[boost] [GSOC 19] Boost.ublas project Idea

Subject: [boost] [GSOC 19] Boost.ublas project Idea
From: Amit Singh (amitsingh19975_at_[hidden])
Date: 2019-02-02 12:46:36

Hi,

     My name is Amit Singh and I’m a 2nd year student.

   1. I have studied last years ublas stats project and before that ublas
   lacked any statistics features, I would like to take boost.ublas stats
   further with

   1. Linear Regression
   2. Logistic Regression
   3. Rigid Regression
   4. Kernels ( Exponential Kernel, Gaussian Kernel, etc)
   5. Correlation (Pearson product-moment correlation, Linear correlation,
   etc)

2.  In ublas there is no expression solver which can solve a expression and
assign the value to

     variable for ex:-

A. matrix<double> m = “A^2 * (B * A^-1) + (det(A) * I)”;

B. matrix<double> m = “A + 3” ;  =>[ m = A + 3*I ]

3. In ublas to modify matrix you have to iterate using for loop. I’m hoping
to create a function

    similar to transform in c++ or map in javascript

   1. This is my small machine learning library that I implemented in c++
   https://github.com/amitsingh19975/ML

Would anybody be interested in mentoring my project for gsoc 19? or help me
to improve my project ideas.