# ttopt
## Description
Gradient-free optimization method for multivariable functions based on the low rank tensor train (TT) format and maximal-volume principle.
> Please, see also our software product [teneva](https://github.com/AndreiChertkov/teneva) which provides a very compact implementation of basic operations in the TT-format.
## Installation
You can install the `ttopt` package for `python >= 3.7` with pip:
```bash
pip install ttopt==0.6.2
```
## Examples
The demo-scripts with detailed comments are collected in the folder `demo`:
- `base.py` - we find the minimum for the 10-dimensional function with vectorized input;
- `qtt.py` - we do almost the same as in the `base.py` script, but use the QTT-based approach (note that results are much more better then in the `base.py` example);
- `qtt_max.py` - we do almost the same as in the `qtt.py`, but consider the maximization task;
- `qtt_100d.py` - we do almost the same as in the `qtt.py` script, but approximate the 100-dimensional function;
- `vect.py` - we find the minimum for the simple analytic function with "simple input" (the function is not vectorized);
- `cache.py` - we find the minimum for the simple analytic function to demonstrate the usage of the cache;
- `tensor.py` - in this example we find the minimum for the multidimensional array/tensor (i.e., discrete function);
- `tensor_init` - we do almost the same as in the `tensor.py` script, but we use special method of initialization (instead of a random tensor, we select a set of starting multi-indices for the search).
## Authors
- [Andrei Chertkov](https://github.com/AndreiChertkov)
- [Ivan Oseledets](https://github.com/oseledets)
- [Roman Schutski](https://github.com/Qbit-)
- [Konstantin Sozykin](https://github.com/gogolgrind)
## Citation
If you find this approach and/or code useful in your research, please consider citing:
```bibtex
@article{sozykin2022ttopt,
author = {Sozykin, Konstantin and Chertkov, Andrei and Schutski, Roman and Phan, Anh-Huy and Cichocki, Andrzej and Oseledets, Ivan},
year = {2022},
title = {{TTOpt}: {A} maximum volume quantized tensor train-based optimization and its application to reinforcement learning},
journal = {Advances in Neural Information Processing Systems},
volume = {35},
pages = {26052--26065},
url = {https://proceedings.neurips.cc/paper_files/paper/2022/hash/a730abbcd6cf4a371ca9545db5922442-Abstract-Conference.html}
}
```
> Please, note that the calculations presented in this paper correspond to version `<0.5.0` of the `ttopt` package (and to very old version of the `teneva` package), to run the calculations, please use the appropriate version. In the new versions `>=0.6.0`, we have removed all the corresponding folders in the folder `computations_old`. In the future, we will try to update the interface of these experiments.
---
> ✭__🚂 The stars that you give to **ttopt**, motivate us to develop faster and add new interesting features to the code 😃
Raw data
{
"_id": null,
"home_page": "https://github.com/AndreiChertkov/ttopt",
"name": "ttopt",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.7",
"maintainer_email": "",
"keywords": "function optimization function minimization low-rank representation tensor train format TT-decomposition cross approximation",
"author": "Andrei Chertkov",
"author_email": "andre.chertkov@gmail.com",
"download_url": "https://files.pythonhosted.org/packages/06/2f/591163d11d33de8a691808ba13a54bc370b9e5556653e8dc26cd5a714ccf/ttopt-0.6.2.tar.gz",
"platform": null,
"description": "# ttopt\n\n\n## Description\n\nGradient-free optimization method for multivariable functions based on the low rank tensor train (TT) format and maximal-volume principle.\n\n> Please, see also our software product [teneva](https://github.com/AndreiChertkov/teneva) which provides a very compact implementation of basic operations in the TT-format.\n\n\n## Installation\n\nYou can install the `ttopt` package for `python >= 3.7` with pip:\n```bash\npip install ttopt==0.6.2\n```\n\n\n## Examples\n\nThe demo-scripts with detailed comments are collected in the folder `demo`:\n\n- `base.py` - we find the minimum for the 10-dimensional function with vectorized input;\n- `qtt.py` - we do almost the same as in the `base.py` script, but use the QTT-based approach (note that results are much more better then in the `base.py` example);\n- `qtt_max.py` - we do almost the same as in the `qtt.py`, but consider the maximization task;\n- `qtt_100d.py` - we do almost the same as in the `qtt.py` script, but approximate the 100-dimensional function;\n- `vect.py` - we find the minimum for the simple analytic function with \"simple input\" (the function is not vectorized);\n- `cache.py` - we find the minimum for the simple analytic function to demonstrate the usage of the cache;\n- `tensor.py` - in this example we find the minimum for the multidimensional array/tensor (i.e., discrete function);\n- `tensor_init` - we do almost the same as in the `tensor.py` script, but we use special method of initialization (instead of a random tensor, we select a set of starting multi-indices for the search).\n\n\n## Authors\n\n- [Andrei Chertkov](https://github.com/AndreiChertkov)\n- [Ivan Oseledets](https://github.com/oseledets)\n- [Roman Schutski](https://github.com/Qbit-)\n- [Konstantin Sozykin](https://github.com/gogolgrind)\n\n\n## Citation\n\nIf you find this approach and/or code useful in your research, please consider citing:\n\n```bibtex\n@article{sozykin2022ttopt,\n author = {Sozykin, Konstantin and Chertkov, Andrei and Schutski, Roman and Phan, Anh-Huy and Cichocki, Andrzej and Oseledets, Ivan},\n year = {2022},\n title = {{TTOpt}: {A} maximum volume quantized tensor train-based optimization and its application to reinforcement learning},\n journal = {Advances in Neural Information Processing Systems},\n volume = {35},\n pages = {26052--26065},\n url = {https://proceedings.neurips.cc/paper_files/paper/2022/hash/a730abbcd6cf4a371ca9545db5922442-Abstract-Conference.html}\n}\n```\n\n> Please, note that the calculations presented in this paper correspond to version `<0.5.0` of the `ttopt` package (and to very old version of the `teneva` package), to run the calculations, please use the appropriate version. In the new versions `>=0.6.0`, we have removed all the corresponding folders in the folder `computations_old`. In the future, we will try to update the interface of these experiments.\n\n\n---\n\n\n> \u272d__\ud83d\ude82 The stars that you give to **ttopt**, motivate us to develop faster and add new interesting features to the code \ud83d\ude03\n",
"bugtrack_url": null,
"license": "",
"summary": "Multivariate function optimizer based on the tensor train approach.",
"version": "0.6.2",
"project_urls": {
"Homepage": "https://github.com/AndreiChertkov/ttopt",
"Source": "https://github.com/AndreiChertkov/ttopt"
},
"split_keywords": [
"function",
"optimization",
"function",
"minimization",
"low-rank",
"representation",
"tensor",
"train",
"format",
"tt-decomposition",
"cross",
"approximation"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "51a29479574706205649a7e316fb178bc32f882df43f21607a8bbb36dd4c742e",
"md5": "f44567565e322c112c5933db035cc9ae",
"sha256": "ab6ac5e1261eaa4e34fb0208de27f296904de78ff62b4b6a318ad2f36dfa52ee"
},
"downloads": -1,
"filename": "ttopt-0.6.2-py3-none-any.whl",
"has_sig": false,
"md5_digest": "f44567565e322c112c5933db035cc9ae",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.7",
"size": 15520,
"upload_time": "2023-07-27T20:14:23",
"upload_time_iso_8601": "2023-07-27T20:14:23.847836Z",
"url": "https://files.pythonhosted.org/packages/51/a2/9479574706205649a7e316fb178bc32f882df43f21607a8bbb36dd4c742e/ttopt-0.6.2-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "062f591163d11d33de8a691808ba13a54bc370b9e5556653e8dc26cd5a714ccf",
"md5": "e2c0cfb1f6cf7f5a3d779bb82235bf89",
"sha256": "77d7cd58e41e4f70f2451f013c96c7992db0cd034b99c5288ab5014dca502b83"
},
"downloads": -1,
"filename": "ttopt-0.6.2.tar.gz",
"has_sig": false,
"md5_digest": "e2c0cfb1f6cf7f5a3d779bb82235bf89",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.7",
"size": 19042,
"upload_time": "2023-07-27T20:14:25",
"upload_time_iso_8601": "2023-07-27T20:14:25.645657Z",
"url": "https://files.pythonhosted.org/packages/06/2f/591163d11d33de8a691808ba13a54bc370b9e5556653e8dc26cd5a714ccf/ttopt-0.6.2.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-07-27 20:14:25",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "AndreiChertkov",
"github_project": "ttopt",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"requirements": [],
"lcname": "ttopt"
}
JSON
