Google Summer of Code
Forschungszentrum Jülich
About the Organization
Shaping change: open source for Big Science
Primary Technologies & Languages
pythonmpihigh performance computingpytorchgpu
Technical Topics & Focus Areas
neurosciencedata analyticsastrophysicsdata-intensive scienceearth-system monitoring
Participation Years
Completed Projects Distribution
Annual Completed Projects History
Forschungszentrum JülichHover over bars to view completed counts
4
3
2
1
0
2022
Projects4
Years Active1
Stackpython
TierTier 1
Past Projects
Showing 4 of 4 projects. Click any project card for scope, mentors, and proposal studio.
RFI Mitigation
intermediateMentors: Student: Pratham Shah
This project explores the use of image analysis in a two-dimensional spectrum to detect RFI (Radio Frequency Interference) in large amounts of memory-distributed radioastronomical data. As a result, this project aims to supports and develops 1D and 2D memory-distributed convolution within the Heat domain decomposition (or data distribution) scheme.
Interactive benchmarking analysis via Ginko Performance Explorer
beginnerMentors: Student: Tewodros Mesfin
HeAT, an array-based numerical programming framework for large-scale parallel processing with an easy-to-use NumPy-like API. HeAT utilizes PyTorch as a node-local eager execution engine and distributes the workload on arbitrarily large high-performance computing systems via MPI. This summer I will be working closely with the community and mentors to produce automatized benchmarks/scaling tests for faster detection of performance degradation and make the results easily accessible on the project’s GitHub repository.
Python array API standard compatibility
advancedMentors: Student: Neo Sun Han
Heat is a Python library for high-performance data analytics. It gives users access to multi-node processing and GPU support by seamlessly replacing NumPy operations with Heat operations. This project aims to implement an additional API for Heat that complies with the Python array API standard. This allows downstream array-consuming libraries to adopt Heat modularly with other tensor/array libraries, and makes the learning curve less steep for new users who wish to switch from other array libraries to Heat.
Memory-Distributed Singular value decomposition.
intermediateMentors: Student: V. Sai Suraj
The major goal of the project is to develop a distributed SVD algorithm that is both efficient and numerically stable in Heat. This will be a major boost as the number of applications of the SVD algorithm is high, In most of the applications basic principle of Dimensionality Reduction is used. Applications of SVD algorithm are: Image Compression, For recognition of faces, Removing Background from Videos, and Finally, the SVD algorithm is also the backbone of recommender systems such as Amazon, YouTube, Netflix, and many others.