Scientific computing requires handling large composed or structured matrices. Fastmat is a framework for handling large composed or structured matrices. It allows expressing and using them in a mathematically intuitive way while storing and handling them internally in an efficient way. This approach allows huge savings in computational time and memory requirements compared to using dense matrix representations.
- Python
- Cython
- Python >= 2.7 or >=3.4
- Numpy >= 1.07
- Scipy >= 1.08
- Cython >= 1.18
- matplotlib: for demos and tools that make use of plotting functions
- Sebastian Semper - Technische Universität Ilmenau, Institute for Mathematics
- Christoph Wagner - Technische Universität Ilmenau, Institute for Information Technology, EMS Group
Please have a look at the documentation, which is included in the source
distribution at github or may be built locally on your machine by running
make doc
If you experience any trouble please do not hesitate to contact us or to open an issue on our github projectpage: https://github.com/EMS-TU-Ilmenau/fastmat
Something went wrong with resolving the dependencies of fastmat during setup.
This issue will be addressed in release 0.1.1. Please check if the problem
persists with this version. You may try to bypass the problem by running
pip install cython numpy scipy
and retrying the installation of fastmat.
Often, this is causedby missing header files. Unfortunately windows ships without a c-compiler and the header files necessary to compile native binary code. If you use the Intel Distribution for Python this can be resolved by installing the Visual Studio Build tools with the version as recommended by the version of the Intel Distribution for Python that you are using.
Please contact us or leave your bug report in the issue section. Thank You!
If you want to use fastmat or parts of it in your private, scientific or commercial project or work you are required to acknowledge visibly to all users that fastmat was used for your work and put a reference to the project and the EMS Group at TU Ilmenau.
If you use fastmat for your scientific work you are further required to cite the following publication affiliated with the project:
to be announced soon. Please tune back in regularly to check on updates.
fastmat currently supports Linux, Windows and Mac OS. You may choose one of these installation methods:
fastmat is included in the Python Package Index (PyPI) and can be installed
from the commandline by running one easy and straightforward command:
pip install fastmat
When installing with pip all dependencies of the package will be installed along. With release 0.1.1 python wheels will be offered for many versions greatly improving installation time and effort.
- download the source distribution from our github repository: https://github.com/EMS-TU-Ilmenau/fastmat/archive/stable.zip
- unpack its contents and navigate to the project root directory
- run
pip install .to install fastmat on your computer - you may also install fastmat without pip, using the offered makefile:
- type
make installto install fastmat on your computer - If you intend to install the package locally for your user type
make install MODE=--userinstead. - If you only would like to compile the package to use it from this local
directory without installing it, type
make compile
- type
- if you install fastmat on a windows system please ensure you have a working and compatible compiler installed, all necessary header files in place and all setup prerequisites fulfilled.
Feel free to have a look at the demos in the demo/ directory of the source
distribution. Please make sure to have fastmat already installed when running
these.
Please note that the edgeDetect demo requires the Python Imaging Library (PIL) installed and the SAFT demos do compile a cython-core of a user defined matrix class beforehand thus having a delaying the first time they're executed.