Evolving Objects / News: Recent posts

This 1.3.1 release bring an easier to use parallelization module:
- the eo::mpi modules is no longer dependent from boost::mpi
- parallel multi-start example
- bugfix: an error is now thrown when accessing best_element of an empty population

The new 1.3.0 version of the "Evolving Objects" framework is available.

EO is a template-based, C++ evolutionary computation library which
helps you to write your own stochastic optimization algorithms
insanely fast.

Learn more about EO on the official website:
http://eodev.sourceforge.net/

You will find the release 1.3.0 at the following address:
https://sourceforge.net/projects/eodev/files/eo/1.3.0/... read more

A new version of the "Evolving Objects" framework is available.

EO is a template-based, C++ evolutionary computation library which
helps you to write your own stochastic optimization algorithms
insanely fast.

Learn more about EO on the official website:
http://eodev.sourceforge.net/

You will find the release 1.2.0 at the following address:
https://sourceforge.net/projects/eodev/files/eo/1.2.0/

Here is a summary of the change log:
- new class eoParallel enabling parallelization in EO using OpenMP. At this
time, it only concerns the evaluation operator, for advanced details go to
the file test/t-eoParallel.cpp. The default parallelization mode is off, to
switch it on, see the parameters prefixed by --parallelize-*
- pyeo (binding for using EO within python) compatible with the last version of Boost and Python libraries
- updated instructions on how to build and install (at last!)
- fixed the incremental allocation issue in variation operators which were
taking too much time for big population sizes
- package generators for easy installation
- script installing EO manually, using symbolic links
- evaluators that throw an exception if a maximum time has been reached
(wallclock and CPU user time for POSIX systems), independently of the number of generations
- an advanced logging system available out-of-the-box
- some bugfixes for windows systems
- and more…... read more

The stable release of the version 1.1.0 is out!

Since the beta(s) a portability bug has been fixed in the logger, and some compilation problems have been removed.

Enjoy.

EO version 1.1.0 have been released!

Summary of the change log:
- provide cmake build system, remove the old autotools one
- package generation system
- GCC 4.3 compatibility
- new versatile log system with several nested verbose levels
- classes using intern verbose parameters marked as deprecated, please update your code accordingly if you use one of the following files:
eo/src/eoCombinedInit.h
eo/src/eoGenContinue.h
eo/src/eoProportionalCombinedOp.h
eo/src/utils/eoData.h
eo/src/utils/eoStdoutMonitor.h
- an evaluator that throw an exception if a maximum eval numbers has been reached, independently of the number of generations
- new monitor that can write on any ostream
- new continuator that can catch POSIX system user signals
- dual fitness class to handle feasibility of individual with guarantee that feasible fitness will always be better than unfeasible one
- feasible fitness ratio stat
- interquartile range stat
- average size of individuals stat
- uniform(min,max) random function
- compatibility macros for compiling paradiseo with CUDACC
- removed old multi-objective classes, deprecated by the Paradiseo-MOEO project
- new website, switch from CVS to GIT, and a cool logo!

Due to spam and configurations problems, the project has switch its tracker system to the Trac hosted app.

To submit bug reports or feature requests, go to the Trac page: https://sourceforge.net/apps/trac/eodev/

We have lost all the mailing-list subscriptions in a crash some weeks ago, if you want to stay tuned, please subscribe again on the eodev-main subscription page: https://lists.sourceforge.net/lists/listinfo/eodev-main

We have migrated the old version control system (CVS) to a more modern one (GIT). Now, you can commit code offline and work easily on separated branches.

Additionally, we also have a new web site, which display more informations about the current status of EO, with a new logo!

The long awaited version 1.0 of Evolving Objects (EO) is released!

The second beta-version of eo-1.0 is available at SourceForge now.

A first beta-version of eo-1.0 is available at SourceForge now.

We have released a new version of Evolving Objects.

EO is a templates-based, ANSI-C++ compliant evolutionary computation library. It contains classes for almost any kind of evolutionary computation you might come up to - at least for the ones we could think of. It is component-based, so that if you don't find the class you need in it, it is very easy to subclass existing abstract or concrete classes. See http://eodev.sourceforge.net for details.... read more

TIred of receiving messages of people trying to compile 0.9.2 with gcc 3.x.

But no major change in the code, though.

PyEO, a library that exports the C++ classes to python classes has been uploaded to the CVS tree.

The C++-class library is fully integrated into python, which means that you can subclass EO (abstract) classes in python, and then add them to C++ based algorithms.

We still have to work out a couple of issues with building it as it depends on python2.2 and something called boost.python, but it seems that we can add a language to the hopefully long list of EO-enabled computer languages

The first version of paradisEO, the parallel version of EO, has been uploaded to the CVS repository, thanks to Sebastien Cahon.
It allows to use your existing eoAlgo's algorithms within the Island- or the Master-Slave models with minimal modifications.

You need to install MPICH, a distribution of MPI that you can find at http://www.mcs.anl.gov/mpi/mpich/download.html

All feedback are welcome.l

To stop downloads of the comletely obsolete 0.9.1, and even if 1.0 comes soon, here is release 0.9.2, the current state - almost atble :-)

CVS version is now stable, and it will very soon called 0.9.2. Hold on there!