ECML - Evolutionary Computation and Machine Learning Group at RMIT

ECML	RMIT Genetic Programming Package (RMIT GP)
About ECML Research Projects Teaching Involved Selected Publications Associated Events Contact Information A/P Vic Ciesielski School of Computer Science and Information Technology RMIT University VIC 3001, Melbourne, Australia Tel: +613 99252926 Fax:+613 96621617	What is RMIT-GP? Current versions How to run RMIT GP in RMIT? Basic and Advanced graphing tools Support and FAQ What is RMIT GP? RMIT-GP is a genetic programming package written in C++. It is intended for serious genetic programming research. It comes with a full implementation of all of the components necessary to implement a GP solution to a problem. It is designed to permit easy experimentation with the various components of a GP system. It was originally designed and built by Vic Ciesielski and Peter Wilson for a robocup application. It was subsequently totally re-engineered by Vic Ciesielski and Dylan Mawhinney. This was partially funded by the Victorian Partnership for Advanced Computing (VPAC). Further modifications were made by Emil, Michael, Xiang Li , Andrew Innes and Teja Nanduri. Some of the history can be found in the release notes. Current Version There are currently two versions which have slightly different features and bug fixes. Work is currently underway to integrate the two versions. README rmitgp.v1.5 (compatible with different GCC, optional optimisation patch 1, 2) rmitgp.v1.4 (including decimation and ADF) rmitgp.v1.32 (basic and simple log file graphing) How to run RMIT GP at RMIT? On a Solaris system RMIT-GP needs version gcc-3.3.2 or lower of gcc. This can be downloaded from the aarnet site. RMIT CS machines already have this version installed. The methods for setting up the environment and running the example programs can be viewed here. The distribution runs without the need to modify any files on the Linux distributions we have used (Redhat, Ubuntu). Basic and advanced RMIT GP run-log graphing tools A run-log can easily be converted into a simple fitness graph by scripts and some scripts are included in the 'graph' sub-directory in version 1.32 and version 1.4 Some comprehensive scripts for analysing runs have been developed by Michael Foster and can be downloaded from here. Some limited documentation can be viewed at Michael's web page. Support and FAQ RMIT students interested in Evolutionary Computing can find a list of possible projects here. Accounts on the ECML group's Linux cluster or the VPAC super computer facility.

ECML

RMIT Genetic Programming Package (RMIT GP)

A/P Vic Ciesielski
School of Computer Science and Information Technology
RMIT University
VIC 3001, Melbourne, Australia
Tel: +613 99252926
Fax:+613 96621617

What is RMIT-GP?
Current versions
How to run RMIT GP in RMIT?
Basic and Advanced graphing tools
Support and FAQ

What is RMIT GP?

Current Version

There are currently two versions which have slightly different features and bug fixes. Work is currently underway to integrate the two versions.

README
rmitgp.v1.5 (compatible with different GCC, optional optimisation patch 1, 2)
rmitgp.v1.4 (including decimation and ADF)
rmitgp.v1.32 (basic and simple log file graphing)

How to run RMIT GP at RMIT?

On a Solaris system RMIT-GP needs version gcc-3.3.2 or lower of gcc. This can be downloaded from the aarnet site.
RMIT CS machines already have this version installed. The methods for setting up the environment and running the example programs can be viewed here.
The distribution runs without the need to modify any files on the Linux distributions we have used (Redhat, Ubuntu).

Basic and advanced RMIT GP run-log graphing tools

A run-log can easily be converted into a simple fitness graph by scripts and some scripts are included in the 'graph' sub-directory in version 1.32 and version 1.4
Some comprehensive scripts for analysing runs have been developed by Michael Foster and can be downloaded from here. Some limited documentation can be viewed at Michael's web page.

Support and FAQ

RMIT students interested in Evolutionary Computing can find a list of possible projects here. Accounts on the ECML group's Linux cluster or the VPAC super computer facility.