Multiagent Systems and Service-Oriented Computing Laboratory
This page assembles code for some of the experimental systems that
we have developed in the lab. Unless otherwise specified, NCSU owns
this code and retains all rights to it. However, we allow
noncommercial use (i.e., for academic and research purposes) only. In
general, the systems have not packaged where they would run directly
out of the box.
Action Descriptions in C+
- Various action description examples in C+
Protocol Verification
OWL-P (ask me)
Marmara
- The Marmara archive includes
the source code for the simulation environment with an example input
file that specifies a population, a configuration file that sets the
simulation parameters, a build.xml file to compile and run the
code.
- The input file (.STATE) contains the initial state of the agents.
The first line states how many agents the population has and how many
dimensions expertise vectors have. Afterwards, you specify an
expertise vector, an interest vector, and a set of neighbors for each
agent. The configuration file (.conf) sets the thresholds, timeouts,
and so on. There is a detailed description of each parameter in the
appendix of pInar
Yolum's dissertation on Emergent
Properties of Referral Networks.
- The build.xml file will work with Jakarta ant. The simulation uses
JBOSS implementation of Java Messaging Service. You need to install
that separately. The build.xml file states where the JBOSS files
should be installed. After the configuration, typing
ant compile
will compile the entire code and put the Java class files in the
classes directory. Then, you run the code with
ant run
and four parameters. The first one is the .STATE file, the second is
the .conf file, and the third is a directory name where the simulator
will use for logging. The fourth parameter is the random seed and is
optional. If you don't supply a seed, the simulator will use its
own.