Bogus, a template C++ sparse block matrix library

Overview

This free software consist of two complementary libraries, the first being fairly generic while the other targets more specific problems. Both are open-source and can be downloaded from the BitBucket repository.

bogus

is a template C++ 98 header-only sparse block matrix library. It allows to express in a concise way algebraic operations on sparse matrices whose elements can be a variety of block types, such as Eigen dense or sparse matrices.

So-bogus

is a library that uses and completes bogus to solve systems with second order cone constraints, such as Coulomb friction problems.

Main features

bogus

Simple creation of sparse block matrices with several storage options (row or column major, symmetric, ...)
Complex arithmetic expressions using standard C++ operators
Lazy evaluation
Parallelization of most sparse block matrix operations using OpenMP (if available)
Generic Gauss-Seidel and Projected Gradient solvers for linear problems under normal cone inclusions (such as LCP or Coulomb friction problems; see the documentation)
Several linear iterative solvers ( Conjugate Gradient and variations, GMRES, ...)
Support for matrix-free iterative solvers (using expression templates)
Optional serialization using boost::serialization
Compatibility with standard Block Sparse Row storage

So-bogus

Second Order Cone complementarity function with local root-finding solver
Analytical solver for local Coulomb friction and SOC complementarity problems in 2d and 3d
Global Coulomb friction and SOCQP solvers for arbitrary dimensions

License

bogus is released under the terms of the Mozilla Public License v2.0.

So-bogus is released under the terms of the GNU General Public License version 2 or, at your option, any later version.

For more information, see LICENCE.md.

Documentation

bogus includes documentation in the doxygen format, along with usage samples. Run doxygen from the doc/doxygen folder, or browse it online.

Requirements

Apart from a C++98 compiler, the only requirement is Eigen.

bogus has been successfully tested on a variety of Linux distributions with gcc 4.5+ and clang 3.5+, and on Mac OSX 10.8+ with clang 3.2+. Visual Studio 2012 ( and maybe earlier ) should also be able to compile and run the test suite, though no particular attention is given to this platform.

For more information, see INSTALL.md.

About

So-bogus implements some algorithms initially developed within the BiPop team at Inria Rhônes-Alpes. Its name is derived from the main use case of the library, a Block-oriented Gauss-Seidel solver for Second Order cone problems.