Python implementation of the Tensor Train (TT) -Toolbox. It contains several important packages for working with the TT-format in Python. It is able to do TT-interpolation, solve linear systems, eigenproblems, solve dynamical problems. Several computational routines are done in Fortran (which can be used separatedly), and are wrapped with the f2py tool.

It is recommended that you use Anaconda Python distribution which has MKL library built-in. Anaconda Python is used for the development of ttpy.

Install dependencies (numpy and cython)

   conda install numpy cython

Install ttpy

   pip install ttpy

To install the development version, you need to install from the source. First, clone the repository with all submodules:

   git clone --recursive git://github.com/oseledets/ttpy.git

   python setup.py install

To update to the latest version (with all submodules) run

git pull
git submodule update --init --recursive *

They have the following functionality

• tt : The main package, with tt.vector and tt.matrix classes, basic arithmetic, norms, scalar products, rounding full -> tt and tt -> full conversion routines, and many others
• tt.amen : AMEN solver for linear systems (Python wrapper for Fortran code written by S. V. Dolgov and D. V. Savostyanov) it can be also used for fast matrix-by-vector products.
• tt.eigb : Block eigenvalue solver in the TT-format
• tt.ksl : Solution of the linear dynamic problems in the TT-format, using the projector-splitting KSL scheme. A Python wrapper for a Fortran code (I. V. Oseledets)
• tt.cross : Has a working implementation of the black-box cross method.

The package provides Sphinx-generated documentation. To build HTML version, just do

cd tt/doc
make html

A few examples are available right now under examples directory

For any questions, please create an issue on Github.

This project is now following the git flow approach. Namely:

• branch master is only for stable versions and releases;
• branch develop is main working branch;
• contributor should create new branch for certain feature and then merge with develop branch as feature was done;
• each release on master branch should correspond to package on PyPI;
• A maintainer checks all the pull request

A pull request should satisfy the following requirements:

• style and quality description of pull request;
• new changes should be tested and shouldn't break anything;
• pull request for one fix or one feature(could be several commits);
• try to keep the code style of the project;

Current maintainer is Ivan Oseledets.