A library for solving an elliptic differential equation in 2D using Finite Element Method (FEM) and Quantized Tensor Train (QTT) decomposition.
This library uses a special operation Z-kron to avoid approximation ranks growing in the QTT-format. This operation enables to build a matrix in the QTT-format with special rows and columns ordering -- z-order. As a result, this approach has slow approximation ranks growing and consumes much less memory than FEniCS.
The left plot is an energy error from vertices number. The middle plot is a erank growing from approximation accuracy. The right plot is a peak memory consumption from vertices number.
Coming soon
Clone the repository:
git clone git://github.com/RerRayne/qtt-laplace.git
Run setup.py from cloned dir:
python setup.py install
Jupyter notebooks are available right now under examples directory.
For any questions, please create an issue on Github.
This package is made by Markeeva L. and Tsybulin I.