(GSoC '21) uBlas Project 2/3 Guidance

From: Laxman (desai.laxman2001_at_[hidden])
Date: 2021-03-18 01:05:25

   Hi,

   I, Laxman Desai, am interested in ublas Project 2 and 3.

   I’ve built a few C++ libraries before (for personal use) and I am
   familiar with C++17, MATLAB, TensorFlow, NumPy and Pandas to a good
   extent. I am currently learning R. I also have a good mathematical
   background, especially in Linear Algebra.

   I have currently completed the Competency Test for uBlas, and would
   request the mentors: David Bellot, Cem Bassoy or Amit Singh to guide me
   further for Project 2: creating new matrix and vector types and Project
   3: Finalize Dataframe Implementation.

   Project 2:

   I dont understand what experimental templates and
   ublas::experimental::matrix means. I although have a strong feeling
   that it just refers to the tensor implementation of the same. I haven’t
   used uBlas tensor_core before, so could a mentor explain further (a
   basic intuition of the improvement, and the workflow for this project)?

   I would also appreciate more information on submatrix, and its
   implementation! To my understanding, it is to solve the problem that
   C++ doesn’t allow overloading of the colon operator.

   Project 3:

   I also am interested in project 3 (Finalising the DataFrames
   Implementation), and am currently exploring the available features. I
   have been using Pandas for a while now, and recently learnt R.

   “Finalize and productionize” seemed a bit vague to me, could I get more
   information on what all the current implementation is lacking (except
   for documentation)?

   I would also like to know if the mentors have a shortlist of features
   to add in mind?

   Any feedback on this would be greatly appreciated.

   Kind Regards,

   Laxman Desai

   Mail: desai.laxman2001_at_[hidden]