| Name | decc JSON |
| Version |
0.2.2
JSON |
| download |
| home_page | |
| Summary | Cooperative Co-evolutionary Differential Evolution algorithms in Python. |
| upload_time | 2023-09-17 23:31:01 |
| maintainer | |
| docs_url | None |
| author | |
| requires_python | >=3.8 |
| license | |
| keywords |
decc
|
| VCS |
 |
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
# decc: Cooperative Co-evolutionary Differential Evolution
Cooperative Co-evolutionary Differential Evolution algorithms in Python with NumPy.
This repository provides simple implementations for some CC-based DE algorithms. All the implementations are from scratch using NumPy, they might slightly differ from the original papers.
# Quick Start
The `decc` library works with two entities:
- `Problem` represents a minimization problem (objective function, dimensionality and bounds).
- The objective function must receives a matrix as input (each row represents a solution) and produce a matrix as output (each row is the objective value for the respective solution).
- `Optimizer` represents a DE algorithm for solving a minimization problem.
- Maximization of a real function `g` is equivalent to minimize `-g`.
In order to use an optimizer, you must first define a problem. Currently, there are two optimizer available: (i) [**DECC**](src/decc/optimizers/decc.py) (also known as CCDE), with the variants **DECC-O** and **DECC-H**; and (ii) [**DECC-G**](src/decc/optimizers/decc_g.py). A basic example is found below:
```python
import numpy as np
from decc import Problem
from decc.optimizers.decc import DECCOptimizer
from decc.optimizers.decc_g import DECCGOptimizer
def objective(x: np.ndarray) -> np.ndarray:
# The Sphere function is a common
# benchmark for optimizers
return np.sum(x ** 2, axis=-1)
# First, we must define the problem
problem = Problem(objective,
dims=100,
lower_bound=-100.0,
upper_bound=100.0)
# Then, we can instantiate the optimizers
F = 0.5
CR = 0.8
seed = 42
max_fn = int(1e5)
pop_size = 50
decc_h = DECCOptimizer(problem,
seed,
pop_size,
grouping='halve',
F=F,
CR=CR,
max_fn=max_fn)
decc_o = DECCOptimizer(problem,
seed,
pop_size,
grouping='dims',
F=F,
CR=CR,
max_fn=max_fn)
decc_g = DECCGOptimizer(problem,
seed,
pop_size,
n_subproblems=max(1, problem.dims // 4),
sansde_evaluations=max_fn // 3,
de_evaluations=max_fn // 5,
F=F,
CR=CR,
max_fn=max_fn)
# Lastly, we can optimize the objective and
# retrieve the results.
for optimizer in [decc_o, decc_h, decc_g]:
result = optimizer.optimize()
print(f'{optimizer}: {result["best_fitness"]}')
# DECC-O: 1.638248431845568e-05
# DECC-H: 0.0006988301174715161
# DECC-G: 1.5340782547163752e-10
```
# Roadmap
- Add support for `DECC-DG`;
- Add support for `DECC-gDG`;
- Improve documentation;
- Add unit tests;
- Add validation benchmark functions from the original papers;
Raw data
{
"_id": null,
"home_page": "",
"name": "decc",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": "",
"keywords": "decc",
"author": "",
"author_email": "",
"download_url": "https://files.pythonhosted.org/packages/ef/1a/145f9b0bc3b159ba41e3256e13e0d0ea44e5778b129396f6b9b5383efd61/decc-0.2.2.tar.gz",
"platform": null,
"description": "# decc: Cooperative Co-evolutionary Differential Evolution\n\nCooperative Co-evolutionary Differential Evolution algorithms in Python with NumPy. \n\nThis repository provides simple implementations for some CC-based DE algorithms. All the implementations are from scratch using NumPy, they might slightly differ from the original papers.\n\n# Quick Start\n\nThe `decc` library works with two entities:\n\n- `Problem` represents a minimization problem (objective function, dimensionality and bounds).\n - The objective function must receives a matrix as input (each row represents a solution) and produce a matrix as output (each row is the objective value for the respective solution).\n- `Optimizer` represents a DE algorithm for solving a minimization problem.\n - Maximization of a real function `g` is equivalent to minimize `-g`.\n\nIn order to use an optimizer, you must first define a problem. Currently, there are two optimizer available: (i) [**DECC**](src/decc/optimizers/decc.py) (also known as CCDE), with the variants **DECC-O** and **DECC-H**; and (ii) [**DECC-G**](src/decc/optimizers/decc_g.py). A basic example is found below:\n\n```python\nimport numpy as np\n\nfrom decc import Problem\nfrom decc.optimizers.decc import DECCOptimizer\nfrom decc.optimizers.decc_g import DECCGOptimizer\n\ndef objective(x: np.ndarray) -> np.ndarray:\n # The Sphere function is a common\n # benchmark for optimizers\n return np.sum(x ** 2, axis=-1)\n\n\n# First, we must define the problem\nproblem = Problem(objective,\n dims=100,\n lower_bound=-100.0,\n upper_bound=100.0)\n\n# Then, we can instantiate the optimizers\nF = 0.5\nCR = 0.8\nseed = 42\nmax_fn = int(1e5)\npop_size = 50\ndecc_h = DECCOptimizer(problem,\n seed,\n pop_size,\n grouping='halve',\n F=F,\n CR=CR,\n max_fn=max_fn)\ndecc_o = DECCOptimizer(problem,\n seed,\n pop_size,\n grouping='dims',\n F=F,\n CR=CR,\n max_fn=max_fn)\ndecc_g = DECCGOptimizer(problem,\n seed,\n pop_size,\n n_subproblems=max(1, problem.dims // 4),\n sansde_evaluations=max_fn // 3,\n de_evaluations=max_fn // 5,\n F=F,\n CR=CR,\n max_fn=max_fn)\n\n# Lastly, we can optimize the objective and\n# retrieve the results.\nfor optimizer in [decc_o, decc_h, decc_g]:\n result = optimizer.optimize()\n print(f'{optimizer}: {result[\"best_fitness\"]}')\n\n# DECC-O: 1.638248431845568e-05\n# DECC-H: 0.0006988301174715161\n# DECC-G: 1.5340782547163752e-10\n```\n\n# Roadmap\n\n- Add support for `DECC-DG`;\n- Add support for `DECC-gDG`;\n- Improve documentation;\n- Add unit tests;\n- Add validation benchmark functions from the original papers;\n",
"bugtrack_url": null,
"license": "",
"summary": "Cooperative Co-evolutionary Differential Evolution algorithms in Python.",
"version": "0.2.2",
"project_urls": {
"Github": "https://github.com/moesio-f/decc"
},
"split_keywords": [
"decc"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "875cd642d4e9fe1f5720be4183f2410fa90e05559861fbfc0c5114b6a42bbf45",
"md5": "968d96b4ac15f037029204308d85e1d6",
"sha256": "72440d463195db81ec1ae28280684663df88ef55cf245d46f025164dbe8806e8"
},
"downloads": -1,
"filename": "decc-0.2.2-py3-none-any.whl",
"has_sig": false,
"md5_digest": "968d96b4ac15f037029204308d85e1d6",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 19308,
"upload_time": "2023-09-17T23:30:59",
"upload_time_iso_8601": "2023-09-17T23:30:59.248091Z",
"url": "https://files.pythonhosted.org/packages/87/5c/d642d4e9fe1f5720be4183f2410fa90e05559861fbfc0c5114b6a42bbf45/decc-0.2.2-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "ef1a145f9b0bc3b159ba41e3256e13e0d0ea44e5778b129396f6b9b5383efd61",
"md5": "6f373b397df353e0dee33870a00351a4",
"sha256": "71907be25c7f213d831ad4cb7d08d50ff57e273d636b9d2b2656b1f9eec1679e"
},
"downloads": -1,
"filename": "decc-0.2.2.tar.gz",
"has_sig": false,
"md5_digest": "6f373b397df353e0dee33870a00351a4",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 14243,
"upload_time": "2023-09-17T23:31:01",
"upload_time_iso_8601": "2023-09-17T23:31:01.145997Z",
"url": "https://files.pythonhosted.org/packages/ef/1a/145f9b0bc3b159ba41e3256e13e0d0ea44e5778b129396f6b9b5383efd61/decc-0.2.2.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-09-17 23:31:01",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "moesio-f",
"github_project": "decc",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "decc"
}
