SIDEMADE - Simple Dense Matrix Decomposition

A simple demo code to read dense matrices from files and solve linear systems with them.

Folder Structure

• cmake: Third-party CMake modules
• CMakeLists.txt: CMake configuration file
• data: Various dense matrices
  • config.yml: SIDEMADE configuration file
  • m3.csv: A 3x3 matrix with randomly generated entries
  • m10.csv: An ill-conditioned 10x10 Hilbert matrix
  • m50.csv: A 50x50 FEM mass matrix for a 1D problem with Lagrange basis
  • m100.csv: A 100x100 FEM stiffness matrix for a 1D problem with Lagrange basis
  • m250.csv: A 250x250 matrix with randomly generated entries
• LICENSE: MIT license file
• src: Source code
  • Configuration.hpp/cpp: A yaml configuration reader
  • MatrixSolver.hpp/cpp: A convenience hull to solve dense matrix system by matrix decompositions
  • matrixIO.hpp/cpp: IO methods to read Eigen matrices from or write them to csv files
  • main.cpp: Main file
• tests: Unit tests

Building

You can build the code using CMake:

$ mkdir build
$ cd build
$ cmake ..
$ make -j

Dependencies

• A C++ compiler supporting at least C++11
• CMake: At least version 3.12
• Eigen: At least version 3.2
• yaml-cpp: At least version 0.6

For unit testing moreover:

• Boost Unit Test Framework: Should work with most recent versions. The code was tested with 1.71.

Running

You need to call the code with a configuration file, e.g. config.yml. If you call the program from the build folder:

$ ./sidemade ../data/config.yml

Configuration

An example configuration file:

DecompositionType: QR
MatrixFileName: ../data/m250.csv
MatrixSize: 250

Tools

Formatting

You can format the code base using clang-format:

$ clang-format -i src/*