Overview
Explore GPU-accelerated crystallography techniques using Python in this EuroPython 2017 conference talk. Discover how the Max Planck Computing and Data Facility leverages Python's data ecosystem for atom probe crystallography, scaling across multiple GPUs. Learn about the PyNX software package, which utilizes pyCUDA and pyOpenCL libraries for fast parallel computation scattering. Gain insights into the data workflow analysis process, from initial exploratory data analysis using Jupyter notebooks and Python packages like pandas, matplotlib, and plotly, to production-stage interactive visualization with Paraview. Understand the comparison between CPU and GPU performance, Python scripting techniques, and data inspection methods. Dive into topics such as VTK file generation, user interface development, and the creation of filter collections and contour plots for effective data visualization.
Syllabus
Intro
CPU vs GPU
Visualization
VTK
PartU
PartU Python
PartU User Interface
Python Confidence
Python scripting
Filter collection
Contour plot
Data inspection
Viewing data
Conclusion
Questions
Taught by
EuroPython Conference