- Short Description:
This PYADOLC, a Python module to differentiate complex algorithms written in Python. It wraps the functionality of the library ADOL-C (C++).
- Author:
Sebastian F. Walter
- Licence (new BSD):
Copyright (c) 2008, Sebastian F. Walter All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: * Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. * Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. * Neither the name of the HU Berlin nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY Sebastian F. Walter ''AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL Sebastian F. Walter BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
EXAMPLE USAGE:
import numpy
from adolc import *
N = M = 10
A = numpy.zeros((M,N))
A[:] = [[ 1./N +(n==m) for n in range(N)] for m in range(M)]
def f(x):
return numpy.dot(A,x)
# tape a function evaluation
ax = numpy.array([adouble(0) for n in range(N)])
trace_on(1)
independent(ax)
ay = f(ax)
dependent(ay)
trace_off()
x = numpy.array([n+1 for n in range(N)])
# compute jacobian of f at x
J = jacobian(1,x)
# compute gradient of f at x
if M==1:
g = gradient(1,x)
- REQUIREMENTS:
- Python and Numpy, both with header files
- ADOL-C http://www.coin-or.org/projects/ADOL-C.xml
- boost::python from http://www.boost.org/
- scons build tool
- OPTIONAL REQUIREMENTS:
- Colpack from http://www.cscapes.org/coloringpage/software.htm or alternatively http://github.com/b45ch1/colpack . Colpack is needed for sparse Jacobians and sparse Hessians
INSTALLATION (my recommendation): At the moment, the installation using setup.py is lagging behind. Below, the way I would install PYADOLC
. If you do it another way, send me an email or fork this project and I'll incorporate it. Alternatively you can _try to use distutils, but support for it lags behind scons: i.e. 1) rename setup.py.EXAMPLE
to setup.py
to fit your system, then 2) run python setup.py build_ext --inplace
.
- Install ADOL-C, to use PYADOLC with sparse support, you do _not have to do ./configure --with-sparse. A simple ./configure does the job
- (OPTIONAL) Install ColPack
cd $HOME
git clone git://github.com/b45ch1/pyadolc.git
cd pyadolc
- Rename
SConstruct.EXAMPLE
toSConstruct
and modify it to fit your system- Run
scons
, this will create the shared libraries_adolc.so
and_sparse.so
. If you don't have colpack installed calling scons will only compile_adolc.so
but fail to compile_sparse.so
.- Using
scons
is more convenient to work with when you often pull new versions.- Add the directory to your
PYTHONPATH
. E.g. add the following line in your~/.bashrc
file:export PYTHONPATH=$PYTHONPATH:/home/walter/workspace/pyadolc