MNE-Python reimplements common M/EEG processing algorithms in pure Python. In addition, it also implements new algorithms, proposed and only recently published by the MNE-Python authors, making them publicly available for the first time (Gramfort et al., 2010a, 2011, 2013b; Larson and Lee, 2013). To achieve this task, MNE-Python is built on the foundation of core libraries provided by the scientific Python environment: NumPy (Van der Walt et al., 2011) offers the n-dimensional array data structure used to efficiently store and manipulate numerical data; SciPy is used mainly for linear algebra, signal processing and sparse matrices manipulation; matplotlib (Hunter, 2007) is used for 2D graphics; Mayavi (Ramachandran and Varoquaux, 2010) is employed for 3D rendering; Scikit-Learn [Pedregosa et al. (2011) and Buitinck et al. (2013)] is required for decoding tasks; and the Python Data Analysis Library (Pandas) is used for interfacing with spreadsheet table oriented data processing tools as often used in econometrics and behavioral sciences. Mayavi, Scikit-Learn and Pandas are only required by a small subset of the code, and are therefore considered optional dependencies. Besides these general libraries, MNE-Python
