anl

======== Overview

ANAC Automated Negotiations League Platform

Overview

This repository is the official platform for running ANAC Automated Negotiation Leagues (starting 2024). It will contain a package called anlXXXX for the competition run in year XXXX. For example anl2024 will contain all files related to the 2024's version of the competition.

Installation

::

pip install anl

You can also install the in-development version with::

pip install https://github.com/autoneg/anl/archive/master.zip

Documentation

https://yasserfarouk.github.io/anl/

Changelog

0.1.10 (2024.04.03)

• fixing issue #1 (anlv failure on macOS)
• Correcting git installation method in docs
• requiring negmas 0.10.21
• Better comments on micro
• Defaulting to 3min hidden time limit per negotiation
• pypi workflow update
• Fix tutorial links

0.1.9 (2024.02.14)

• Adding divide-the-pies scenarios
• Adding workflow to test on negmas master
• Tutorial and docs update
• Update faq

0.1.8 (2023.12.31)

• bugfix in visualizer initial tournament list
• Correcting auto pushing to PyPi

0.1.7 (2023.12.31)

• Adding simple dockerfile
• Adding --port, --address to anlv show. You can now set the port and address of the visualizer
• Visualizer parses folders recursively
• minor: faster saving of figs
• Adding mkdocs to dev requirements
• Removing NaiveTitForTat from the default set of competitors
• Improving tutorial

0.1.6 (2023.12.27)

• Improved visualizer

  • Adding filtering by scenario or strategy to the main view.
  • Adding new options to show scenario statistics, scenario x strategy statistics, and cases with no agreements at all.
  • You can show multiple negotiations together
  • You can show the descriptive statistics of any metric according to strategy or scenario
  • More plotting options for metrics

• Improved CLI

  • Adding the ability to pass parameters to competitors in the CLI.
  • Removing NaiveTitForTat from the default set of competitors
  • Making small tournaments even smaller

• New and improved strategies

  • Adding RVFitter strategy which showcases simple implementation of curve fitting for reserved value estimation and using logging.
  • Adding more comments to NashSeeker strategy
  • Simplified implementation of MiCRO
  • Adding a simple test for MiCRO
  • Avoid failure when Nash cannot be found in NashSeeker

• Migrating to NegMAS 0.10.11. Needed for logging (and 0.10.10 is needed for self.oppponent_ufun)

0.1.5 (2023.12.24)

• Changing default order of agents
• Adding a basic visualizer
• Adding make-scenarios to the CLI
• Passing opponent ufun in the private info
• Separating implementation of builtin agents
• requiring NegMAS 0.10.9

0.1.4 (2023.12.24)

• Retrying scenario generation if it failed
• Defaulting to no plotting in windows

0.1.3 (2023.12.23)

• Defaulting to no-plotting on windows to avoid an error caused by tkinter
• Retry scenario generation on failure. This is useful for piece-wise linear which will fail (by design) if n_pareto happened to be less than n_segments + 1

0.1.2 (2023.12.18)

• Adding better scenario generation and supporting mixtures of zero-sum, monotonic and general scenarios.
• Requiring negmas 0.10.8

0.1.2 (2023.12.11)

• Controlling log path in anl2024_tournament() through the added base_path argument

0.1.1 (2023.12.09)

• Added anl cli for running tournaments.
• Added the ability to hide or show type names during negotiations
• Corrected a bug in importing unique_name
• Now requires negmas 0.10.6

0.1.0 (2023.11.30)

• Adding ANL 2024 placeholder