tSVD is a python library for calculating Singular Value Decomposition with threshold. User could provide input threshold for singular values if the threshold is known.
tSVD is a Python library based on Eigen3. Before installing tSVD, please make sure that you have Eigen3 installed in your system.
In MacOs, entrance point of the Eigen header file is normally under the path
/usr/local/include/eigen3.
In Linuxs or Windows systems, the Eigen header file is normally under the path:
/usr/include/eigen3 .
Once you have Eigen3 in your system, setup this path in setup.py.
For example :
eigen_path = ['/usr/local/include/eigen3']
One can install tSVD by building Python extension using pip .
python setup.py build_ext --inplace
python setup.py bdist_wheel
cd dist && pip install tSVD-0.1-cp37-cp37m-macosx_10_15_x86_64.whlfrom cylib.tSVD import *
import numpy as np
threshold = 1
accuracy = 0.0001
a = np.arange(1,16, dtype = np.float64).reshape(3,5)
u,s,vt = partial_svd(a, threshold, accuracy)For this naive version, the input data type could only be limited to np.float64
In the Jupyter Notebook under demo/demo.ipynb, you could find a
demonstration of one of the applications of this library. With tSVD, we could
use a threshold value to compress and decompress an image. Even more, selecting
only the large singular values avoid expensive computation of the smaller eigen
values, and reduce the noises in the image.
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.