A Python package for optimization
Latest commit 63da829 Oct 22, 2013 @dpo dpo Merge pull request #16 from bitdeli-chef/master
Add a Bitdeli Badge to README



Welcome to the NLPy optimization toolkit for Python. If you read documentation, the next section is for you. Otherwise, skip to the Installation section below.


The NLPy documentation is typeset using the Sphinx documentation system `http://sphinx.pocoo.org`_ which is particularly suited to documenting Python packages.

The manual in PDF format is in the doc subfolder. The HTML documentation is accessible by pointing your browser to doc/build/contents.html.

Check the paper NLPy---A Large-Scale Optimization Toolkit in Python by Dominique Orban available from `http://www.gerad.ca`_.

If the user so desires, the documentation may be generated afresh as follows:

  • To re-generate the HTML documentation, change to the doc subfolder and type:

    make html
  • To re-generate the PDF documentation, change to the doc subfolder and type:

    make latex

    Then change to the build/latex subfolder and type:

    make all-pdf


For now, just:

cp site.template.cfg site.cfg
# Edit site.cfg to adjust to your local settings.
python setup.py build
python setup.py install [--prefix=...]

To select another C compiler:

python setup.py config --compiler=<name> build

To select another Fortran compiler:

python setup.py config_fc --fcompiler=<name> build

To see a list of available Fortran compilers and their names:

python setup.py config_fc --help-fcompiler

To select both a C and a Fortran compiler:

python setup.py config --compiler=<name> config_fc --fcompiler=<name> build

For example, you can force compilation with gfortran by specifying:


and wit g95 by specifiying:


(note the subtle name difference.)


  • On an OpenSuSE system I obtain the following error message with both gfortran and g95:

    ld: build/temp.linux-x86_64-2.6/libnlpy_ma27.a(ma27ad.o): relocation
    R_X86_64_32S against 'a local symbol' can not be used when making a shared
    object; recompile with -fPIC build/temp.linux-x86_64-2.6/libnlpy_ma27.a:
    could not read symbols: Bad value

    The source of the problem is that on this platform, distutils appears to be compiling a static library without the -fPIC flags, which generates relocatable objects. When subsequently linking against this library to build a shared object, the linker complains. This problem can be resolved by restarting the installation and adding the following flags to the python setup.py command line:

    --f77flags="-ffixed-form -fno-second-underscore -fPIC -O"
    --f90flags="-fno-second-underscore -fPIC -O"


http://www.gnu.org/graphics/lgplv3-88x31.png Bitdeli badge