Introduction to the Pycroscopy Ecosystem: https://pycroscopy.github.io/pycroscopy/ecosystem.html
Brief Overview of Pycroscopy Package for scientific analysis of nanoscience data: https://pycroscopy.github.io/pycroscopy/about.html
Homepage: https://github.com/pycroscopy
Download Anaconda: https://www.anaconda.com (This should automatically install Anaconda-Navigator and allow you to lauch Jupyter Notebooks.)
Install NumPy: https://numpy.org/install/
Install sidpy package through Anaconda: https://anaconda.org/conda-forge/sidpy
Install h5py: https://docs.h5py.org/en/stable/build.html
The notebooks for their respective respositories are available in the files section.
Tools for extracting data and metadata from scientific data files
SciFiReaders is a collection of sidpy.Reader python classes that extract data and metadata from scientific data files. The extracted information are returned as sidpy.Dataset objects which are standardized and exchangable data objects across all packages in the pycroscopy ecosystem.
About: https://pycroscopy.github.io/SciFiReaders/about.html
• Notebook - Developing a reader: https://github.com/pycroscopy/SciFiReaders/blob/master/notebooks/00_developing_a_reader/developing_a_reader.ipynb
• Notebook - Using Readers: https://github.com/pycroscopy/SciFiReaders/blob/master/notebooks/01_using_readers/convert_hyperspy.ipynb
Coming soon: short-video with example
Python utilities for storing and visualizing Spectroscopic and Imaging Data (SID)
Installation: https://pycroscopy.github.io/sidpy/install.html
• Notebook - Basic Usage: https://pycroscopy.github.io/sidpy/notebooks/00_basic_usage/create_dataset.html
• Notebook - Parallel Computing: https://github.com/pycroscopy/sidpy/blob/master/notebooks/01_parallel_computing/parallel_compute.ipynb
• Notebooks - Visualization: https://github.com/pycroscopy/sidpy/tree/master/notebooks/02_visualization
• Notebooks - hdf5: https://github.com/pycroscopy/sidpy/tree/master/notebooks/03_hdf5
Python framework for storing, visualizing, and processing spectroscopy, imaging or any observational / experimental data
About: https://pycroscopy.github.io/pyNSID/about.html
• Notebook - Write SIDpy Dataset via pyNSID: https://github.com/pycroscopy/pyNSID/blob/master/notebooks/00_basic_usage/write_read_sidpy_dataset.ipynb
Why not just use h5py? -> https://pycroscopy.github.io/pyNSID/nsid.html
An open-source file format that supports large, complect, heterogeneous data. Think of it as a file that has a file directory structure, allowing you to organize data within the file like you would on your own computer.
-
There are two important terms you want to recognize within these types of files:
o Group: a folder like element within your HDF5 that can contain even more groups OR datasets (referred to as “keys” in python code).
o Dataset: this is the actual data contained within the HDF5 file
If you already have an HDF5, follow the following structure for identifying contents, shapes of datasets, and plotting.
In this example, we are using a small sample of PDMS Gel from Bruker premounted on a 12mm magnetic disk.
After setting your file path as directed in the Basic Usage notebook, use the following command to print out the contents of your HDF5 file:
usid.hdf_utils.print_tree(h5_f)
Here, you can see that you have two groups or "keys" in your file with the following datasets.
To see the keys and shape of your datasets use the following command:
print(h5_f.keys())
print(h5_f['FFM']['Defl'][:].shape, h5_f['FFM']['Drive'][:].shape, h5_f['FFM']['Raw'][:].shape)
The following Jupyter Notebook will walk through the whole process of visualizing your data: