VIATRA2/GRaTS: Stochastic Graph Transformation System

Tags: VIATRA2, stochastic-simulation, graph-transformation, GRaTS, semi-Markov-processes

Abstract

VIATRA2/GRaTS is a stochastic graph transformation system simulator developed in collaboration between the VIATRA2 Group at Budapest University of Technology and Economics and Dr. Paolo Torrini from Prof. Reiko Heckel's group at the University of Leicester. The system implements a framework for running discrete-event stochastic simulations of semi-Markov processes based on generalized stochastic graph transformation, enabling structured simulation experiments with statistical analysis capabilities.

Main Publication: Stochastic Simulation of Graph Transformation Systems (FASE 2010)
Authors: Paolo Torrini, Reiko Heckel, István Ráth

Overview

VIATRA2/GRaTS is a stochastic graph transformation system simulator built in co-operation between the VIATRA2 Group and Dr. Paolo Torrini from Prof. Reiko Heckel's group at the University of Leicester.

GRaTS implements a framework to run discrete-event stochastic simulations of semi-Markov processes based on generalised stochastic graph transformation. Graph transformation rules are used to define observable events that lead from one state to another (action rules), as well as quantitative observations over states (probe rules). GRaTS enables the execution of structured simulation experiments made of multiple batches of bounded runs, allowing for variations in sensitive parameters that may include maximum length. It can compute statistics about action rule application, timing, probe rule values and covariance matrix, based on different forms of aggregation of the observations throughout an experiment.

Technical Implementation

GRaTS is implemented in Java as a functionality of VIATRA, a rule-based language interpreter based on abstract state machines and incremental pattern matching, available as an Eclipse plugin. GRaTS relies on the implementation of graph transformation given by VIATRA2, taking advantage of incremental pattern matching to deal efficiently with semi-Markov processes.

Key Features

• Semi-Markov Process Simulation: Supports discrete-event stochastic simulations based on generalized stochastic graph transformation
• Statistical Analysis: Computes comprehensive statistics including action rule applications, timing, probe rule values, and covariance matrices
• Structured Experiments: Enables multiple batches of bounded runs with parameter variations
• Visualization Support: Individual run visualization using Eclipse Zest graph visualization
• External Libraries: Integrates SSJ libraries for random number generation, distribution functions, and confidence intervals

Obtaining GRaTS

Currently, GRaTS is available as a set of source projects that need to be checked out from SVN and run as an "Eclipse Application" configuration from your workspace.

Requirements

• Java 1.6.x
• Eclipse 3.6
• Recent version of VIATRA2

Source Code

📁 Code Archive Notice: The stochastic graph transformation tools are archived at github.com/ftsrg/viatra-svn-archive. This is research prototype code for historical reference.

Check out the main system from:

• https://github.com/ftsrg/viatra-svn-archive/tree/main/stochsim/

Simulation project samples can be obtained from:

• https://github.com/ftsrg/viatra-svn-archive/tree/main/transformations/collab/trunk/stochastic_gt

Documentation

The most current documentation is available as a PDF attachment: GRaTS Documentation

Related Publications

Publications at Budapest University of Technology and Economics

1. Stochastic simulation of graph transformation systems (FASE 2010)
   Paolo Torrini, Reiko Heckel, István Ráth

2. Stochastic graph transformation with regions (GT-VMT 2010)
   Paolo Torrini, Reiko Heckel, István Ráth, Gábor Bergmann

3. Advances in Model Transformations by Graph Transformation: Specification, Execution and Analysis (SENSORIA Book 2010)
   Gábor Bergmann, Artur Boronat, Paolo Torrini, Reiko Heckel, István Ráth, Dániel Varró

4. Incremental pattern matching for the efficient computation of transitive closure (ICGT 2012)
   Gábor Bergmann, István Ráth, Tamás Szabó, Paolo Torrini, Dániel Varró
   See also the online addendum

Publications at University of Leicester

1. Model-based Simulation of VoIP Network Reconfigurations using Graph Transformation Systems (ICGT 2008)
   Ajab Khan, Paolo Torrini, Reiko Heckel

2. Model-based stochastic simulation of P2P VoIP using graph transformation system (ASMTA 2010)
   Ajab Khan, Reiko Heckel, Paolo Torrini, István Ráth
   Download PDF

3. Model-based stochastic simulation of Super Peer Promotion in P2P VoIP using graph transformation (DCNET 2011)
   Ajab Khan, Reiko Heckel

Research Applications

GRaTS has been successfully applied to various domains including:

• VoIP Network Simulation: Modeling network reconfigurations and P2P VoIP systems
• Peer-to-Peer Networks: Simulation of super peer promotion mechanisms
• Service-Oriented Systems: Analysis of complex distributed systems
• Performance Evaluation: Statistical analysis of system behavior under different parameters

Technical Architecture

The system combines several key technologies:

• VIATRA2 Framework: Provides the core graph transformation engine with incremental pattern matching
• Eclipse Platform: Serves as the development and runtime environment
• SSJ Statistical Libraries: Handles random number generation and statistical computations
• Zest Visualization: Enables graphical representation of transformation sequences

This integration allows GRaTS to efficiently handle large-scale stochastic simulations while providing comprehensive analysis capabilities for complex systems modeled as graph transformation systems.