The Model Checking Contest (MCC) is a yearly event that assesses existing verification tools for concurrent systems on a set of models (i.e., benchmarks) proposed by the scientific community. All tools are compared on the same benchmarks and using the same computing platform, so that a fair comparison can be made, contrary to most scientific publications, in which different benchmarks are executed on different platforms.
Another goal of the Model Checking Contest is to infer conclusions about the respective efficiency of verification techniques for Petri nets (decision diagrams, partial orders, symmetries, etc.) depending on the particular characteristics of models under analysis. Through the feedback on tools efficiency, we aim at identifying which techniques can best tackle a given class of models.
Finally, the Model Checking Contest seeks to be a friendly place where developers meet, exchange, and collaborate to enhance their verification tools.
The Model Checking Contest is organized in three steps:
- the Call for Models,
- the present Call for Tools,
- and the Contest itself.
Call for Tools
For the 2017 edition, we kindly ask the developers of verification tools for concurrent systems to participate in the MCC competition. Each tool will be assessed on both the accumulated collection of MCC models (these are the "known" models, see http://mcc.lip6.fr/models.php) and on the new models selected during the 2017 edition (these are the "surprise" models, see http://mcc.lip6.fr/cfm.php).
The benchmarks on which tools will be assessed, are colored Petri nets and/or P/T nets. Some P/T nets are provided with additional information giving a hierarchical decomposition into sequential machines (these models are called Nested-Units Petri nets - see http://mcc.lip6.fr/nupn.php for more information): tools may wish to exploit this information to increase performance and scalability.
Each tool may compete in one or more categories of problems, such as reachability analysis, evaluation of CTL formulas, of LTL formulas, etc.
Tools have to be submitted in binary-code form. Each submitted tool will be run by the MCC organizers in a virtual machine (typically configured with 4 cores, 4 Gbytes RAM per core, and a time confinement of 60 minutes per run, i.e., per instance of a model). Last year, more than 1500 days of CPU time have been invested in the MCC competition. The MCC relies on BenchKit (https://github.com/fkordon/BenchKit), a dedicated execution environment for monitoring the execution of processes and gathering of data.
By submitting a tool, you explicitly allow the organizers of the Model Checking Contest to publish this tool in binary form on the MCC web site, so that experiments can be reproduced by others after the contest. Detailed information is available from http://mcc.lip6.fr/rules.php.
Note: to submit a tool, it is not required to have submitted any model to the MCC Call for Models.
IMPORTANT : it is strongly recommended to pre-register your tool using the dedicated form before Feb. 1, 2017. You will then be informed of the way the contest is going.
- Dec. 19, 2016: publication of the present Call for Tools
- Jan. 15, 2017: publication of the Tool Submission Kit, which will be made available from http://mcc.lip6.fr/archives/ToolSubmissionKit.tar.gz
- Jan. 15, 2017: publication of the updated 2017 contest rules at http://mcc.lip6.fr/rules.php
- Feb. 1st, 2017: deadline for tool pre-registration If you plan to submit a tool to the contest, please fill in the pre-registration form (you may retire if you finally decide not to do so)
- Apr. 15, 2017: deadline for tool submission
- Apr. 30, 2017: early feedback to tool submitters, following the preliminary qualification runs, which are performed using a few small instances of the "known" models
- June 1st, 2017: more feedback to tool submitters, following the competition runs
- June 27, 2017: official announcement of MCC'2017 results during the Petri Net conference (Zaragoza, Spain)
- Didier Buchs - Univ. Geneva, Switzerland
- Fabrice Kordon - UPMC, France
Execution Monitoring Board
- Francis Hulin-Hubard - CNRS and ENS de Cachan, France
- Fabrice Kordon - UPMC, France
- Marco Beccuti - Univ. Torino, Italy
- Monika Heiner - Univ. Cottbus, Germany
- Jeroen Meijer - Univ. Twente, Netherlands
- Franck Pommereau - Univ. Evry, France
- Christian Rohr - Univ. Cottbus, Germany
- Jiri Srba - Univ. Aalborg, Denmark