% FlexBox
%
% Author Information:
% Hendrik Dirks
% Institute for Computational and Applied Mathematics
% University of Muenster, Germany
%
% Contact: hendrik.dirks@wwu.de
%
%
% Version 1.03
% Date: 2016-07-14
% FlexBox is copyright ©2016 by Hendrik Dirks and is distributed under the terms of the GNU General Public License (GPL) version 3 (or later).
%
% If you plan to distribute the software (commercially or not). Please contact me for more information.
%%%%%%%%%%%%%%%%
%%% Citation %%%
%%%%%%%%%%%%%%%%
If you use this toolbox please use the following citation
@Article{dirks2015flexbox,
Title = {A Flexible Primal-Dual Toolbox},
Author = {Dirks, Hendrik},
Journal = {ArXiv e-prints},
Year = {2016},
Month = mar,
Keywords = {Mathematics - Optimization and Control, Computer Science - Computer Vision and Pattern Recognition, Computer Science - Mathematical Software, I.4, G.1.6, G.4},
Primaryclass = {math.OC}
}
% A preprint of the article can be found at http://arxiv.org/abs/1603.05835
% Examples can be found in the 'examples' folder
% In case you experience any problems, please create an issue at https://github.com/HendrikMuenster/flexBox/issues
%%%%%%%%%%%%%%%%%%%
%%% Compilation %%%
%%%%%%%%%%%%%%%%%%%
C++:
To compile the C++ module, simply run the MATLAB script "compileMexModule.m" in the folder "CPP_module"
CUDA:
To compile the Cuda module, open the "compileCudaModule.m" and change the given PATH variables according to your computer. Afterwards, run "compileCudaModule.m"
Issues for University of Münster users:
There are some known problems for CUDA computers at the University of Münster. You may have to preload libraries. To do so, simply start Matlab with:
env LD_PRELOAD="/usr/lib/x86_64-linux-gnu/libstdc++.so.6 /usr/local/cuda-7.5/lib64/libcudart.so.7.5 /usr/local/cuda-7.5/lib64/libcusparse.so.7.5" matlab
%%%%%%%%%%%%%%%%%
%%% Changelog %%%
%%%%%%%%%%%%%%%%%
14.07.2016 (Version 1.03)
- CUDA module added
07.04.2016 (Version 1.02)
- Added FFT-based convolution operator, downsampling and concatenation operator
- Added examples for deblurring and operator concatenation
- Speed improvements for C++ version
31.03.2016 (Version 1.01)
- generalized the concept of matrices to linear operators and created classes for gradient, diagonal, identity and zero operator for MATLAB and C++ version to improve speed and flexibility