-
Download the EPOS source code from the latest release 0.0.4.
-
Make sure that a version of Oracle Java 8 is installed. You can download it from here.
-
Downloaded files should include:
datasets/
,conf/
, andsrc/
.
-
Choose dataset in
datasets/
. Default is gaussian with 10 agents' plans file and the goal-signal target. Each plan has the same size, consisting of plan cost (before ":") and plan values (after ":"). The goal-signal has the same dimensions as the plan dimensions. You can download other dataset from here. -
Config parameters in
conf/epos.properties
based on the dataset you choose, including the number of agents, plans, iterations, simulations, children and the plan dimension. Fixed configs constainconf/log4j.properties
,conf/measurement.conf
, andconf/protopeer.conf
. -
Run from command line. Navigate to the project directory and execute:
cd EPOS
mvn clean install –U #or where the project pom.xml is
Or, navigate the Release-0.0.4.zip, and run from command line:
java -jar IEPOS-Tutorial.jar
Once it successfully finishes, the project can be opened via your IDE of choice as a Maven project.
After running, results are shown in output/
directory. The output constains:
- Log output from the loggers in ".csv" format.
- A copy of the applied configuration for sanity checks named "used_conf.txt"
- Visualization of the tree & the plan selections
For more information of EPOS, you can read the EPOS-Documentation from here.
Pournaras, Evangelos, Peter Pilgerstorfer, and Thomas Asikis. "Decentralized collective learning for self-managed sharing economies." ACM Transactions on Autonomous and Adaptive Systems (TAAS) 13.2 (2018): 1-33.