## Orloge
[](https://travis-ci.org/pchtsp/orloge)
## What and why
The idea of this project is to permit a fast and easy parsing of the log files from different solvers, specifically 'operations research' (OR) logs.
There exist bigger, and more robust libraries. In particular, [IPET](https://github.com/GregorCH/ipet/). The trouble I had was that it deals with too many benchmarking and GUI things and I wanted something simple I could modify and build on top.
In any case, a lot of the ideas and parsing strings were obtained or adapted from IPET, to whom I am graceful.
The supported solvers for the time being are: GUROBI, CPLEX and CBC. Specially the two first ones.
## How
The basic idea is just to provide a unique interface function like the following:
import orloge as ol
ol.get_info_solver(path_to_solver_log, solver_name)
This returns a python dictionary with a lot of information from the log (see *Examples* below).
## Installation
pip install orloge
or, for the development version:
pip install https://github.com/pchtsp/orloge/archive/master.zip
## Testing
Run the command
python -m unittest test
if the output says OK, all tests were passed.
## Reference
### Main parameters
The most common parameters to extract are: `best_bound`, `best_solution` and `time`. These three parameters are obtained at the end of the solving process and summarize the best relaxed objective value obtained, the best integer objective value obtained and the time it took to do the solving.
### Cuts
The cuts information can be accessed by the `cuts_info` key. It offers the best known bound after the cut phase has ended, the best solution (if any) after the cuts and the number of cuts made of each type.
### Matrix
There are two matrices that are provided. The `matrix` key returns the number of variables, constraints and non-zero values before the pre-processing of the solver. The `matrix_post` key returns these same values after the pre-processing has been done.
### Progress
The `progress` key returns a raw pandas Dataframe with the all the progress information the solver gives. Including the times, the gap, the best bound, the best solution, the iterations, nodes, among other. This table can vary in number of columns between solvers but the names of the columns are normalized so as to have the same name for the same information.
### Status
The status is given in several ways. First, a raw string extraction is returned in `status`. Then, a normalized one using codes is given via `sol_code` and `status_code` keys. `sol_code` gives information about the quality of the solution obtained. `status_code` gives details about the status of the solver after finishing (mainly, the reason it stopped).
### Other
There is also information about the pre-solving phase, the first bound and the first solution. Also, there's information about the time it took to solve the root node.
## Examples
import orloge as ol
ol.get_info_solver('tests/data/cbc298-app1-2.out', 'CBC')
Would produce the following:
{'best_bound': -96.111283,
'best_solution': None,
'cut_info': {'best_bound': -210.09571,
'best_solution': 1e+50,
'cuts': None,
'time': None},
'first_relaxed': -210.09571,
'first_solution': 1e+50,
'gap': None,
'matrix': {'constraints': 53467, 'nonzeros': 199175, 'variables': 26871},
'matrix_post': {'constraints': 26555, 'nonzeros': 195875, 'variables': 13265},
'nodes': 31867,
'presolve': None,
'progress':
Node NodesLeft BestInteger CutsBestBound Time
0 0 1 1e+50 -210.09571 32.83
1 100 11 1e+50 -210.09571 124.49
.. ... ... ... ... ...
[319 rows x 5 columns],
'rootTime': None,
'sol_code': 0,
'solver': 'CBC',
'status': 'Stopped on time limit',
'status_code': -4,
'time': 7132.49,
'version': '2.9.8'}
And another example, this time using GUROBI:
import orloge as ol
ol.get_info_solver('tests/data/gurobi700-app1-2.out', 'GUROBI')
Creates the following output:
{'best_bound': -41.0,
'best_solution': -41.0,
'cut_info': {'best_bound': -167.97894,
'best_solution': -41.0,
'cuts': {'Clique': 1,
'Gomory': 16,
'Implied bound': 23,
'MIR': 22},
'time': 21.0},
'first_relaxed': -178.94318,
'first_solution': -41.0,
'gap': 0.0,
'matrix': {'constraints': 53467, 'nonzeros': 199175, 'variables': 26871},
'matrix_post': {'constraints': 35616, 'nonzeros': 149085, 'variables': 22010},
'nodes': 526.0,
'presolve': {'cols': 4861, 'rows': 17851, 'time': 3.4},
'progress':
Node NodesLeft Objective Depth ... CutsBestBound Gap ItpNode Time
0 0 0 -178.94318 0 ... -178.94318 336% None 4s
1 0 0 -171.91701 0 ... -171.91701 319% None 15s
2 0 0 -170.97660 0 ... -170.97660 317% None 15s
[26 rows x 10 columns],
'rootTime': 0.7,
'sol_code': 1,
'solver': 'GUROBI',
'status': 'Optimal solution found',
'status_code': 1,
'time': 46.67,
'version': '7.0.0'}
Parsing the complete progress table helps anyone who later wants to analyze the raw solution process. I've tried to use the status codes and solution codes present in [PuLP](https://github.com/coin-or/pulp).
Raw data
{
"_id": null,
"home_page": "https://github.com/pchtsp/orloge",
"name": "orloge",
"maintainer": "Franco Peschiera",
"docs_url": null,
"requires_python": "",
"maintainer_email": "pchtsp@gmail.com",
"keywords": "Mathematical Optimization solver log parser",
"author": "Franco Peschiera",
"author_email": "pchtsp@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/73/42/4df880442737ce17ebe9e0ce5d491b3af1b69c7c54387e6b5be2e25023f5/orloge-0.17.2.tar.gz",
"platform": "",
"description": "## Orloge\n[](https://travis-ci.org/pchtsp/orloge)\n\n\n## What and why\n\nThe idea of this project is to permit a fast and easy parsing of the log files from different solvers, specifically 'operations research' (OR) logs.\n\nThere exist bigger, and more robust libraries. In particular, [IPET](https://github.com/GregorCH/ipet/). The trouble I had was that it deals with too many benchmarking and GUI things and I wanted something simple I could modify and build on top.\n\nIn any case, a lot of the ideas and parsing strings were obtained or adapted from IPET, to whom I am graceful.\n\nThe supported solvers for the time being are: GUROBI, CPLEX and CBC. Specially the two first ones.\n\n## How\n\nThe basic idea is just to provide a unique interface function like the following:\n\n import orloge as ol\n ol.get_info_solver(path_to_solver_log, solver_name)\n\nThis returns a python dictionary with a lot of information from the log (see *Examples* below).\n\n## Installation\n\n pip install orloge\n\nor, for the development version:\n\n pip install https://github.com/pchtsp/orloge/archive/master.zip\n\n## Testing\n\nRun the command \n\n python -m unittest test\n\n if the output says OK, all tests were passed.\n\n## Reference\n\n### Main parameters\n\nThe most common parameters to extract are: `best_bound`, `best_solution` and `time`. These three parameters are obtained at the end of the solving process and summarize the best relaxed objective value obtained, the best integer objective value obtained and the time it took to do the solving.\n\n### Cuts\n\nThe cuts information can be accessed by the `cuts_info` key. It offers the best known bound after the cut phase has ended, the best solution (if any) after the cuts and the number of cuts made of each type.\n\n### Matrix\n\nThere are two matrices that are provided. The `matrix` key returns the number of variables, constraints and non-zero values before the pre-processing of the solver. The `matrix_post` key returns these same values after the pre-processing has been done.\n\n### Progress\n\nThe `progress` key returns a raw pandas Dataframe with the all the progress information the solver gives. Including the times, the gap, the best bound, the best solution, the iterations, nodes, among other. This table can vary in number of columns between solvers but the names of the columns are normalized so as to have the same name for the same information.\n\n### Status\n\nThe status is given in several ways. First, a raw string extraction is returned in `status`. Then, a normalized one using codes is given via `sol_code` and `status_code` keys. `sol_code` gives information about the quality of the solution obtained. `status_code` gives details about the status of the solver after finishing (mainly, the reason it stopped).\n\n### Other\n\nThere is also information about the pre-solving phase, the first bound and the first solution. Also, there's information about the time it took to solve the root node.\n\n## Examples\n\n import orloge as ol\n ol.get_info_solver('tests/data/cbc298-app1-2.out', 'CBC')\n\nWould produce the following:\n\n {'best_bound': -96.111283,\n 'best_solution': None,\n 'cut_info': {'best_bound': -210.09571,\n 'best_solution': 1e+50,\n 'cuts': None,\n 'time': None},\n 'first_relaxed': -210.09571,\n 'first_solution': 1e+50,\n 'gap': None,\n 'matrix': {'constraints': 53467, 'nonzeros': 199175, 'variables': 26871},\n 'matrix_post': {'constraints': 26555, 'nonzeros': 195875, 'variables': 13265},\n 'nodes': 31867,\n 'presolve': None,\n 'progress': \n Node NodesLeft BestInteger CutsBestBound Time\n 0 0 1 1e+50 -210.09571 32.83\n 1 100 11 1e+50 -210.09571 124.49\n .. ... ... ... ... ...\n [319 rows x 5 columns],\n 'rootTime': None,\n 'sol_code': 0,\n 'solver': 'CBC',\n 'status': 'Stopped on time limit',\n 'status_code': -4,\n 'time': 7132.49,\n 'version': '2.9.8'}\n\nAnd another example, this time using GUROBI:\n\n import orloge as ol\n ol.get_info_solver('tests/data/gurobi700-app1-2.out', 'GUROBI')\n\nCreates the following output:\n\n {'best_bound': -41.0,\n 'best_solution': -41.0,\n 'cut_info': {'best_bound': -167.97894,\n 'best_solution': -41.0,\n 'cuts': {'Clique': 1,\n 'Gomory': 16,\n 'Implied bound': 23,\n 'MIR': 22},\n 'time': 21.0},\n 'first_relaxed': -178.94318,\n 'first_solution': -41.0,\n 'gap': 0.0,\n 'matrix': {'constraints': 53467, 'nonzeros': 199175, 'variables': 26871},\n 'matrix_post': {'constraints': 35616, 'nonzeros': 149085, 'variables': 22010},\n 'nodes': 526.0,\n 'presolve': {'cols': 4861, 'rows': 17851, 'time': 3.4},\n 'progress': \n Node NodesLeft Objective Depth ... CutsBestBound Gap ItpNode Time\n 0 0 0 -178.94318 0 ... -178.94318 336% None 4s\n 1 0 0 -171.91701 0 ... -171.91701 319% None 15s\n 2 0 0 -170.97660 0 ... -170.97660 317% None 15s\n [26 rows x 10 columns],\n 'rootTime': 0.7,\n 'sol_code': 1,\n 'solver': 'GUROBI',\n 'status': 'Optimal solution found',\n 'status_code': 1,\n 'time': 46.67,\n 'version': '7.0.0'}\n\nParsing the complete progress table helps anyone who later wants to analyze the raw solution process. I've tried to use the status codes and solution codes present in [PuLP](https://github.com/coin-or/pulp).\n\n\n",
"bugtrack_url": null,
"license": "",
"summary": "OR log extractor",
"version": "0.17.2",
"project_urls": {
"Download": "https://github.com/pchtsp/orloge/archive/master.zip",
"Homepage": "https://github.com/pchtsp/orloge"
},
"split_keywords": [
"mathematical",
"optimization",
"solver",
"log",
"parser"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "91921307fc6cd85edc51d95f3c6b4ceb999f50569522abf90551eab6d93b2d2a",
"md5": "8b717a7dbdc74af0e5ca8d633d9e09b8",
"sha256": "d3310d814cd574a906354a1f9ad8370c45c897b934eb4affff54ccbe0056b287"
},
"downloads": -1,
"filename": "orloge-0.17.2-py3-none-any.whl",
"has_sig": false,
"md5_digest": "8b717a7dbdc74af0e5ca8d633d9e09b8",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 11816,
"upload_time": "2021-09-08T09:49:23",
"upload_time_iso_8601": "2021-09-08T09:49:23.492062Z",
"url": "https://files.pythonhosted.org/packages/91/92/1307fc6cd85edc51d95f3c6b4ceb999f50569522abf90551eab6d93b2d2a/orloge-0.17.2-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "73424df880442737ce17ebe9e0ce5d491b3af1b69c7c54387e6b5be2e25023f5",
"md5": "e20646bc5b1542e5331467956286b6a1",
"sha256": "e44cb97e3fae5411b095e994fcb9f555d674003f48851166c92104123cc46e33"
},
"downloads": -1,
"filename": "orloge-0.17.2.tar.gz",
"has_sig": false,
"md5_digest": "e20646bc5b1542e5331467956286b6a1",
"packagetype": "sdist",
"python_version": "source",
"requires_python": null,
"size": 12929,
"upload_time": "2021-09-08T09:49:24",
"upload_time_iso_8601": "2021-09-08T09:49:24.781252Z",
"url": "https://files.pythonhosted.org/packages/73/42/4df880442737ce17ebe9e0ce5d491b3af1b69c7c54387e6b5be2e25023f5/orloge-0.17.2.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2021-09-08 09:49:24",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "pchtsp",
"github_project": "orloge",
"travis_ci": true,
"coveralls": false,
"github_actions": true,
"requirements": [],
"lcname": "orloge"
}
JSON
