red):
131 | ax.scatter(genotypes[g + 1][:, 0], genotypes[g + 1][:, 1], s=10, c='red')
132 | plt.savefig(f"pp_{folder_path}/{g+1}_3.jpeg")
133 |
134 | def deepga_render(folder_path, nb_generations, objective, sigma, num_parents, num_elite):
135 |
136 | import numpy as np
137 | import matplotlib.pyplot as plt
138 |
139 | from DNE4py.optimizers.deepga.mutation import Member
140 |
141 | def start_image():
142 | fig, ax = plt.subplots()
143 | ax.set_xlim([-1, 1])
144 | ax.set_ylim([-1, 1])
145 | ax.set_xticks(np.arange(-1, 1, 0.2))
146 | ax.set_yticks(np.arange(-1, 1, 0.2))
147 | return fig, ax
148 |
149 | # Read Input:
150 | raw_folder_path = f"{folder_path}/raw_data"
151 | costs = load_mpidata(f"{raw_folder_path}", "costs", nb_generations)
152 | genotypes = load_mpidata(f"{raw_folder_path}", "genotypes", nb_generations)
153 | initial_guess = load_mpidata(f"{raw_folder_path}", "initial_guess", 1)[0]
154 | print(' costs toto ', folder_path)
155 | # Start figure:
156 | fig, ax = start_image()
157 |
158 | # Plot function:
159 | resolution = 100
160 | x1, x2 = np.meshgrid(np.linspace(-1, 1, resolution), np.linspace(-1, 1, resolution))
161 | X = np.array([x1.flatten(), x2.flatten()]).T
162 | Y = np.array([objective(x) for x in X]).reshape(resolution, resolution)
163 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
164 |
165 | # Show Initial Guess
166 | ax.scatter(initial_guess[0], initial_guess[1], c='black', s=20)
167 | plt.savefig(f"pp_{folder_path}/0.jpeg")
168 |
169 | # Loop:
170 | for g in range(nb_generations - 1):
171 |
172 | # Plot function:
173 | fig, ax = start_image()
174 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
175 |
176 | # Image 1:
177 | phenotypes = []
178 | for genotype in genotypes[g]:
179 | phenotype = Member(initial_guess, genotype, sigma).phenotype
180 | phenotypes.append(phenotype)
181 | phenotypes = np.array(phenotypes)
182 |
183 | ax.scatter(phenotypes[:, 0], phenotypes[:, 1], c='black', s=10)
184 | plt.savefig(f"pp_{folder_path}/{g+1}_1.jpeg")
185 |
186 | # Image 2:
187 | order = np.argsort(costs[g])
188 | rank = np.argsort(order)
189 | parents_mask = rank < num_parents
190 | elite_mask = rank < num_elite
191 |
192 | parents_phenotypes = phenotypes[parents_mask]
193 | elite_phenotypes = phenotypes[elite_mask]
194 | ax.scatter(parents_phenotypes[:, 0], parents_phenotypes[:, 1], c='green', s=10)
195 | ax.scatter(elite_phenotypes[:, 0], elite_phenotypes[:, 1], c='blue', s=10)
196 | plt.savefig(f"pp_{folder_path}/{g+1}_2.jpeg")
197 |
198 | # Image 3
199 | phenotypes = []
200 | for genotype in genotypes[g + 1]:
201 | phenotype = Member(initial_guess, genotype, sigma).phenotype
202 | phenotypes.append(phenotype)
203 | phenotypes = np.array(phenotypes)
204 | ax.scatter(phenotypes[:, 0], phenotypes[:, 1], c='red', s=10)
205 | plt.savefig(f"pp_{folder_path}/{g+1}_3.jpeg")
206 |
207 | def composite_deepga_render(folder_path, nb_generations, objective, sigma, num_parents, num_elite):
208 |
209 | import numpy as np
210 | import matplotlib.pyplot as plt
211 |
212 | from DNE4py.optimizers.deepga.mutation import Member
213 |
214 | def ranking(data):
215 | """
216 | @TODO : gèrer les ex-aequo
217 | """
218 | for inst in range(data.shape[1]):
219 | data[:,inst] = rankdata(data[:,inst])
220 | return np.mean(data, axis=1)
221 |
222 | def start_image():
223 | fig, ax = plt.subplots()
224 | ax.set_xlim([-1, 1])
225 | ax.set_ylim([-1, 1])
226 | ax.set_xticks(np.arange(-1, 1, 0.2))
227 | ax.set_yticks(np.arange(-1, 1, 0.2))
228 | return fig, ax
229 |
230 | # Read Input:
231 | raw_folder_path = f"{folder_path}/raw_data"
232 | costs = load_mpidata(f"{raw_folder_path}", "costs", nb_generations)
233 | genotypes = load_mpidata(f"{raw_folder_path}", "genotypes", nb_generations)
234 | initial_guess = load_mpidata(f"{raw_folder_path}", "initial_guess", 1)[0]
235 | # Start figure:
236 | fig, ax = start_image()
237 |
238 | # Plot function:
239 | resolution = 100
240 | x1, x2 = np.meshgrid(np.linspace(-1, 1, resolution), np.linspace(-1, 1, resolution))
241 | X = np.array([x1.flatten(), x2.flatten()]).T
242 | Y = np.array([objective(x) for x in X]).reshape(resolution, resolution)
243 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
244 |
245 | # Show Initial Guess
246 | ax.scatter(initial_guess[0], initial_guess[1], c='black', s=20)
247 | plt.savefig(f"pp_{folder_path}/0.jpeg")
248 |
249 | # Loop:
250 | for g in range(nb_generations - 1):
251 |
252 | # Plot function:
253 | fig, ax = start_image()
254 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
255 |
256 | # Image 1:
257 | phenotypes = []
258 | for genotype in genotypes[g]:
259 | phenotype = Member(initial_guess, genotype, sigma).phenotype
260 | phenotypes.append(phenotype)
261 |
262 | phenotypes = np.array(phenotypes)
263 | ax.scatter(phenotypes[:, 0], phenotypes[:, 1], c='black', s=10)
264 | plt.savefig(f"pp_{folder_path}/{g+1}_1.jpeg")
265 |
266 | # Image 2:
267 | average_ranking = ranking(costs[g])
268 | order = np.argsort(average_ranking)
269 | rank = np.argsort(order)
270 | parents_mask = rank < num_parents
271 | elite_mask = rank < num_elite
272 |
273 | parents_phenotypes = phenotypes[parents_mask]
274 | elite_phenotypes = phenotypes[elite_mask]
275 | ax.scatter(parents_phenotypes[:, 0], parents_phenotypes[:, 1], c='green', s=10)
276 | ax.scatter(elite_phenotypes[:, 0], elite_phenotypes[:, 1], c='blue', s=10)
277 | plt.savefig(f"pp_{folder_path}/{g+1}_2.jpeg")
278 |
279 | # Image 3
280 | phenotypes = []
281 | for genotype in genotypes[g + 1]:
282 | phenotype = Member(initial_guess, genotype, sigma).phenotype
283 | phenotypes.append(phenotype)
284 | phenotypes = np.array(phenotypes)
285 | ax.scatter(phenotypes[:, 0], phenotypes[:, 1], c='red', s=10)
286 | plt.savefig(f"pp_{folder_path}/{g+1}_3.jpeg")
287 |
288 | # Move somewhere else
289 | def rankdata(a):
290 | """Ranks the data in a, dealing with ties appropriately.
291 | Equal values are assigned a rank that is the average of the ranks that
292 | would have been otherwise assigned to all of the values within that set.
293 | Ranks begin at 1, not 0.
294 | Example:
295 | In [15]: stats.rankdata([0, 2, 2, 3])
296 | Out[15]: array([ 1. , 2.5, 2.5, 4. ])
297 | Parameters:
298 | - *a* : array
299 | This array is first flattened.
300 | Returns:
301 | An array of length equal to the size of a, containing rank scores.
302 | """
303 | a = np.ravel(a)
304 | n = len(a)
305 | svec, ivec = fastsort(a)
306 | sumranks = 0
307 | dupcount = 0
308 | newarray = np.zeros(n, float)
309 | for i in range(n):
310 | sumranks += i
311 | dupcount += 1
312 | if i == n - 1 or svec[i] != svec[i + 1]:
313 | averank = sumranks / float(dupcount) + 1
314 | for j in range(i - dupcount + 1, i + 1):
315 | newarray[ivec[j]] = averank
316 | sumranks = 0
317 | dupcount = 0
318 | return newarray
319 | # Dimensions
320 | # Calcul des ranks (non trivial dans le cas composite)
321 |
322 | def fastsort(a):
323 | it = np.argsort(a)
324 | as_ = a[it]
325 | return as_, it
--------------------------------------------------------------------------------
/DNE4py/version_v2/new_tutorials/tutorials2/run.py:
--------------------------------------------------------------------------------
1 | import sys
2 | import numpy as np
3 |
4 | from DNE4py.optimizers.deepga2 import TruncatedRealMutatorCompactGA, TruncatedRealMutatorCompositeGA
5 | from DNE4py.optimizers.cmaes import CMAES
6 | from DNE4py.optimizers.random import BatchRandomSearch
7 |
8 | # Compact example task
9 | def objective_function(x):
10 | result = np.sum(x**2)
11 | return result
12 |
13 | def get_deepga_TruncatedRealMutatorCompactGA():
14 |
15 | initial_guess = np.array([-0.3, 0.7])
16 | workers_per_rank = 10
17 | num_elite = 1
18 | num_parents = 3
19 | sigma = 0.05
20 | seed = 100
21 |
22 | optimizer = TruncatedRealMutatorCompactGA(objective_function,
23 | {'initial_guess': initial_guess,
24 | 'workers_per_rank': workers_per_rank,
25 | 'num_elite': num_elite,
26 | 'num_parents': num_parents,
27 | 'sigma': sigma,
28 | 'global_seed': seed,
29 | 'save': 1,
30 | 'verbose': 1,
31 | 'output_folder': 'results/DeepGA/TruncatedRealMutatorCompactGA'})
32 | return optimizer
33 |
34 | def get_cmaes_CMAES():
35 |
36 | initial_guess = np.array([-0.3, 0.7])
37 | workers_per_rank = 10
38 | sigma = 0.05
39 | seed = 100
40 |
41 | optimizer = CMAES(objective_function,
42 | {'initial_guess': initial_guess,
43 | 'workers_per_rank': workers_per_rank,
44 | 'sigma': sigma,
45 | 'global_seed': seed,
46 | 'save': 1,
47 | 'verbose': 1,
48 | 'output_folder': 'results/CMAES'})
49 | return optimizer
50 |
51 | def get_random_BatchRandomSearch():
52 |
53 | dim = 2
54 | workers_per_rank = 10
55 | bounds = np.array((-1, 1))
56 | global_seed = 100
57 |
58 | optimizer = BatchRandomSearch(objective_function,
59 | {'dim': dim,
60 | 'bounds': bounds,
61 | 'workers_per_rank': workers_per_rank,
62 | 'global_seed': global_seed,
63 | 'save': 1,
64 | 'verbose': 1,
65 | 'output_folder': 'results/BatchRandomSearch'})
66 | return optimizer
67 |
68 |
69 | # Composite example task
70 | def composite_objective_function(x, n_tasks=50):
71 | """
72 | The composite objective is composed of n_tasks functions of the form f_i:(x,y) -> (x^2 + y^2) + i
73 | for i in {-n_task/2, ..., n_task/2}.
74 | Thus mean(f_i) = (x^2 + y^2)
75 | """
76 | result = np.array([np.sum(x**2) + i for i in range(-(n_tasks//2), n_tasks//2)])
77 | return result
78 |
79 |
80 | def get_deepga_TruncatedRealMutatorCompositeGA():
81 |
82 | initial_guess = np.array([-0.3, 0.7])
83 | workers_per_rank = 10
84 | num_elite = 1
85 | num_parents = 3
86 | sigma = 0.05
87 | seed = 100
88 |
89 | optimizer = TruncatedRealMutatorCompositeGA(composite_objective_function,
90 | {'initial_guess': initial_guess,
91 | 'workers_per_rank': workers_per_rank,
92 | 'num_elite': num_elite,
93 | 'num_parents': num_parents,
94 | 'sigma': sigma,
95 | 'global_seed': seed,
96 | 'save': 1,
97 | 'verbose': 1,
98 | 'output_folder': 'results/DeepGA/TruncatedRealMutatorCompositeGA'})
99 | return optimizer
100 |
101 |
102 | if "__main__" == __name__:
103 | # Options:
104 | switcher = {
105 | 0: get_deepga_TruncatedRealMutatorCompactGA,
106 | 1: get_deepga_TruncatedRealMutatorCompositeGA,
107 | 2: get_cmaes_CMAES,
108 | 3: get_random_BatchRandomSearch,
109 | }
110 | optimizer = switcher.get(int(sys.argv[1]), lambda: "Invalid index for optimizer")()
111 | optimizer.run(20)
112 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/postprocessing/__init__.py:
--------------------------------------------------------------------------------
1 | from .utils import load_mpidata # get_best_phenotype, get_best_phenotype_generator, print_statistics, plot_cost_over_generation, save_meta_losses
2 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/postprocessing/test.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 |
3 |
4 | a = np.array([[12341], [5123, 123], [1234]], dtype=object)
5 | b = np.array([[5123], [1234], [1237, 4124, 1231]], dtype=object)
6 |
7 | print()
8 | print(a)
9 | print(b)
10 | print()
11 |
12 | with open('test.npy', 'ab') as f:
13 | np.save(f, a)
14 |
15 | with open('test.npy', 'rb') as f:
16 | for _ in range(1):
17 | print(np.load(f, allow_pickle=True))
18 | print()
19 |
20 | with open('test.npy', 'ab') as f:
21 | np.save(f, b)
22 |
23 | with open('test.npy', 'rb') as f:
24 | print('!!!')
25 | for _ in range(2):
26 | print(np.load(f, allow_pickle=True))
27 | print()
28 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/postprocessing/utils.py:
--------------------------------------------------------------------------------
1 | import os
2 | import glob
3 | import logging
4 | import numpy as np
5 | import json
6 |
7 | def load_mpidata(name, folder_path):
8 |
9 | # Internals:
10 | nb_files = len(glob.glob1(f'{folder_path}', f'{name}*'))
11 | with open(f'{folder_path}/info.json', 'rb') as f:
12 | info = json.load(f)
13 | nb_generations = info['nb_generations']
14 |
15 | full_data = [[]] * nb_generations
16 | for w in range(nb_files):
17 | with open(f'{folder_path}/{name}_w{w}.npy', 'rb') as f:
18 | for g in range(nb_generations):
19 | full_data[g] = np.load(f, allow_pickle=True).tolist()
20 |
21 | return np.array(full_data, object)
22 |
23 | # def get_best_x(folder_path):
24 |
25 | # from DNE4py.optimizers.deepga.mutation import Member
26 |
27 | # # Read Input:
28 | # costs = load_mpidata("costs", f"{folder_path}")
29 | # genotypes = load_mpidata("genotypes", f"{folder_path}")
30 | # initial_guess = load_mpidata("initial_guess", f"{folder_path}")[0]
31 |
32 | # print(initial_guess)
33 | # exit()
34 |
35 | # def get_best_phenotype(folder_path, nb_generations, sigma):
36 |
37 | # from DNE4py.optimizers.deepga.mutation import Member
38 |
39 | # # Read Input:
40 | # costs = load_mpidata(f"{folder_path}", "costs", nb_generations)
41 | # genotypes = load_mpidata(f"{folder_path}", "genotypes", nb_generations)
42 | # initial_guess = load_mpidata(f"{folder_path}", "initial_guess", 1)[0]
43 |
44 | # # Select Best Idxs:
45 | # best_idxs = np.unravel_index(costs.argmin(), costs.shape)
46 |
47 | # # Create member and get phonetype:
48 | # phenotype = Member(initial_guess, genotypes[best_idxs], sigma).phenotype
49 | # return phenotype
50 |
51 | # def get_best_phenotype_generator(folder_path, nb_generations, sigma):
52 |
53 | # from DNE4py.optimizers.deepga.mutation import Member
54 |
55 | # # Read Input:
56 | # costs = load_mpidata(f"{folder_path}", "costs", nb_generations)
57 | # genotypes = load_mpidata(f"{folder_path}", "genotypes", nb_generations)
58 | # initial_guess = load_mpidata(f"{folder_path}", "initial_guess", 1)[0]
59 |
60 | # # Select Best Idxs:
61 | # min_idxs = np.argmin(costs, axis=1)
62 | # for i in range(nb_generations):
63 | # genotype = genotypes[i, min_idxs[i]]
64 | # yield Member(initial_guess, genotype, sigma).phenotype
65 |
66 | # def print_statistics(folder_path, nb_generations):
67 |
68 | # costs = load_mpidata(f"{folder_path}", "costs", nb_generations)
69 |
70 | # final_best_cost = np.min(costs[-1])
71 | # best_cost = np.min(np.min(costs, axis=1))
72 |
73 | # print(f"Final Best cost: {final_best_cost}")
74 | # print(f"Best cost: {best_cost}")
75 |
76 | # def plot_cost_over_generation(folder_path, nb_generations, sigma=None, test_objective=None):
77 |
78 | # import matplotlib.pyplot as plt
79 |
80 | # # Load data:
81 | # costs = load_mpidata(f"{folder_path}", "costs", nb_generations)
82 |
83 | # # Calculate data:
84 | # means = np.mean(costs, axis=1)
85 | # stds = np.std(costs, axis=1)
86 | # mins = np.min(costs, axis=1)
87 | # maxes = np.max(costs, axis=1)
88 |
89 | # # Plot Population errorbar:
90 | # plt.errorbar(np.arange(nb_generations), means, stds, fmt='ok', lw=3)
91 | # plt.errorbar(np.arange(nb_generations), means, [means - mins, maxes - means],
92 | # fmt='.k', ecolor='gray', lw=1)
93 |
94 | # # Plot Best Individuals (blue line):
95 | # plt.plot(np.arange(nb_generations), mins)
96 |
97 | # # Plot Test Performance of Best Individuals (red line):
98 | # if test_objective is not None:
99 | # best_phenotypes = get_best_phenotype_generator(folder_path, nb_generations, sigma)
100 |
101 | # test_evaluations = []
102 | # for i, phenotype in enumerate(best_phenotypes):
103 | # if i % 4 == 0:
104 | # print(f'{i}/{nb_generations}\r', end='')
105 | # evaluation = test_objective(phenotype)
106 | # test_evaluations.append(evaluation)
107 | # test_evaluations = np.array(test_evaluations)
108 | # plt.plot(np.arange(0, nb_generations, 4), test_evaluations)
109 |
110 | # # Configuration:
111 | # plt.xlim(-1, nb_generations)
112 | # plt.xticks(np.arange(-1, nb_generations + 1, nb_generations // 10.0))
113 |
114 | # # Save
115 | # plt.savefig(f"{folder_path}/cost_over_generation.png")
116 |
117 |
118 | # def save_meta_losses(input_path, output_path, nb_generations, test_objective=None, sigma=None):
119 |
120 | # # Graph 1 (generation x (meta_train_loss, meta_test_loss))
121 |
122 | # # Meta-Train Loss:
123 | # # Load data:
124 | # costs = load_mpidata(f"{input_path}", "costs", nb_generations)
125 |
126 | # # Calculate data:
127 | # meta_train_loss_y = np.min(costs, axis=1)
128 |
129 | # # Meta-Test Loss:
130 | # if test_objective is not None:
131 | # best_phenotypes = get_best_phenotype_generator(input_path, nb_generations, sigma)
132 |
133 | # meta_test_loss_y = []
134 | # for i, phenotype in enumerate(best_phenotypes):
135 | # if i % 1 == 0:
136 | # print(f'{i}/{nb_generations}\r', end='')
137 | # evaluation = test_objective(phenotype)
138 | # meta_test_loss_y.append(evaluation)
139 | # meta_test_loss_y = np.array(meta_test_loss_y)
140 |
141 | # with open(f"{output_path}/meta_train_loss_y.npy", "wb") as f:
142 | # np.save(f, meta_train_loss_y)
143 |
144 | # with open(f"{output_path}/meta_test_loss_y.npy", "wb") as f:
145 | # np.save(f, meta_test_loss_y)
146 |
--------------------------------------------------------------------------------
/DNE4py/version_v2/sliceops.py:
--------------------------------------------------------------------------------
1 | import numpy as np
2 |
3 |
4 | def multi_slice_add(x1_inplace, x2, x1_slices=(), x2_slices=()):
5 | """
6 | Does an inplace addition on x1 given a list of slice objects
7 | If slices for both are given, it is assumed that they will be of
8 | the same size and each slice will have the same # of elements
9 | """
10 |
11 | if (len(x1_slices) != 0) and (len(x2_slices) == 0):
12 | for x1_slice in x1_slices:
13 | x1_inplace[x1_slice] += x2
14 |
15 | elif (len(x1_slices) != 0) and (len(x2_slices) != 0) \
16 | and (len(x2_slices) == len(x1_slices)):
17 |
18 | for i in range(len(x1_slices)):
19 | x1_inplace[x1_slices[i]] += x2[x2_slices[i]]
20 |
21 | elif (len(x1_slices) == 0) and (len(x2_slices) == 0):
22 | x1_inplace += x2
23 |
24 |
25 | def multi_slice_subtract(x1_inplace, x2, x1_slices=(), x2_slices=()):
26 | """
27 | Does an inplace addition on x1 given a list of slice objects
28 | If slices for both are given, it is assumed that they will be of
29 | the same size and each slice will have the same # of elements
30 | """
31 |
32 | if (len(x1_slices) != 0) and (len(x2_slices) == 0):
33 | for x1_slice in x1_slices:
34 | x1_inplace[x1_slice] -= x2
35 |
36 | elif (len(x1_slices) != 0) and (len(x2_slices) != 0) \
37 | and (len(x2_slices) == len(x1_slices)):
38 |
39 | for i in range(len(x1_slices)):
40 | x1_inplace[x1_slices[i]] -= x2[x2_slices[i]]
41 |
42 | elif (len(x1_slices) == 0) and (len(x2_slices) == 0):
43 | x1_inplace -= x2
44 |
45 |
46 | def multi_slice_multiply(x1_inplace, x2, x1_slices=(), x2_slices=()):
47 | """
48 | Does an inplace multiplication on x1 given a list of slice objects
49 | If slices for both are given, it is assumed that they will be of
50 | the same size and each slice will have the same # of elements
51 | """
52 |
53 | if (len(x1_slices) != 0) and (len(x2_slices) == 0):
54 | for x1_slice in x1_slices:
55 | x1_inplace[x1_slice] *= x2
56 |
57 | elif (len(x1_slices) != 0) and (len(x2_slices) != 0) \
58 | and (len(x2_slices) == len(x1_slices)):
59 |
60 | for i in range(len(x1_slices)):
61 | x1_inplace[x1_slices[i]] *= x2[x2_slices[i]]
62 |
63 | elif (len(x1_slices) == 0) and (len(x2_slices) == 0):
64 | x1_inplace *= x2
65 |
66 |
67 | def multi_slice_divide(x1_inplace, x2, x1_slices=(), x2_slices=()):
68 | """
69 | Does an inplace multiplication on x1 given a list of slice objects
70 | If slices for both are given, it is assumed that they will be of
71 | the same size and each slice will have the same # of elements
72 | """
73 |
74 | if (len(x1_slices) != 0) and (len(x2_slices) == 0):
75 | for x1_slice in x1_slices:
76 | x1_inplace[x1_slice] /= x2
77 |
78 | elif (len(x1_slices) != 0) and (len(x2_slices) != 0) \
79 | and (len(x2_slices) == len(x1_slices)):
80 |
81 | for i in range(len(x1_slices)):
82 | x1_inplace[x1_slices[i]] /= x2[x2_slices[i]]
83 |
84 | elif (len(x1_slices) == 0) and (len(x2_slices) == 0):
85 | x1_inplace /= x2
86 |
87 |
88 | def multi_slice_assign(x1_inplace, x2, x1_slices=(), x2_slices=()):
89 | """
90 | Does an inplace assignment on x1 given a list of slice objects
91 | If slices for both are given, it is assumed that they will be of
92 | the same size and each slice will have the same # of elements
93 | """
94 |
95 | if (len(x1_slices) != 0) and (len(x2_slices) == 0):
96 | for x1_slice in x1_slices:
97 | x1_inplace[x1_slice] = x2
98 |
99 | elif (len(x1_slices) != 0) and (len(x2_slices) != 0) \
100 | and (len(x2_slices) == len(x1_slices)):
101 |
102 | for i in range(len(x1_slices)):
103 | x1_inplace[x1_slices[i]] = x2[x2_slices[i]]
104 |
105 | elif (len(x1_slices) == 0) and (len(x2_slices) == 0):
106 | x1_inplace = x2
107 |
108 |
109 | def multi_slice_mod(x1_inplace, x2, x1_slices=(), x2_slices=()):
110 | """
111 | Does an inplace modulo on x1 given a list of slice objects
112 | If slices for both are given, it is assumed that they will be of
113 | the same size and each slice will have the same # of elements
114 | """
115 |
116 | if (len(x1_slices) != 0) and (len(x2_slices) == 0):
117 | for x1_slice in x1_slices:
118 | x1_inplace[x1_slice] %= x2
119 |
120 | elif (len(x1_slices) != 0) and (len(x2_slices) != 0) \
121 | and (len(x2_slices) == len(x1_slices)):
122 |
123 | for i in range(len(x1_slices)):
124 | x1_inplace[x1_slices[i]] %= x2[x2_slices[i]]
125 |
126 | elif (len(x1_slices) == 0) and (len(x2_slices) == 0):
127 | x1_inplace %= x2
128 |
129 |
130 | def multi_slice_fabs(x1_inplace, x1_slices=()):
131 | """
132 | Does an inplace fabs on x1 given a list of slice objects
133 | """
134 |
135 | if len(x1_slices) != 0:
136 | for x1_slice in x1_slices:
137 | np.fabs(x1_inplace[x1_slice], out=x1_inplace[x1_slice])
138 |
139 | else:
140 | np.fabs(x1_inplace, out=x1_inplace)
141 |
142 |
143 | def multi_slice_clip(x1_inplace, lower, upper, xslices=None,
144 | lslices=None, uslices=None):
145 | """
146 | Does an inplace clip on x1
147 | """
148 |
149 | if (lslices is None) and (uslices is None) and (xslices is None):
150 | np.clip(x1_inplace, lower, upper, out=x1_inplace)
151 |
152 | elif (lslices is None) or (uslices is None) and (xslices is not None):
153 | for xslice in xslices:
154 | np.clip(x1_inplace[xslice], lower, upper, out=x1_inplace[xslice])
155 |
156 | elif (lslices is not None) and (uslices is not None) \
157 | and (len(lslices) == len(uslices) and (xslices is not None)):
158 | for i in range(len(xslices)):
159 | np.clip(x1_inplace[xslices[i]], lower[lslices[i]], upper[uslices[i]],
160 | out=x1_inplace[xslices[i]])
161 |
162 | else:
163 | raise NotImplementedError("Invalid arguments in multi_slice_clip")
164 |
165 |
166 | def random_slices(rng, iterator_size, size_of_slice, max_step=1):
167 | """
168 | Returns a list of slice objects given the size of the iterator it
169 | will be used for and the number of elements desired for the slice
170 | This will return additional slice each time it wraps around the
171 | iterator
172 |
173 | iterator_size - the number of elements in the iterator
174 | size_of_slice - the number of elements the slices will cover
175 | max_step - the maximum number of steps a slice will take.
176 | This affects the number of slice objects created, as
177 | larger max_step will create more wraps around the iterator
178 | and so return more slice objects
179 |
180 | The number of elements is not guaranteed when slices overlap themselves
181 | """
182 |
183 | step_size = rng.randint(1, max_step + 1) # randint is exclusive
184 | start_step = rng.randint(0, iterator_size)
185 |
186 | return build_slices(start_step, iterator_size, size_of_slice, step_size)
187 |
188 |
189 | def build_slices(start_step, iterator_size, size_of_slice, step_size):
190 | """
191 | Given a starting index, the size of the total members of the window,
192 | a step size, and the size of the iterator the slice will act upon,
193 | this function returns a list of slice objects that will cover that full
194 | window. Upon reaching the endpoints of the iterator, it will wrap around.
195 | """
196 |
197 | if step_size >= iterator_size:
198 | raise NotImplementedError("Error: step size must be less than the " +
199 | "size of the iterator")
200 | end_step = start_step + step_size * size_of_slice
201 | slices = []
202 | slice_start = start_step
203 | for i in range(1 + (end_step - step_size) // iterator_size):
204 | remaining = end_step - i * iterator_size
205 | if remaining > iterator_size:
206 | remaining = iterator_size
207 |
208 | slice_end = (slice_start + 1) + ((remaining -
209 | (slice_start + 1)) // step_size) * step_size
210 | slices.append(np.s_[slice_start:slice_end:step_size])
211 | slice_start = (slice_end - 1 + step_size) % iterator_size
212 |
213 | return slices
214 |
215 |
216 | def match_slices(slice_list1, slice_list2):
217 | """
218 | Will attempt to create additional slices to match the # elements of
219 | each slice from list1 to the corresponding slice of list 2.
220 | Will fail if the total # elements is different for each list
221 | """
222 |
223 | slice_list1 = list(slice_list1)
224 | slice_list2 = list(slice_list2)
225 | if slice_size(slice_list1) == slice_size(slice_list2):
226 | slice_list1.reverse()
227 | slice_list2.reverse()
228 | new_list1_slices = []
229 | new_list2_slices = []
230 |
231 | while len(slice_list1) != 0 and len(slice_list2) != 0:
232 | slice_1 = slice_list1.pop()
233 | slice_2 = slice_list2.pop()
234 | size_1 = slice_size(slice_1)
235 | size_2 = slice_size(slice_2)
236 |
237 | if size_1 < size_2:
238 | new_slice_2, slice_2 = splice_slice(slice_2, size_1)
239 | slice_list2.append(slice_2)
240 | new_list2_slices.append(new_slice_2)
241 | new_list1_slices.append(slice_1)
242 |
243 | elif size_2 < size_1:
244 | new_slice_1, slice_1 = splice_slice(slice_1, size_2)
245 | slice_list1.append(slice_1)
246 | new_list1_slices.append(new_slice_1)
247 | new_list2_slices.append(slice_2)
248 |
249 | elif size_1 == size_2:
250 | new_list1_slices.append(slice_1)
251 | new_list2_slices.append(slice_2)
252 |
253 | else:
254 | raise AssertionError("Error: slices not compatible")
255 |
256 | return new_list1_slices, new_list2_slices
257 |
258 |
259 | def splice_slice(slice_obj, num_elements):
260 | """
261 | Returns two slices spliced from a single slice.
262 | The size of the first slice will be # elements
263 | The size of the second slice will be the remainder
264 | """
265 |
266 | splice_point = slice_obj.step * (num_elements - 1) + slice_obj.start + 1
267 | new_start = splice_point - 1 + slice_obj.step
268 | return np.s_[slice_obj.start: splice_point: slice_obj.step], \
269 | np.s_[new_start: slice_obj.stop: slice_obj.step]
270 |
271 |
272 | def slice_size(slice_objects):
273 | """
274 | Returns the total number of elements in the combined slices
275 | Also works if given a single slice
276 | """
277 |
278 | num_elements = 0
279 |
280 | try:
281 | for sl in slice_objects:
282 | num_elements += (sl.stop - (sl.start + 1)) // sl.step + 1
283 | except TypeError:
284 | num_elements += (slice_objects.stop - (slice_objects.start + 1)) \
285 | // slice_objects.step + 1
286 |
287 | return num_elements
288 |
--------------------------------------------------------------------------------
/LICENSE:
--------------------------------------------------------------------------------
1 | GNU GENERAL PUBLIC LICENSE
2 | Version 3, 29 June 2007
3 |
4 | Copyright (C) 2007 Free Software Foundation, Inc.
5 | Everyone is permitted to copy and distribute verbatim copies
6 | of this license document, but changing it is not allowed.
7 |
8 | Preamble
9 |
10 | The GNU General Public License is a free, copyleft license for
11 | software and other kinds of works.
12 |
13 | The licenses for most software and other practical works are designed
14 | to take away your freedom to share and change the works. By contrast,
15 | the GNU General Public License is intended to guarantee your freedom to
16 | share and change all versions of a program--to make sure it remains free
17 | software for all its users. We, the Free Software Foundation, use the
18 | GNU General Public License for most of our software; it applies also to
19 | any other work released this way by its authors. You can apply it to
20 | your programs, too.
21 |
22 | When we speak of free software, we are referring to freedom, not
23 | price. Our General Public Licenses are designed to make sure that you
24 | have the freedom to distribute copies of free software (and charge for
25 | them if you wish), that you receive source code or can get it if you
26 | want it, that you can change the software or use pieces of it in new
27 | free programs, and that you know you can do these things.
28 |
29 | To protect your rights, we need to prevent others from denying you
30 | these rights or asking you to surrender the rights. Therefore, you have
31 | certain responsibilities if you distribute copies of the software, or if
32 | you modify it: responsibilities to respect the freedom of others.
33 |
34 | For example, if you distribute copies of such a program, whether
35 | gratis or for a fee, you must pass on to the recipients the same
36 | freedoms that you received. You must make sure that they, too, receive
37 | or can get the source code. And you must show them these terms so they
38 | know their rights.
39 |
40 | Developers that use the GNU GPL protect your rights with two steps:
41 | (1) assert copyright on the software, and (2) offer you this License
42 | giving you legal permission to copy, distribute and/or modify it.
43 |
44 | For the developers' and authors' protection, the GPL clearly explains
45 | that there is no warranty for this free software. For both users' and
46 | authors' sake, the GPL requires that modified versions be marked as
47 | changed, so that their problems will not be attributed erroneously to
48 | authors of previous versions.
49 |
50 | Some devices are designed to deny users access to install or run
51 | modified versions of the software inside them, although the manufacturer
52 | can do so. This is fundamentally incompatible with the aim of
53 | protecting users' freedom to change the software. The systematic
54 | pattern of such abuse occurs in the area of products for individuals to
55 | use, which is precisely where it is most unacceptable. Therefore, we
56 | have designed this version of the GPL to prohibit the practice for those
57 | products. If such problems arise substantially in other domains, we
58 | stand ready to extend this provision to those domains in future versions
59 | of the GPL, as needed to protect the freedom of users.
60 |
61 | Finally, every program is threatened constantly by software patents.
62 | States should not allow patents to restrict development and use of
63 | software on general-purpose computers, but in those that do, we wish to
64 | avoid the special danger that patents applied to a free program could
65 | make it effectively proprietary. To prevent this, the GPL assures that
66 | patents cannot be used to render the program non-free.
67 |
68 | The precise terms and conditions for copying, distribution and
69 | modification follow.
70 |
71 | TERMS AND CONDITIONS
72 |
73 | 0. Definitions.
74 |
75 | "This License" refers to version 3 of the GNU General Public License.
76 |
77 | "Copyright" also means copyright-like laws that apply to other kinds of
78 | works, such as semiconductor masks.
79 |
80 | "The Program" refers to any copyrightable work licensed under this
81 | License. Each licensee is addressed as "you". "Licensees" and
82 | "recipients" may be individuals or organizations.
83 |
84 | To "modify" a work means to copy from or adapt all or part of the work
85 | in a fashion requiring copyright permission, other than the making of an
86 | exact copy. The resulting work is called a "modified version" of the
87 | earlier work or a work "based on" the earlier work.
88 |
89 | A "covered work" means either the unmodified Program or a work based
90 | on the Program.
91 |
92 | To "propagate" a work means to do anything with it that, without
93 | permission, would make you directly or secondarily liable for
94 | infringement under applicable copyright law, except executing it on a
95 | computer or modifying a private copy. Propagation includes copying,
96 | distribution (with or without modification), making available to the
97 | public, and in some countries other activities as well.
98 |
99 | To "convey" a work means any kind of propagation that enables other
100 | parties to make or receive copies. Mere interaction with a user through
101 | a computer network, with no transfer of a copy, is not conveying.
102 |
103 | An interactive user interface displays "Appropriate Legal Notices"
104 | to the extent that it includes a convenient and prominently visible
105 | feature that (1) displays an appropriate copyright notice, and (2)
106 | tells the user that there is no warranty for the work (except to the
107 | extent that warranties are provided), that licensees may convey the
108 | work under this License, and how to view a copy of this License. If
109 | the interface presents a list of user commands or options, such as a
110 | menu, a prominent item in the list meets this criterion.
111 |
112 | 1. Source Code.
113 |
114 | The "source code" for a work means the preferred form of the work
115 | for making modifications to it. "Object code" means any non-source
116 | form of a work.
117 |
118 | A "Standard Interface" means an interface that either is an official
119 | standard defined by a recognized standards body, or, in the case of
120 | interfaces specified for a particular programming language, one that
121 | is widely used among developers working in that language.
122 |
123 | The "System Libraries" of an executable work include anything, other
124 | than the work as a whole, that (a) is included in the normal form of
125 | packaging a Major Component, but which is not part of that Major
126 | Component, and (b) serves only to enable use of the work with that
127 | Major Component, or to implement a Standard Interface for which an
128 | implementation is available to the public in source code form. A
129 | "Major Component", in this context, means a major essential component
130 | (kernel, window system, and so on) of the specific operating system
131 | (if any) on which the executable work runs, or a compiler used to
132 | produce the work, or an object code interpreter used to run it.
133 |
134 | The "Corresponding Source" for a work in object code form means all
135 | the source code needed to generate, install, and (for an executable
136 | work) run the object code and to modify the work, including scripts to
137 | control those activities. However, it does not include the work's
138 | System Libraries, or general-purpose tools or generally available free
139 | programs which are used unmodified in performing those activities but
140 | which are not part of the work. For example, Corresponding Source
141 | includes interface definition files associated with source files for
142 | the work, and the source code for shared libraries and dynamically
143 | linked subprograms that the work is specifically designed to require,
144 | such as by intimate data communication or control flow between those
145 | subprograms and other parts of the work.
146 |
147 | The Corresponding Source need not include anything that users
148 | can regenerate automatically from other parts of the Corresponding
149 | Source.
150 |
151 | The Corresponding Source for a work in source code form is that
152 | same work.
153 |
154 | 2. Basic Permissions.
155 |
156 | All rights granted under this License are granted for the term of
157 | copyright on the Program, and are irrevocable provided the stated
158 | conditions are met. This License explicitly affirms your unlimited
159 | permission to run the unmodified Program. The output from running a
160 | covered work is covered by this License only if the output, given its
161 | content, constitutes a covered work. This License acknowledges your
162 | rights of fair use or other equivalent, as provided by copyright law.
163 |
164 | You may make, run and propagate covered works that you do not
165 | convey, without conditions so long as your license otherwise remains
166 | in force. You may convey covered works to others for the sole purpose
167 | of having them make modifications exclusively for you, or provide you
168 | with facilities for running those works, provided that you comply with
169 | the terms of this License in conveying all material for which you do
170 | not control copyright. Those thus making or running the covered works
171 | for you must do so exclusively on your behalf, under your direction
172 | and control, on terms that prohibit them from making any copies of
173 | your copyrighted material outside their relationship with you.
174 |
175 | Conveying under any other circumstances is permitted solely under
176 | the conditions stated below. Sublicensing is not allowed; section 10
177 | makes it unnecessary.
178 |
179 | 3. Protecting Users' Legal Rights From Anti-Circumvention Law.
180 |
181 | No covered work shall be deemed part of an effective technological
182 | measure under any applicable law fulfilling obligations under article
183 | 11 of the WIPO copyright treaty adopted on 20 December 1996, or
184 | similar laws prohibiting or restricting circumvention of such
185 | measures.
186 |
187 | When you convey a covered work, you waive any legal power to forbid
188 | circumvention of technological measures to the extent such circumvention
189 | is effected by exercising rights under this License with respect to
190 | the covered work, and you disclaim any intention to limit operation or
191 | modification of the work as a means of enforcing, against the work's
192 | users, your or third parties' legal rights to forbid circumvention of
193 | technological measures.
194 |
195 | 4. Conveying Verbatim Copies.
196 |
197 | You may convey verbatim copies of the Program's source code as you
198 | receive it, in any medium, provided that you conspicuously and
199 | appropriately publish on each copy an appropriate copyright notice;
200 | keep intact all notices stating that this License and any
201 | non-permissive terms added in accord with section 7 apply to the code;
202 | keep intact all notices of the absence of any warranty; and give all
203 | recipients a copy of this License along with the Program.
204 |
205 | You may charge any price or no price for each copy that you convey,
206 | and you may offer support or warranty protection for a fee.
207 |
208 | 5. Conveying Modified Source Versions.
209 |
210 | You may convey a work based on the Program, or the modifications to
211 | produce it from the Program, in the form of source code under the
212 | terms of section 4, provided that you also meet all of these conditions:
213 |
214 | a) The work must carry prominent notices stating that you modified
215 | it, and giving a relevant date.
216 |
217 | b) The work must carry prominent notices stating that it is
218 | released under this License and any conditions added under section
219 | 7. This requirement modifies the requirement in section 4 to
220 | "keep intact all notices".
221 |
222 | c) You must license the entire work, as a whole, under this
223 | License to anyone who comes into possession of a copy. This
224 | License will therefore apply, along with any applicable section 7
225 | additional terms, to the whole of the work, and all its parts,
226 | regardless of how they are packaged. This License gives no
227 | permission to license the work in any other way, but it does not
228 | invalidate such permission if you have separately received it.
229 |
230 | d) If the work has interactive user interfaces, each must display
231 | Appropriate Legal Notices; however, if the Program has interactive
232 | interfaces that do not display Appropriate Legal Notices, your
233 | work need not make them do so.
234 |
235 | A compilation of a covered work with other separate and independent
236 | works, which are not by their nature extensions of the covered work,
237 | and which are not combined with it such as to form a larger program,
238 | in or on a volume of a storage or distribution medium, is called an
239 | "aggregate" if the compilation and its resulting copyright are not
240 | used to limit the access or legal rights of the compilation's users
241 | beyond what the individual works permit. Inclusion of a covered work
242 | in an aggregate does not cause this License to apply to the other
243 | parts of the aggregate.
244 |
245 | 6. Conveying Non-Source Forms.
246 |
247 | You may convey a covered work in object code form under the terms
248 | of sections 4 and 5, provided that you also convey the
249 | machine-readable Corresponding Source under the terms of this License,
250 | in one of these ways:
251 |
252 | a) Convey the object code in, or embodied in, a physical product
253 | (including a physical distribution medium), accompanied by the
254 | Corresponding Source fixed on a durable physical medium
255 | customarily used for software interchange.
256 |
257 | b) Convey the object code in, or embodied in, a physical product
258 | (including a physical distribution medium), accompanied by a
259 | written offer, valid for at least three years and valid for as
260 | long as you offer spare parts or customer support for that product
261 | model, to give anyone who possesses the object code either (1) a
262 | copy of the Corresponding Source for all the software in the
263 | product that is covered by this License, on a durable physical
264 | medium customarily used for software interchange, for a price no
265 | more than your reasonable cost of physically performing this
266 | conveying of source, or (2) access to copy the
267 | Corresponding Source from a network server at no charge.
268 |
269 | c) Convey individual copies of the object code with a copy of the
270 | written offer to provide the Corresponding Source. This
271 | alternative is allowed only occasionally and noncommercially, and
272 | only if you received the object code with such an offer, in accord
273 | with subsection 6b.
274 |
275 | d) Convey the object code by offering access from a designated
276 | place (gratis or for a charge), and offer equivalent access to the
277 | Corresponding Source in the same way through the same place at no
278 | further charge. You need not require recipients to copy the
279 | Corresponding Source along with the object code. If the place to
280 | copy the object code is a network server, the Corresponding Source
281 | may be on a different server (operated by you or a third party)
282 | that supports equivalent copying facilities, provided you maintain
283 | clear directions next to the object code saying where to find the
284 | Corresponding Source. Regardless of what server hosts the
285 | Corresponding Source, you remain obligated to ensure that it is
286 | available for as long as needed to satisfy these requirements.
287 |
288 | e) Convey the object code using peer-to-peer transmission, provided
289 | you inform other peers where the object code and Corresponding
290 | Source of the work are being offered to the general public at no
291 | charge under subsection 6d.
292 |
293 | A separable portion of the object code, whose source code is excluded
294 | from the Corresponding Source as a System Library, need not be
295 | included in conveying the object code work.
296 |
297 | A "User Product" is either (1) a "consumer product", which means any
298 | tangible personal property which is normally used for personal, family,
299 | or household purposes, or (2) anything designed or sold for incorporation
300 | into a dwelling. In determining whether a product is a consumer product,
301 | doubtful cases shall be resolved in favor of coverage. For a particular
302 | product received by a particular user, "normally used" refers to a
303 | typical or common use of that class of product, regardless of the status
304 | of the particular user or of the way in which the particular user
305 | actually uses, or expects or is expected to use, the product. A product
306 | is a consumer product regardless of whether the product has substantial
307 | commercial, industrial or non-consumer uses, unless such uses represent
308 | the only significant mode of use of the product.
309 |
310 | "Installation Information" for a User Product means any methods,
311 | procedures, authorization keys, or other information required to install
312 | and execute modified versions of a covered work in that User Product from
313 | a modified version of its Corresponding Source. The information must
314 | suffice to ensure that the continued functioning of the modified object
315 | code is in no case prevented or interfered with solely because
316 | modification has been made.
317 |
318 | If you convey an object code work under this section in, or with, or
319 | specifically for use in, a User Product, and the conveying occurs as
320 | part of a transaction in which the right of possession and use of the
321 | User Product is transferred to the recipient in perpetuity or for a
322 | fixed term (regardless of how the transaction is characterized), the
323 | Corresponding Source conveyed under this section must be accompanied
324 | by the Installation Information. But this requirement does not apply
325 | if neither you nor any third party retains the ability to install
326 | modified object code on the User Product (for example, the work has
327 | been installed in ROM).
328 |
329 | The requirement to provide Installation Information does not include a
330 | requirement to continue to provide support service, warranty, or updates
331 | for a work that has been modified or installed by the recipient, or for
332 | the User Product in which it has been modified or installed. Access to a
333 | network may be denied when the modification itself materially and
334 | adversely affects the operation of the network or violates the rules and
335 | protocols for communication across the network.
336 |
337 | Corresponding Source conveyed, and Installation Information provided,
338 | in accord with this section must be in a format that is publicly
339 | documented (and with an implementation available to the public in
340 | source code form), and must require no special password or key for
341 | unpacking, reading or copying.
342 |
343 | 7. Additional Terms.
344 |
345 | "Additional permissions" are terms that supplement the terms of this
346 | License by making exceptions from one or more of its conditions.
347 | Additional permissions that are applicable to the entire Program shall
348 | be treated as though they were included in this License, to the extent
349 | that they are valid under applicable law. If additional permissions
350 | apply only to part of the Program, that part may be used separately
351 | under those permissions, but the entire Program remains governed by
352 | this License without regard to the additional permissions.
353 |
354 | When you convey a copy of a covered work, you may at your option
355 | remove any additional permissions from that copy, or from any part of
356 | it. (Additional permissions may be written to require their own
357 | removal in certain cases when you modify the work.) You may place
358 | additional permissions on material, added by you to a covered work,
359 | for which you have or can give appropriate copyright permission.
360 |
361 | Notwithstanding any other provision of this License, for material you
362 | add to a covered work, you may (if authorized by the copyright holders of
363 | that material) supplement the terms of this License with terms:
364 |
365 | a) Disclaiming warranty or limiting liability differently from the
366 | terms of sections 15 and 16 of this License; or
367 |
368 | b) Requiring preservation of specified reasonable legal notices or
369 | author attributions in that material or in the Appropriate Legal
370 | Notices displayed by works containing it; or
371 |
372 | c) Prohibiting misrepresentation of the origin of that material, or
373 | requiring that modified versions of such material be marked in
374 | reasonable ways as different from the original version; or
375 |
376 | d) Limiting the use for publicity purposes of names of licensors or
377 | authors of the material; or
378 |
379 | e) Declining to grant rights under trademark law for use of some
380 | trade names, trademarks, or service marks; or
381 |
382 | f) Requiring indemnification of licensors and authors of that
383 | material by anyone who conveys the material (or modified versions of
384 | it) with contractual assumptions of liability to the recipient, for
385 | any liability that these contractual assumptions directly impose on
386 | those licensors and authors.
387 |
388 | All other non-permissive additional terms are considered "further
389 | restrictions" within the meaning of section 10. If the Program as you
390 | received it, or any part of it, contains a notice stating that it is
391 | governed by this License along with a term that is a further
392 | restriction, you may remove that term. If a license document contains
393 | a further restriction but permits relicensing or conveying under this
394 | License, you may add to a covered work material governed by the terms
395 | of that license document, provided that the further restriction does
396 | not survive such relicensing or conveying.
397 |
398 | If you add terms to a covered work in accord with this section, you
399 | must place, in the relevant source files, a statement of the
400 | additional terms that apply to those files, or a notice indicating
401 | where to find the applicable terms.
402 |
403 | Additional terms, permissive or non-permissive, may be stated in the
404 | form of a separately written license, or stated as exceptions;
405 | the above requirements apply either way.
406 |
407 | 8. Termination.
408 |
409 | You may not propagate or modify a covered work except as expressly
410 | provided under this License. Any attempt otherwise to propagate or
411 | modify it is void, and will automatically terminate your rights under
412 | this License (including any patent licenses granted under the third
413 | paragraph of section 11).
414 |
415 | However, if you cease all violation of this License, then your
416 | license from a particular copyright holder is reinstated (a)
417 | provisionally, unless and until the copyright holder explicitly and
418 | finally terminates your license, and (b) permanently, if the copyright
419 | holder fails to notify you of the violation by some reasonable means
420 | prior to 60 days after the cessation.
421 |
422 | Moreover, your license from a particular copyright holder is
423 | reinstated permanently if the copyright holder notifies you of the
424 | violation by some reasonable means, this is the first time you have
425 | received notice of violation of this License (for any work) from that
426 | copyright holder, and you cure the violation prior to 30 days after
427 | your receipt of the notice.
428 |
429 | Termination of your rights under this section does not terminate the
430 | licenses of parties who have received copies or rights from you under
431 | this License. If your rights have been terminated and not permanently
432 | reinstated, you do not qualify to receive new licenses for the same
433 | material under section 10.
434 |
435 | 9. Acceptance Not Required for Having Copies.
436 |
437 | You are not required to accept this License in order to receive or
438 | run a copy of the Program. Ancillary propagation of a covered work
439 | occurring solely as a consequence of using peer-to-peer transmission
440 | to receive a copy likewise does not require acceptance. However,
441 | nothing other than this License grants you permission to propagate or
442 | modify any covered work. These actions infringe copyright if you do
443 | not accept this License. Therefore, by modifying or propagating a
444 | covered work, you indicate your acceptance of this License to do so.
445 |
446 | 10. Automatic Licensing of Downstream Recipients.
447 |
448 | Each time you convey a covered work, the recipient automatically
449 | receives a license from the original licensors, to run, modify and
450 | propagate that work, subject to this License. You are not responsible
451 | for enforcing compliance by third parties with this License.
452 |
453 | An "entity transaction" is a transaction transferring control of an
454 | organization, or substantially all assets of one, or subdividing an
455 | organization, or merging organizations. If propagation of a covered
456 | work results from an entity transaction, each party to that
457 | transaction who receives a copy of the work also receives whatever
458 | licenses to the work the party's predecessor in interest had or could
459 | give under the previous paragraph, plus a right to possession of the
460 | Corresponding Source of the work from the predecessor in interest, if
461 | the predecessor has it or can get it with reasonable efforts.
462 |
463 | You may not impose any further restrictions on the exercise of the
464 | rights granted or affirmed under this License. For example, you may
465 | not impose a license fee, royalty, or other charge for exercise of
466 | rights granted under this License, and you may not initiate litigation
467 | (including a cross-claim or counterclaim in a lawsuit) alleging that
468 | any patent claim is infringed by making, using, selling, offering for
469 | sale, or importing the Program or any portion of it.
470 |
471 | 11. Patents.
472 |
473 | A "contributor" is a copyright holder who authorizes use under this
474 | License of the Program or a work on which the Program is based. The
475 | work thus licensed is called the contributor's "contributor version".
476 |
477 | A contributor's "essential patent claims" are all patent claims
478 | owned or controlled by the contributor, whether already acquired or
479 | hereafter acquired, that would be infringed by some manner, permitted
480 | by this License, of making, using, or selling its contributor version,
481 | but do not include claims that would be infringed only as a
482 | consequence of further modification of the contributor version. For
483 | purposes of this definition, "control" includes the right to grant
484 | patent sublicenses in a manner consistent with the requirements of
485 | this License.
486 |
487 | Each contributor grants you a non-exclusive, worldwide, royalty-free
488 | patent license under the contributor's essential patent claims, to
489 | make, use, sell, offer for sale, import and otherwise run, modify and
490 | propagate the contents of its contributor version.
491 |
492 | In the following three paragraphs, a "patent license" is any express
493 | agreement or commitment, however denominated, not to enforce a patent
494 | (such as an express permission to practice a patent or covenant not to
495 | sue for patent infringement). To "grant" such a patent license to a
496 | party means to make such an agreement or commitment not to enforce a
497 | patent against the party.
498 |
499 | If you convey a covered work, knowingly relying on a patent license,
500 | and the Corresponding Source of the work is not available for anyone
501 | to copy, free of charge and under the terms of this License, through a
502 | publicly available network server or other readily accessible means,
503 | then you must either (1) cause the Corresponding Source to be so
504 | available, or (2) arrange to deprive yourself of the benefit of the
505 | patent license for this particular work, or (3) arrange, in a manner
506 | consistent with the requirements of this License, to extend the patent
507 | license to downstream recipients. "Knowingly relying" means you have
508 | actual knowledge that, but for the patent license, your conveying the
509 | covered work in a country, or your recipient's use of the covered work
510 | in a country, would infringe one or more identifiable patents in that
511 | country that you have reason to believe are valid.
512 |
513 | If, pursuant to or in connection with a single transaction or
514 | arrangement, you convey, or propagate by procuring conveyance of, a
515 | covered work, and grant a patent license to some of the parties
516 | receiving the covered work authorizing them to use, propagate, modify
517 | or convey a specific copy of the covered work, then the patent license
518 | you grant is automatically extended to all recipients of the covered
519 | work and works based on it.
520 |
521 | A patent license is "discriminatory" if it does not include within
522 | the scope of its coverage, prohibits the exercise of, or is
523 | conditioned on the non-exercise of one or more of the rights that are
524 | specifically granted under this License. You may not convey a covered
525 | work if you are a party to an arrangement with a third party that is
526 | in the business of distributing software, under which you make payment
527 | to the third party based on the extent of your activity of conveying
528 | the work, and under which the third party grants, to any of the
529 | parties who would receive the covered work from you, a discriminatory
530 | patent license (a) in connection with copies of the covered work
531 | conveyed by you (or copies made from those copies), or (b) primarily
532 | for and in connection with specific products or compilations that
533 | contain the covered work, unless you entered into that arrangement,
534 | or that patent license was granted, prior to 28 March 2007.
535 |
536 | Nothing in this License shall be construed as excluding or limiting
537 | any implied license or other defenses to infringement that may
538 | otherwise be available to you under applicable patent law.
539 |
540 | 12. No Surrender of Others' Freedom.
541 |
542 | If conditions are imposed on you (whether by court order, agreement or
543 | otherwise) that contradict the conditions of this License, they do not
544 | excuse you from the conditions of this License. If you cannot convey a
545 | covered work so as to satisfy simultaneously your obligations under this
546 | License and any other pertinent obligations, then as a consequence you may
547 | not convey it at all. For example, if you agree to terms that obligate you
548 | to collect a royalty for further conveying from those to whom you convey
549 | the Program, the only way you could satisfy both those terms and this
550 | License would be to refrain entirely from conveying the Program.
551 |
552 | 13. Use with the GNU Affero General Public License.
553 |
554 | Notwithstanding any other provision of this License, you have
555 | permission to link or combine any covered work with a work licensed
556 | under version 3 of the GNU Affero General Public License into a single
557 | combined work, and to convey the resulting work. The terms of this
558 | License will continue to apply to the part which is the covered work,
559 | but the special requirements of the GNU Affero General Public License,
560 | section 13, concerning interaction through a network will apply to the
561 | combination as such.
562 |
563 | 14. Revised Versions of this License.
564 |
565 | The Free Software Foundation may publish revised and/or new versions of
566 | the GNU General Public License from time to time. Such new versions will
567 | be similar in spirit to the present version, but may differ in detail to
568 | address new problems or concerns.
569 |
570 | Each version is given a distinguishing version number. If the
571 | Program specifies that a certain numbered version of the GNU General
572 | Public License "or any later version" applies to it, you have the
573 | option of following the terms and conditions either of that numbered
574 | version or of any later version published by the Free Software
575 | Foundation. If the Program does not specify a version number of the
576 | GNU General Public License, you may choose any version ever published
577 | by the Free Software Foundation.
578 |
579 | If the Program specifies that a proxy can decide which future
580 | versions of the GNU General Public License can be used, that proxy's
581 | public statement of acceptance of a version permanently authorizes you
582 | to choose that version for the Program.
583 |
584 | Later license versions may give you additional or different
585 | permissions. However, no additional obligations are imposed on any
586 | author or copyright holder as a result of your choosing to follow a
587 | later version.
588 |
589 | 15. Disclaimer of Warranty.
590 |
591 | THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY
592 | APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT
593 | HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY
594 | OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO,
595 | THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
596 | PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM
597 | IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF
598 | ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
599 |
600 | 16. Limitation of Liability.
601 |
602 | IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
603 | WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS
604 | THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY
605 | GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE
606 | USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF
607 | DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD
608 | PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS),
609 | EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF
610 | SUCH DAMAGES.
611 |
612 | 17. Interpretation of Sections 15 and 16.
613 |
614 | If the disclaimer of warranty and limitation of liability provided
615 | above cannot be given local legal effect according to their terms,
616 | reviewing courts shall apply local law that most closely approximates
617 | an absolute waiver of all civil liability in connection with the
618 | Program, unless a warranty or assumption of liability accompanies a
619 | copy of the Program in return for a fee.
620 |
621 | END OF TERMS AND CONDITIONS
622 |
623 | How to Apply These Terms to Your New Programs
624 |
625 | If you develop a new program, and you want it to be of the greatest
626 | possible use to the public, the best way to achieve this is to make it
627 | free software which everyone can redistribute and change under these terms.
628 |
629 | To do so, attach the following notices to the program. It is safest
630 | to attach them to the start of each source file to most effectively
631 | state the exclusion of warranty; and each file should have at least
632 | the "copyright" line and a pointer to where the full notice is found.
633 |
634 |
635 | Copyright (C)
636 |
637 | This program is free software: you can redistribute it and/or modify
638 | it under the terms of the GNU General Public License as published by
639 | the Free Software Foundation, either version 3 of the License, or
640 | (at your option) any later version.
641 |
642 | This program is distributed in the hope that it will be useful,
643 | but WITHOUT ANY WARRANTY; without even the implied warranty of
644 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
645 | GNU General Public License for more details.
646 |
647 | You should have received a copy of the GNU General Public License
648 | along with this program. If not, see .
649 |
650 | Also add information on how to contact you by electronic and paper mail.
651 |
652 | If the program does terminal interaction, make it output a short
653 | notice like this when it starts in an interactive mode:
654 |
655 | Copyright (C)
656 | This program comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
657 | This is free software, and you are welcome to redistribute it
658 | under certain conditions; type `show c' for details.
659 |
660 | The hypothetical commands `show w' and `show c' should show the appropriate
661 | parts of the General Public License. Of course, your program's commands
662 | might be different; for a GUI interface, you would use an "about box".
663 |
664 | You should also get your employer (if you work as a programmer) or school,
665 | if any, to sign a "copyright disclaimer" for the program, if necessary.
666 | For more information on this, and how to apply and follow the GNU GPL, see
667 | .
668 |
669 | The GNU General Public License does not permit incorporating your program
670 | into proprietary programs. If your program is a subroutine library, you
671 | may consider it more useful to permit linking proprietary applications with
672 | the library. If this is what you want to do, use the GNU Lesser General
673 | Public License instead of this License. But first, please read
674 | .
675 |
--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | # DNE4py: Deep-Neuroevolution with mpi4py
2 |
3 | Status: Maintenance (expect bug fixes and major updates)
4 |
5 | DNE4py is a python library that aims to run and visualize many different evolutionary algorithms with high performance using mpi4py. It allows easy evaluation of evolutionary algorithms in high dimension (e.g. neural networks for reinforcement learning)
6 |
7 | Implementation available:
8 |
9 | * [Genetic Algorithms Are a Competitive Alternative for Training Deep Neural Networks for Reinforcement Learning](https://arxiv.org/pdf/1712.06567.pdf)
10 |
11 | ## Installation
12 |
13 | ```console
14 | foo@bar:~$ git clone https://github.com/optimization-toolbox/DNE4py
15 | foo@bar:~$ cd deep-neuroevolution-mpi4py/
16 | foo@bar:~$ python3 -m pip install -e .
17 | ```
18 |
19 | ## Running
20 |
21 | Create main.py:
22 |
23 | ```python
24 | from DNE4py.optimizers.deepga import TruncatedRealMutatorGA
25 |
26 | def objective(x):
27 | result = np.sum(x**2)
28 | return result
29 |
30 | initial_guess = [1.0, 1.0]
31 |
32 | optimizer = TruncatedRealMutatorGA(objective=objective,
33 | initial_guess=initial_guess,
34 | workers_per_rank=10,
35 | num_elite=1,
36 | num_parents=3,
37 | sigma=0.01,
38 | seed=100,
39 | save=1,
40 | verbose=1,
41 | output_folder='results/experiment')
42 |
43 | optimizer(100)
44 | ```
45 |
46 | Execute main.py (relies on MPI):
47 |
48 | ```console
49 | foo@bar:~$ mpiexec -n 4 python3 main.py
50 | ```
51 |
52 | This will create a result folder based on output_folder
53 |
54 | ##### DeepGA
55 |
56 | 
57 |
58 |
59 |
60 | ##### CMA-ES
61 |
62 | 
63 |
64 | ##### RandomSearch
65 |
66 | 
67 |
68 | ## Post-processing
69 |
70 | You can import and generate some visualizations:
71 | ```python
72 | from DNE4py import load_mpidata, get_best_phenotype, get_best_phenotype_generator
73 | ```
74 |
75 |
76 |
77 |
--------------------------------------------------------------------------------
/setup.py:
--------------------------------------------------------------------------------
1 | import setuptools
2 |
3 | with open("README.md", "r") as fh:
4 | long_description = fh.read()
5 |
6 | setuptools.setup(
7 | name="DNE4py",
8 | version="0.2",
9 | author="Hugo Dovs",
10 | author_email="hugodovs@gmail.com",
11 | description="DNE4py: Deep Neuroevolution Algorithms using mpi4py",
12 | long_description=long_description,
13 | long_description_content_type="text/markdown",
14 | url="https://github.com/optimization-toolbox/DNE4py",
15 | packages=setuptools.find_packages(),
16 | classifiers=[
17 | "Programming Language :: Python :: 3",
18 | "License :: GNU General Public License v3.0",
19 | "Operating System :: OS Independent",
20 | ],
21 | python_requires='>=3.5',
22 | )
23 |
--------------------------------------------------------------------------------
/tutorials/1_optimizing_user_defined_function/TruncatedRealMutatorGA.yaml:
--------------------------------------------------------------------------------
1 | optimizer:
2 | id: TruncatedRealMutatorGA
3 | workers_per_rank: 64
4 | num_elite: 1
5 | num_parents: 1
6 | sigma_initial: 0.1
7 | sigma_min: 0.01
8 | sigma_decay: 0.95
9 | global_seed: 42
10 | output_folder: '../2_postprocessing_the_results/results/TruncatedRealMutatorGA'
11 | save_steps: 1
12 | verbose: 2
13 |
--------------------------------------------------------------------------------
/tutorials/1_optimizing_user_defined_function/run.py:
--------------------------------------------------------------------------------
1 | import yaml
2 | import numpy as np
3 | from DNE4py import load_optimizer
4 | np.random.seed(10)
5 |
6 |
7 | def objective_function(x):
8 | return np.sum(x**2)
9 |
10 |
11 | if "__main__" == __name__:
12 |
13 | # Read config:
14 | with open(f'TruncatedRealMutatorGA.yaml') as f:
15 | config = yaml.load(f, Loader=yaml.FullLoader)
16 | optimizer_config = config.get('optimizer')
17 |
18 | # Declare Optimizer:
19 | optimizer_config['initial_guess'] = np.random.random(7000)
20 | optimizer = load_optimizer(optimizer_config)
21 |
22 | # Run Optimizer:
23 | optimizer.run(objective_function, 2)
24 |
--------------------------------------------------------------------------------
/tutorials/2_postprocessing_the_results/gif/cmaes.gif:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/optimization-toolbox/DNE4py/4639ff4f869f07d6f5858edf12bbc96e2c2c1824/tutorials/2_postprocessing_the_results/gif/cmaes.gif
--------------------------------------------------------------------------------
/tutorials/2_postprocessing_the_results/gif/deepga_truncatedrealmutatorga.gif:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/optimization-toolbox/DNE4py/4639ff4f869f07d6f5858edf12bbc96e2c2c1824/tutorials/2_postprocessing_the_results/gif/deepga_truncatedrealmutatorga.gif
--------------------------------------------------------------------------------
/tutorials/2_postprocessing_the_results/gif/randomsearch.gif:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/optimization-toolbox/DNE4py/4639ff4f869f07d6f5858edf12bbc96e2c2c1824/tutorials/2_postprocessing_the_results/gif/randomsearch.gif
--------------------------------------------------------------------------------
/tutorials/2_postprocessing_the_results/pp_run.py:
--------------------------------------------------------------------------------
1 | import sys
2 | import numpy as np
3 |
4 | from DNE4py import load_mpidata, get_best_phenotype, get_best_phenotype_generator
5 | import matplotlib.pyplot as plt
6 |
7 |
8 | def objective_function(x):
9 | return np.sum(x**2)
10 |
11 |
12 | if "__main__" == __name__:
13 |
14 | costs = load_mpidata('costs', 'results/TruncatedRealMutatorGA/')
15 | genotypes = load_mpidata('genotypes', 'results/TruncatedRealMutatorGA/')
16 | initial_guess = load_mpidata('initial_guess', 'results/TruncatedRealMutatorGA/')
17 |
18 | # for i, c in enumerate(costs):
19 | # print(i)
20 | # print(f'{np.min(c)}')
21 | # print(f'{c}')
22 | # print()
23 |
24 | print(f'costs shape: {costs.shape}')
25 | print(f'genotypes shape: {genotypes.shape}')
26 | print(f'initial_guess shape: {initial_guess.shape}')
27 |
28 | print(genotypes[0])
29 | print(genotypes[-1])
30 | #print(initial_guess)
31 | #print(costs[-1][0])
32 |
33 | x = get_best_phenotype('results/TruncatedRealMutatorGA/')
34 |
35 | #exit()
36 | #print(x)
37 | #print(objective_function(x))
38 | #exit()
39 |
40 | generator = get_best_phenotype_generator('results/TruncatedRealMutatorGA/')
41 |
42 | y = np.min(costs, axis=1)
43 | plt.plot(y)
44 | plt.show()
45 | # for i in generator:
46 | # print(i)
47 |
--------------------------------------------------------------------------------
/tutorials/2_postprocessing_the_results/render.py:
--------------------------------------------------------------------------------
1 | # After running these methods, you can generate a compressed .gif in the following command:
2 | # $: convert -layers OptimizeTransparency -delay 20 -loop 0 `ls -v` myimage.gif
3 | # To Optimizer: (Use Optimize Transparency 10%) https://ezgif.com/optimize/
4 |
5 |
6 | import pickle
7 |
8 | from scipy.stats import multivariate_normal
9 |
10 | from DNE4py.postprocessing.utils import load_mpidata
11 | from DNE4py.optimizers.cmaes import CMAES
12 |
13 |
14 | def randomsearch_render(folder_path, nb_generations, objective):
15 |
16 | import numpy as np
17 | import matplotlib.pyplot as plt
18 |
19 | def start_image():
20 | fig, ax = plt.subplots()
21 | ax.set_xlim([-1, 1])
22 | ax.set_ylim([-1, 1])
23 | ax.set_xticks(np.arange(-1, 1, 0.2))
24 | ax.set_yticks(np.arange(-1, 1, 0.2))
25 | return fig, ax
26 |
27 | # Read Input:
28 | costs = load_mpidata(f"{folder_path}", "costs", nb_generations)
29 | genotypes = load_mpidata(f"{folder_path}", "genotypes", nb_generations)
30 | initial_guess = load_mpidata(f"{folder_path}", "initial_guess", 1)[0]
31 |
32 | # Start figure:
33 | fig, ax = start_image()
34 |
35 | # Plot function:
36 | resolution = 100
37 | x1, x2 = np.meshgrid(np.linspace(-1, 1, resolution), np.linspace(-1, 1, resolution))
38 | X = np.array([x1.flatten(), x2.flatten()]).T
39 | Y = np.array([objective(x) for x in X]).reshape(resolution, resolution)
40 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
41 |
42 | for g in range(nb_generations - 1):
43 |
44 | # Image 1:
45 |
46 | # Plot function:
47 | fig, ax = start_image()
48 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
49 |
50 | # Plot points:
51 | ax.scatter(genotypes[g][:, 0], genotypes[g][:, 1], s=10, c='black')
52 | plt.savefig(f"pp_{folder_path}/{g+1}_1.jpeg")
53 |
54 | def cmaes_render(folder_path, nb_generations, objective, sigma):
55 |
56 | import numpy as np
57 | import matplotlib.pyplot as plt
58 |
59 | def start_image():
60 | fig, ax = plt.subplots()
61 | ax.set_xlim([-1, 1])
62 | ax.set_ylim([-1, 1])
63 | ax.set_xticks(np.arange(-1, 1, 0.2))
64 | ax.set_yticks(np.arange(-1, 1, 0.2))
65 | return fig, ax
66 |
67 | # Read Input:
68 | costs = load_mpidata(f"{folder_path}", "costs", nb_generations)
69 | genotypes = load_mpidata(f"{folder_path}", "genotypes", nb_generations)
70 | initial_guess = load_mpidata(f"{folder_path}", "initial_guess", 1)[0]
71 |
72 | # Start figure:
73 | fig, ax = start_image()
74 |
75 | # Plot function:
76 | resolution = 100
77 | x1, x2 = np.meshgrid(np.linspace(-1, 1, resolution), np.linspace(-1, 1, resolution))
78 | X = np.array([x1.flatten(), x2.flatten()]).T
79 | Y = np.array([objective(x) for x in X]).reshape(resolution, resolution)
80 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
81 |
82 | # Show Initial Guess
83 | ax.scatter(initial_guess[0], initial_guess[1], c='black', s=20)
84 | plt.savefig(f"pp_{folder_path}/0.jpeg")
85 |
86 | cmaes = CMAES(objective=objective,
87 | initial_guess=initial_guess,
88 | workers_per_rank=10,
89 | sigma=sigma,
90 | seed=100,
91 | save=0,
92 | verbose=0,
93 | output_folder='DNE4py_result')
94 | optimizer = cmaes.optimizer
95 |
96 | # Loop:
97 | contour_x0, contour_x1 = np.mgrid[-1:1:.01, -1:1:.01]
98 | pos = np.dstack((contour_x0, contour_x1))
99 | for g in range(nb_generations - 1):
100 |
101 | # Image 1:
102 |
103 | # Plot function:
104 | fig, ax = start_image()
105 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
106 |
107 | # Plot current distribution (black):
108 | mu_x0, mu_x1 = optimizer.mean
109 | variance = optimizer.sigma
110 | rv = multivariate_normal([mu_x0, mu_x1], [[variance, 0], [0, variance]])
111 | ax.contour(contour_x0, contour_x1, rv.pdf(pos), colors='black', alpha=0.3)
112 | # plt.savefig(f"{folder_path}/pp/{g+1}_1.jpeg")
113 |
114 | # Plot current points (black):
115 | ax.scatter(genotypes[g][:, 0], genotypes[g][:, 1], s=10, c='black')
116 | plt.savefig(f"pp_{folder_path}/{g+1}_1.jpeg")
117 |
118 | # Update CMAES
119 | solutions = np.array(optimizer.ask())
120 | optimizer.tell(genotypes[g], costs[g])
121 |
122 | # Plot next distribution (red):
123 | mu_x0, mu_x1 = optimizer.mean
124 | variance = optimizer.sigma
125 | rv = multivariate_normal([mu_x0, mu_x1], [[variance, 0], [0, variance]])
126 | ax.contour(contour_x0, contour_x1, rv.pdf(pos), colors='red', alpha=0.3)
127 | plt.savefig(f"pp_{folder_path}/{g+1}_2.jpeg")
128 |
129 | # Plot current points (red):
130 | ax.scatter(genotypes[g + 1][:, 0], genotypes[g + 1][:, 1], s=10, c='red')
131 | plt.savefig(f"pp_{folder_path}/{g+1}_3.jpeg")
132 |
133 | def deepga_render(folder_path, nb_generations, objective, sigma, num_parents, num_elite):
134 |
135 | import numpy as np
136 | import matplotlib.pyplot as plt
137 |
138 | from DNE4py.optimizers.deepga.mutation import Member
139 |
140 | def start_image():
141 | fig, ax = plt.subplots()
142 | ax.set_xlim([-1, 1])
143 | ax.set_ylim([-1, 1])
144 | ax.set_xticks(np.arange(-1, 1, 0.2))
145 | ax.set_yticks(np.arange(-1, 1, 0.2))
146 | return fig, ax
147 |
148 | # Read Input:
149 | costs = load_mpidata(f"{folder_path}", "costs", nb_generations)
150 | genotypes = load_mpidata(f"{folder_path}", "genotypes", nb_generations)
151 | initial_guess = load_mpidata(f"{folder_path}", "initial_guess", 1)[0]
152 |
153 | # Start figure:
154 | fig, ax = start_image()
155 |
156 | # Plot function:
157 | resolution = 100
158 | x1, x2 = np.meshgrid(np.linspace(-1, 1, resolution), np.linspace(-1, 1, resolution))
159 | X = np.array([x1.flatten(), x2.flatten()]).T
160 | Y = np.array([objective(x) for x in X]).reshape(resolution, resolution)
161 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
162 |
163 | # Show Initial Guess
164 | ax.scatter(initial_guess[0], initial_guess[1], c='black', s=20)
165 | plt.savefig(f"pp_{folder_path}/0.jpeg")
166 |
167 | # Loop:
168 | for g in range(nb_generations - 1):
169 |
170 | # Plot function:
171 | fig, ax = start_image()
172 | ax.pcolormesh(x1, x2, Y, cmap=plt.cm.coolwarm)
173 |
174 | # Image 1:
175 | phenotypes = []
176 | for genotype in genotypes[g]:
177 | phenotype = Member(initial_guess, genotype, sigma).phenotype
178 | phenotypes.append(phenotype)
179 | phenotypes = np.array(phenotypes)
180 |
181 | ax.scatter(phenotypes[:, 0], phenotypes[:, 1], c='black', s=10)
182 | plt.savefig(f"pp_{folder_path}/{g+1}_1.jpeg")
183 |
184 | # Image 2:
185 | order = np.argsort(costs[g])
186 | rank = np.argsort(order)
187 | parents_mask = rank < num_parents
188 | elite_mask = rank < num_elite
189 |
190 | parents_phenotypes = phenotypes[parents_mask]
191 | elite_phenotypes = phenotypes[elite_mask]
192 | ax.scatter(parents_phenotypes[:, 0], parents_phenotypes[:, 1], c='green', s=10)
193 | ax.scatter(elite_phenotypes[:, 0], elite_phenotypes[:, 1], c='blue', s=10)
194 | plt.savefig(f"pp_{folder_path}/{g+1}_2.jpeg")
195 |
196 | # Image 3
197 | phenotypes = []
198 | for genotype in genotypes[g + 1]:
199 | phenotype = Member(initial_guess, genotype, sigma).phenotype
200 | phenotypes.append(phenotype)
201 | phenotypes = np.array(phenotypes)
202 | ax.scatter(phenotypes[:, 0], phenotypes[:, 1], c='red', s=10)
203 | plt.savefig(f"pp_{folder_path}/{g+1}_3.jpeg")
204 |
--------------------------------------------------------------------------------
/tutorials/README.md:
--------------------------------------------------------------------------------
1 | ## `run.py`
2 | * It changes results folder
3 | ```console
4 | foo@bar:~$ mpiexec -n 2 python3 run.py 3
5 | ```
6 |
7 | ## `pp_run.py`
8 | * It changes pp_results folder
9 | * To Optimize: (Use Optimize Transparency 10%) https://ezgif.com/optimize/
10 |
11 | ```console
12 | foo@bar:~$ python3 pp_run.py 3
13 | foo@bar:~$ cd pp_results/RandomSearch/
14 | foo@bar:~$ convert -layers OptimizeTransparency -delay 20 -loop 0 `ls -v` randomsearch.gif
15 | ```
16 |
17 | ## `render.py`
18 | * It defines the behaviour to render the optimization procedure
19 |
20 |
--------------------------------------------------------------------------------