Releases: epournaras/EPOS
Releases · epournaras/EPOS
I-EPOS 0.0.4
Agent behaviours:
- Each agent can have its own behaviours (unfairness and local cost weights). The different behaviours are required within the dataset file as "behaviours.csv".
- Adding an output file "agents-position.csv" that presents the agent's position in sequence order in each simulation.
Hard constraints:
- Set two types of hard constraints: one is HARD_PLANS that constrains the global response, the other is HARD_COSTS that constrains the cost (local, global and global complex).
- Adding an output file "hard-constraint-violation.csv" that presents whether the response or cost violates the hard constraint (1 is violate, 0 is not).
I-EPOS 0.0.3
Shuffling Amend in selected-plans
-
Amend the order of the agents in the output file "selected-plans.csv" since agents sequence is shuffled after the first simulation.
-
Generate a matrix to store all agents sequence mappings, and output the right agent index in the logging.
I-EPOS 0.0.2
- updated epos.properties: new format for arguments for complex cost function (see agent.planselection.PlanSelectionOptimizationFunctionCollection:98). This new format support complex cost functions with more than two arguments. See epos.propoerties for the information on the new format.
- updated Configuration.java: now supporting an array of arguments for complex cost functions.
- added tree visualizer: added tree visualizer which shows the local cost of the agents in the tree at the final iteration of each simulation (darker colours show higher local cost)
Epos jar release for experiments.
Possibly final release contains the new maven scheme as well as a much more improved configuration scheme. The executable .jar file can be found in the release-0.0.1.zip. The underlying directory tree should remain as-is. A toy gaussian dataset is found for experimentation under datasets/gaussian/. More information regarding the usage and execution can be found in the project documentation!