Hi Davi,

I don't think ECJ has any native raster graphs output capability. I think your best choice is one of two options:

1) Used ECJ's built in LaTeX output to produce a PDF that then gets converted to PNG;

2) Write custom code that will emit an ECJ tree individual in a format that is compatible with graph/tree drawing packages like Graphviz: http://www.graphviz.org/

That way, you don't have to spend time mucking around with low level graphics drawing stuff that is esentially unrelated to your project.


On Sat, May 4, 2013 at 4:39 AM, Davi Carvalho <[log in to unmask]> wrote:
Congratulations Sean and all ECJ developers!

I am using ECJ on my Master Degree experiments in Genetic Programming.

I would like to know any library that I can use to transform a GP Individual (tree representation) in a PNG file.

Other question is how do I cite ECJ on my work?

Thanks in advance!

Davi Carvalho

On Sat, May 4, 2013 at 1:13 AM, Sean Luke <[log in to unmask]> wrote:
A lot of ECJ version-to-version is bugfixes and cleanups and simplificaiton, and certainly ECJ 21 is a lot of that.  But it also includes (off the top of my head) GE, Push, an entirely new PSO facility, Meta-GAs, improved distributed evaluation, significantly stronger coevolutionary facilities, GP benchmarks, and a bunch of new problems.


On May 3, 2013, at 4:27 PM, Warren Henning wrote:

> What are the biggest changes/new features in ECJ 21?
> Warren
> On Fri, May 3, 2013 at 12:57 PM, Sean Luke <[log in to unmask]> wrote:
> GMU's ECLab, Autonomous Robotics Laboratory, and Center for Social Complexity are proud to announce the official released of ECJ version 21 and MASON version 17.  They are available here:
>         http://cs.gmu.edu/~eclab/projects/ecj/
>         http://cs.gmu.edu/~eclab/projects/mason/
> These released have been a long time in coming: between 1 and 2 years.  They reflect a significant degree of improvement of both the systems, and the assistance of a significant number of people to whom we are thankful.
> ECJ is a high-performance evolutionary computation and stochastic optimization system written in Java.  MASON is a high performance ultralight multiagent simulation toolkit intended for "swarm"-style simulations.
> Sean Luke


Davi Lopes Carvalho
[log in to unmask]