Lightning Talk: Let's Fix Sparse Linear Algebra with C++. It'll Be Fun and Easy!

on YouTube ↗06m

" summary

Claim: C++ can make sparse linear algebra easier by exposing customization points that return iterable views and by leveraging the ranges library plus high-level multi-dimensional iteration. Implementation detail: range adapters and a row iterable concept map CSR and other compressed formats to zipped value/index views so generic algorithms can run without index arithmetic. Demo code compiles and mostly runs during the lightning slot with a brief bug, showing feasibility rather than a finished library.

AI-written summary · about

" chapters (5)

1. 0:00Intro
2. 0:41What is sparse linear algebra
3. 1:47Range adapters
4. 2:29Implementing views
5. 3:42Code demos

" related talks

• 01h 04mCompile-time sparse matricesDaniel Withopf · Meeting C++ 2023
• 22mViewing The World Through Array-Shaped GlassesŁukasz Mendakiewicz · CppCon 2014
• 06mLightning Talk: Quantum Interpretations of the C++ Object ModelDavid Stone · CppCon 2021
• 28mCommunication Avoiding Algorithms for Linear Algebra and BeyondJim Demmel · Parallel and Distributed Algorithms for Inference and Optimization (Simons Institute, Oct. 21–24, 2013)
• 23mPattern-Based Optimization (...)Dániel Berényi · Lambda Days 2018
• 48mC++ Algorithmic Complexity, Data Locality, Parallelism, Compiler Optimizations, & Some ConcurrencyAvi Lachmish · CppCon 2022
• 44mFunctional Array Funhouse Intensive - Part 5 of 5Aaron W Hsu · LambdaConf 2017
• 01h 03mCompute More in Less Time Using C++ Simd Wrapper LibrariesJefferson Amstutz · CppCon 2018