Skip to content

Releases: marketdesignresearch/mechlib

Release v0.2.1

13 Dec 10:25
ef2b4c7
Compare
Choose a tag to compare

Fix

  • Bumped dependencies (i.e., Log4j), due to latest vulnerability findings

Release v0.2.0

23 Sep 14:44
Compare
Choose a tag to compare

Improvements

  • Implementation of the Machine Learning-powered Iterative Combinatorial Auctions with Interval Bidding (iMLCA, https://doi.org/10.1145/3465456.3467535)
  • Added a generic ValueFunction (i.e a function that can be asked for for the value for a certain bundle) which allows easy transformation and replacement of any bidders’ ValueFunction.
  • Cleanup of interfaces

Note: This release touches much of the existing code base and is not backward-compatible.

Release v0.1.0

12 Sep 09:41
34d01ed
Compare
Choose a tag to compare

Mechanism Library

The purpose of this library is to collect common domain and market mechanism implementations in the most general way possible.
For example, you can create a set of bidders with XOR valuation on a set of goods (which already forms a simple standard-domain),
and then do various things with this domain, e.g.:

  • Run Winner Determination Problems (WDPs) to find the optimal allocation
  • Collect sets of bids from the bidders according to a certain strategy (e.g., truthful)
    • Based on this bids, find the resulting allocation with bid-based WDPs, or try out different payment rules (e.g., VCG, or variations of CCG)
  • Ask bidders for their value for a certain bundle
  • Ask bidders for the optimal bundle given certain prices
  • Run instances of complete auction formats (incl. multi-round auctions like the Combinatorial Clock Auction) in this domain

Most importantly, the library should serve as a dependency in your project.
Thus, you will most likely create your own classes that implement the corresponding interfaces in the core/ package.
These interfaces are well-documented and serve as a good starting point to use this library.
It's also worth checking the tests to see how some of the above functionality is used.

This library was developed at the University of Zurich in collaboration with Boston University.