Salomon Sickert

Return to Index

MoChiBA

A prototype using ltl2ldba (Owl) for quantitative model checking of MDPs (Markov decision process). Based on the PRISM model checker.

Download

The development repositories are not public at the moment. But a recent snapshot of the source code and a pre-built version (with startup script) can be obtained here:

System Requirements

MoChiBA only requires an installed Java 8 Runtime Environment.

Usage

The tool cannot be configured - all optimisations are enabled - and does not need to be installed.

It reads a model (given as an MDP, .nm) and a property specification (.pctl) from the file system and prints the results to stdout.

java -jar mochiba.jar model.nm properties.pctl

Benchmarks

We performed benchmarks on two case studies and publish here the models, specifications and the raw log output:

Since in most default configuration the Java JRE is too conservative with the heap and stack size, we recommend using the included sh script to start the tool, which fine tunes the JRE for MoChiBA.

./mochiba.sh model.nm properties.pctl

Naming

MoChiBA is short for Quantitative Probabilistic LTL Model Checking using Limit-Deterministic Büchi Automata.

It is a well-known fact that naming things is a hard problem. After spending some time searching for a good name we only came up with candidates too close1 to already used names (e.g. ProB, PRISM) or unpronounceable ones such as QuPrLimBü or ProLimDeBüchi. So after some toying around we discovered one could form the word mochi (a delicious Japanese rice cake) from Model Checking and adding BA for Büchi automaton yielded something pronounceable. We also noticed that it contains Chiba, the name of the city where ATVA 2016 will take place (the whole word has also another meaning in Japanese: Translations) and at this point we didn’t want to further tempt fate for a project that is submitted to ATVA 2016. :)

1: Edit distance less than 2.

Publications