tibotib / matrix

Sample of matrix class with fast multiplication and others basics functions.

matrix

Sample class Matrix with basics functions and fast multiplication using GPU, AVX2 instructions or openmp.

Methods of class :

• inverse(using gauss)

• decomposition PLU

• decomposition QR using gram schmidt or householder algorithm

• decomposition SVD

• Fast Fourier Transform

• Kernel of the matrix

• inverse modular

• Eigen values using QR algorithm and Eigen vectors

• Smith normal form (only on square matrix)

Example of basics functions :

#include <iostream>
#include "matrix.hpp"

int main() {
  Matrix<int> ma(std::vector<std::vector<int> > { {1, 2, 3, 4}, {4, 5, 6, 7}, {8, 9, 10, 11}, {12, 13, 14, 15}} );
  Matrix<int> mb(std::vector<std::vector<int> > { {1, 2, 3, 4}, {4, 5, 6, 7}, {8, 9, 10, 11}, {12, 13, 14, 15} } );
  
  ma.display();
  mb.display();
  
//basic matrix product
  auto mc = ma.dot(mb);
  mc.display();
  
//faster matrix product
  auto md = ma.strassen(mb);
  md.display();

//if you have NVDIA GPU
  auto m = ma.dotGPU(mb);
  m.display();
  
  return 0;
}

You have to compile with: -fopenmp -mavx2 -O3 to obtain maximum performance.

If you want to use your GPU compile with nvcc and compile the kernel.cu file