anl-agents 0.1.1

Add your description here

This repository contains agents (negotiators) submitted to the ANL league of the ANAC Competition

To install this package just run:

pip install anl-agents

There are two ways to submit agents to this repository:

1. Participate in the ANAC competition https://anac.cs.brown.edu/anl
2. Submit a pull-request with your agent added to the contrib directory.

Getting lists of agents

You can get any specific subset of the agents in the library using get_agents(). This function has the following parameters:

• version: Either a competition year (2024, ...) or the value "contrib" for all other agents. You can also pass "all" or "any" to get all agents.
• track: The track (advantage, utility, welfare, nash, kalai, kalai-smorodinsky)
• qualified_only: If true, only agents that were submitted to ANL and ran in the qualifications round will be returned.
• finalists_only: If true, only agents that were submitted to ANL and passed qualifications will be returned.
• winners_only: If true, only winners of ANL (the given version) will be returned.
• top_only: Either a fraction of finalists or the top n finalists with highest scores in the finals of ANL.
• as_class: If true, the agent classes will be returned otherwise their full class names.

For example, to get the top 10% of the "advantage" track finalists in year 2024 as strings, you can use:

get_agents(version=2024, track="advantage", finalists_only=True, top_only=0.1, as_class=False)

Winners of the ANL 2024 Competition

Advantage Track

• First Place: Shochan
• Second Place: UOAgent
• Third Place: AgentRenting2024

You can get these agents after installing anl-agents by running:

get_agents(2024, track="advantage", winners_only=True)

Nash Track

• First Place: Shochan

You can get this agent after installing anl-agents by running:

get_agents(2024, track="nash", winners_only=True)

Installation Note

If you are on Apple M1, you will need to install tensorflow before installing this package on conda using the method described here