S

spams-devel

Name Last Update
build Loading commit data...
cpp_library Loading commit data...
dags Loading commit data...
data Loading commit data...
decomp Loading commit data...
dictLearn Loading commit data...
doc Loading commit data...
image Loading commit data...
linalg Loading commit data...
prox Loading commit data...
src_release Loading commit data...
swig Loading commit data...
test_release Loading commit data...
wiki Loading commit data...
.gitattributes Loading commit data...
.gitattributes_cpp Loading commit data...
.gitattributes_matlab Loading commit data...
.gitignore Loading commit data...
COPYING Loading commit data...
HOW_TO_INSTALL.txt Loading commit data...
HOW_TO_USE.txt Loading commit data...
README Loading commit data...
TODO Loading commit data...
compile.m Loading commit data...
compile_gcc.m Loading commit data...
compile_intel.m Loading commit data...
compile_mex_linux.m Loading commit data...
compile_mex_macosx.m Loading commit data...
compile_octave.m Loading commit data...
export_cpp_src_git.sh Loading commit data...
export_matlab_src_git.sh Loading commit data...
export_matlab_src_svn.sh Loading commit data...
extract_matlab_src.sh Loading commit data...
get_architecture.m Loading commit data...
make_matlab_linux.sh Loading commit data...
make_matlab_macosx.sh Loading commit data...
start_spams.m Loading commit data...
This is the source code of the software  (you can use the precompiled toolbox
or compile the sources before using it).

There are several possible configurations. In particular,
- it should work with 32 and 64 bits operating systems.
- Linux, Windows (at least for version <= 2.5) and Mac OS.
- it can use either the netlib blas/lapack library, or the intel mkl
  library, or your favorite BLAS library (ACML for instance).
- it can be compiled using either gcc for Linux and Mac (a recent gcc >= 4.5
  is recommended), Microsoft visual studio compiler for Windows and the
  intel compiler for all platforms.
- it can either be used as a C++ library, or interfaced with Matlab, and it
  now comes with a preliminary R/Python interface.


# SPAMS-2.6

SPAMS is now available in version 2.6

## Matlab users

### Precompiled toolbox
You can use the precompiled toolbox `spams-matlab-precompiled-v2.6-%LAST_DATE%-%OS%.tar.gz`
for Linux or MacOS. You extract the tarball and run Matlab in the current
directory. You have to use the command `start_spams` in Matlab to load the
precompiled functions. [HOW_TO_USE.txt](./HOW_TO_USE.txt) for details.

The precompiled version for MacOS was compiled on MacOS X Maverick (10.9.5),
please let us know if you encounter any issues on more recent MacOS version.

At the moment, no precompiled version is available for Windows users, we are
currently working on it.

### DISCLAIMER

In the MacOS precompiled version, the multi-threading with OpenMP is not
available (not supported by the compiler clang at the moment).

To enable multi-threading, you will have to compile the library (c.f. below)
with a different compiler (e.g. gcc, intel or clang-omp).

### Building the library (more advanced users)

The best way to compile the library is to open the file [compile.m](./compile.m)
and follow instructions. You can modify the beginning of the file to choose
the compiler, the blas library, set some options and the different paths, then
run `compile.m` in Matlab.

After all the mex-files are compiled, a script run_matlab.sh is also created,
at least for the Linux and/or Mac OS version. Depending on your configuration,
it might be necessary to launch Matlab with the script in order to use the
toolbox (in order to preload multi-threading libraries).
Otherwise, just type "start_spams;" and use the SPAMS library.


## R/Python

For R/Python users, an interface with some instructions is available in the
folder [swig](./swig)