GSoC '22 contributors WELCOME!

[ClaudiaComito]

Thanks for your interest!

---

UPDATE Mar 30th 2022

A summary of Heat resources:

• Our GSoC22 contributor guidance
• Our docs including Tutorials
• Jupyter Notebook illustrating the basics
• More details on contributing to Heat
• Parallel programming / MPI / mpi4py resources
• we held an AMA on March 29th. You can access it on our Mattermost (sign up with your GitHub credentials, join the GSoC22 channel)

---

Here's our List of Ideas, and our primary mentors:

1. Python array-API standard compatibility @mtar
2. Memory-distributed sparse operations @ClaudiaComito
3. Memory-distributed 2D convolution @ClaudiaComito
4. Memory-distributed Singular Value Decomposition @coquelin77
5. Data-Parallel Neural Network Training with DASO @coquelin77
6. mpi4torch: An automatic-differentiable MPI wrapper for pytorch @d1saster (Philipp Knechtges)
7. Support HIP/ROCm @mtar
8. Interactive benchmarking analysis @ClaudiaComito and @bhagemeier

We're looking forward to hearing from you!

EDIT: feel free to sign up to our Mattermost (you can sign up with your GitHub credentials) and join the GSoC22 channel.

[Mystic-Slice]

Hi!
Hope you are all doing well!
I am Ashwath. I am a sophomore undergraduate student majoring in Computer Science Engineering. I came across your project and found it very interesting. I am particularly interested in "Heat for Neuroscience: Memory-distributed sparse operations" (Project Idea 2) and "Data-Parallel Neural Network Training with DASO" (Project Idea 5). I would love to know more about the projects. Should I ask my questions here or would it be better if I talked with the respective mentors through the mail?
Also, if you have any questions for me, please feel free to ask!

[Mystic-Slice]

Questions about the DASO project:

1. Is there a list of sorts for the features that you would expect this project to implement?

2. Just to be certain, NCCL is Nvidia's Communications library, right?

3. The skill requirements state that I would have to know parallel/distributed computing. I have a basic understanding of parallel programming. I learned it as a part of a Java course that I took in college. I have good experience working with asynchronous programming, which is not exactly parallel programming, but close! I have worked on apps built on Qt(C++) and Electron.js(JavaScript), Android(Java) frameworks where I have employed async programming to make the apps more responsive. I have used a message-passing model as well for communication. I would like to learn more about parallel programming. Are there any resources that you would suggest? I am currently going through this course to know more. [ANSWERED]

[Mystic-Slice]

Questions about Memory-distributed sparse operations:

1. As far as I understand, I have to implement HeAT's version of the Pytorch sparse. Am I right in assuming that? If so, will I be more focussed on implementing these functionalities along with the sparse data structure?

2. What is meant by the memory-distributed linear operator mentioned in the project description?
I tried searching about it but could not find anything. Maybe, I did find it and did not know that I was looking at it.

[Mystic-Slice]

Going through the project ideas again. I liked "Heat for SKA: memory-distributed 2D convolution" also.
"An implementation of one-dimensional distributed convolution for Heat is in progress"
What is the status of this? Is the code available anywhere for me to view? If it is still in progress, maybe in the developers' fork or smth?

[Mystic-Slice]

Sorry for flooding this discussion with questions. I ask these questions only because I am very interested. I hope you don't find it a nuisance.

[ClaudiaComito]

Going through the project ideas again. I liked "Heat for SKA: memory-distributed 2D convolution" also. "An implementation of one-dimensional distributed convolution for Heat is in progress" What is the status of this? Is the code available anywhere for me to view? If it is still in progress, maybe in the developers' fork or smth?

see #939

[samiali12]

Hi, my Self Sami Ali, currently pursuing BS in Software Engineering at COMSATS UNIVERSITY OF ISLAMABAD, WAH CAMPUS. I came across your project and found it very interesting. I am particularly interested in Data-Parallel Neural Network Training with DASO (Project Idea 5), Memory-distributed sparse operations (Project Idea 2). I would love to know more about the projects. Should I ask my questions here or would it be better if I talked with the respective mentors through the mail?

[tikka01]

Hi! I am Chaitika, an undergraduate student majoring in Mathematics. I'm interested in Project idea 01- Python array-API standard compatibility and I was hoping that I could connect with someone with whom I can discuss the project in depth.

[ClaudiaComito]

Thank you so much @Mystic-Slice @samiali12 @tikka01 for your questions! Several general topics are arising, I'll collect them in separate threads.

• Parallel programming resources

More later!

[tikka01]

Hi Claudia , I was hoping that maybe there is someway that I can get in touch with my primary mentor @mtar for discussing Idea -1.

[Mystic-Slice]

Hi @ClaudiaComito. I asked some questions about the project ideas a few days ago. I think you might have missed these questions since the thread is collapsed.
Memory distributed sparse operations: #919 (reply in thread)
Memory distributed 2-D Convolution: #919 (reply in thread)
I am more interested in these two projects. Would love to discuss the ideas more.

Also, I am not able to access Mattermost. Any idea why that might be?

[ClaudiaComito]

Good catch, that is an old Mattermost link. Try this one, you can sign up with your GitHub credential, and join the GSoC22 channel.

I'll get to your other questions tomorrow.

[sahilvora10]

Hi!

I am Sahil Vora. I am currently working as a Software Developer, and I have completed my undergraduate studies majoring in Computer and Communication Engineering.
I came across your project and organization and found it very interesting.
I am particularly interested in Heat general: Python array-API standard compatibility (Project 1) and would love to know more about the projects. Should I ask my questions here, or is there any other channel I could join?
Also, if you have any questions for me, please feel free to ask!

[ClaudiaComito]

Ping, original post updated

[RATED-R-SUNDRAM]

Hello Team, I am shivam sundram 3rd year undergraduate at IIT Mandi, I have 3 years (including one year of professional experience in python). I have been an active contributor in open source and would like to contribute in this community for the this year's GSOC. The idea I am intrested in and would like to contribute is the (PROJECT IDEA 7) . Please help me in getting started and refering me to resources required for this project.

[RATED-R-SUNDRAM]

Hey @mtar please provide some input on the project 7 so I can finalize the draft making.

[mtar]

Hi @RATED-R-SUNDRAM, have you looked at the links to PyTorch and our devices.py file in the List of Ideas? As stated there, we want to use PyTorch's HIP interface for AMD GPU support and set up a testing environment eventually.

Questions regarding idea 7 can now be posed in Discussion #963.

[Choon38]

Hello, I am Sneha first year undergraduate student from India. I am interested in project idea 4 (heat for engineering: memory- distributed SVD) and project 1(Heat general: python array-API standard compatibility). I want to know in depth of project idea 4. Kindly provide me resources related to this.

[anirudha-mirmira]

Hi, I am Anirudha Mirmira, a final year theoretical physics PhD student in JNCASR, India. I work in Condensed Matter physics and predominantly write Fortran programs implementing different methods to solve non-equilibrium quantum systems. I have used MPI for parallelization in my code and have used convolutions (1d) via FFT in my calculations. I have working knowledge of Python and have basic familiarity with the pytorch framework, though am very interested in learning more about the same. I also have some familiarity with OpenACC and OpenMP. This is also the first time I have come across the open-source community and am very enthused to contribute.

I am interested in the sparse matrix and the 2D convolution projects and am very excited to have an opportunity to work on the same. Coming to the time commitment, since I am finalising my thesis, I have quite a hectic schedule and will only be able to dedicate a couple of hours every day in the evenings (IST- UTC+5:30). I'm looking forward to contributing both as a part of GSoC and otherwise.

[BohrCh]

Hello, everyone!
I am Bo Chen, a 2nd-year graduate student at the University of Science & Technology Beijing. My graduation research project is "Molecular dynamics optimization on heterogeneous processors" I use the MPI and the OpenMP in my project. I found this organization by searching "high-performance computing" I was an intern at Baidu Inc. last year. I was responsible for benchmarking the performance of cuDNN on the A100. So I am interested in the "Interactive benchmarking analysis via Ginko Performance Explorer Description"(PROJECT IEDA 8) and "Mpi4torch: An automatic-differentiable MPI wrapper for PyTorch"(PROJECT IEDA 6). Is it too late to apply these projects? @d1saster @ClaudiaComito @bhagemeier @ClaudiaComito

[bhagemeier]

Dear @BohrCh , sorry for the late reply. It should never be too later for handing in something before the deadline. If you prepared anything, you'd have to hand it in today. The contributor application deadline is at 18:00 UTC.

[aditya-shrivastavv]

helmholtz-analytics | is there any updates regarding GSoC 2023

[ClaudiaComito]

Hi @aditya-shrivastavv thanks for your interest!

We will certainly apply as GSoC Mentoring Organization again. We will know on Feb 22nd whether we are in. Feel free to check out our Issues in the meantime, we'll keep posting new issues related to potential GSoC projects.