Installing HyperSpy¶

HyperSpy Bundle for Microsoft Windows¶

The easiest way to install HyperSpy in Windows is installing the HyperSpy Bundle. This is a customised WinPython distribution that includes HyperSpy, all its dependencies and many other scientific Python packages. HyperSpy Bundle does not interact with any other Python installation in your system, so it can be safely installed alongside other Python distributions. Moreover it is portable, so it can be installed in an USB key. When intalling it with administator rights for all users it adds context menu entries to start the Jupyter Notebook and Juypter QtConsole. See start_jupyter_cm for details.

Quick instructions to install HyperSpy using Anaconda (Linux, MacOs, Windows)¶

1. Download and install Anaconda. Anaconda is recommended for the best performance (it is compiled using Intel MKL libraries) and the easiest intallation (all the required libraries are included). The academic license is free.

   $ conda install -c conda-forge traits traitsui
   $ pip install hyperspy
   

$ pip install --upgrade hyperspy==0.8.3-1

For convenience you may also consider installing start_jupyter_cm.

For more options and details read the rest of the documentation.

Install using Python installers¶

HyperSpy is listed in the Python Package Index. Therefore, it can be automatically downloaded and installed pip. You may need to install pip for the following commands to run.

Install using pip:

$ pip install hyperspy

$ pip install --upgrade hyperspy==0.8.3-1

pip installs automatically the stricly required libraries. However, for full functionality you may need to install some other dependencies, see Installing the required libraries.

Creating Conda environment for HyperSpy¶

Anaconda Python distribution can be easily set up using environment files. The two required steps are:

1. Download HyperSpy environment file.
2. Create and activate HyperSpy environment according to instructions here. For Unix, the following should work:

$ conda env create -f anaconda_hyperspy_environment.yml
$ source activate hyperspy

Install from a binary¶

We provide binary distributions for Windows (see the Downloads section of the website). To install easily in other platforms see Install using Python installers

Install from source¶

$ tar -xzf hyperspy.tar.gz
$ cd hyperspy
$ python setup.py install

$ pip install hyperspy.tar.gz

$ git clone https://github.com/hyperspy/hyperspy.git

$ cd hyperspy
$ pip install -e ./

$ cd hyperspy
$ sudo pip install -e ./

$ ./release_debian

Installing the required libraries¶

When installing HyperSpy using Python installers or from source the Python programming language and the following libraries must be installed in the system: numpy, scipy, matplotlib (>= 1.2), ipython, natsort, traits and traitsui. For full functionality it is recommended to also install h5py, scikit-learn and ipywidgets. In addition, since version 0.7.2 the lowess filter requires statsmodels. In Windows HyperSpy uses the Ipython’s QtConsole and therefore Qt and PyQt or PySide are also required.

In Debian/Ubuntu you can install the libraries as follows:

$ sudo apt-get install python-numpy python-matplotlib ipython
ipython-notebook python-traits python-traitsui python-h5py
python-scikits-learn python-nose python-statsmodels

In Arch Linux, the following command should install the required packages to get a fully functional installation:

$ sudo pacman -Sy python2 python2-numpy     python2-matplotlib      python2-pip
python2-traits python2-traitsui python2-h5py python2-scikit-learn python2-nose
python2-statsmodels python2-pillow python2-pyqt4 python2-pyqt5 python2-scipy
python2-pandas python2-setuptools ipython2  python2-jinja python2-pyzmq
python2-pyqt4 python2-tornado python2-sip python2-pygments

# Or, just run this command from the root hyperspy directory to import the
# list of packages and install automatically:
$ xargs sudo pacman -Sy --noconfirm < doc/package_lists/arch_linux_package_list.txt

# Once these are installed, go to the HyperSpy directory and run:
$ sudo pip2 install -e ./

# If desired, the python2-seaborn library can also be installed from AUR for prettier plotting

Known issues¶