Analysis of Large Scale Interacting Systems by Mean Field Method

  title={Analysis of Large Scale Interacting Systems by Mean Field Method},
  author={Andrea Bobbio and Marco Gribaudo and Mikl{\'o}s Telek},
  journal={2008 Fifth International Conference on Quantitative Evaluation of Systems},
  • A. Bobbio, M. Gribaudo, M. Telek
  • Published 14 September 2008
  • Mathematics
  • 2008 Fifth International Conference on Quantitative Evaluation of Systems
Modeling and analysing very large stochastic systems composed of interacting entities is a very challenging and complex task. The usual approach, relying on the generation of the whole state space, is bounded by the state space explosion, even if symmetry properties, often included in the model, allow to apply lumping techniques and building the overall model by means of tensor algebra operations. In this paper we resort to the mean field theory. The main idea of the mean field theory is to… 
Fluid Aggregations for Markovian Process Algebra
Quantitative analysis by means of discrete-state stochastic processes is hindered by the well-known phenomenon of state-space explosion, whereby the size of the state space may have an exponential
Applying Mean-Field Approximation to Continuous Time Markov Chains
This tutorial considers illustrating examples to discuss how the mean-field method is used in different application areas, starting from the application of the classical technique and moving to cases where additional steps have to be used, such as systems with local communication.
Optimization and Control of Large Systems: Fighting the Curse of Dimensionality
The goal of this thesis is to provide methods for the control and the optimization of large-scale systems, starting from stochastic models that approximate its behavior, and shows that solving an optimal control problem for a system with a large number of objects can be reduced to the solving of a problems for a deterministic system as the number of object grows large.
Stochastic Process Algebras: From Individuals to Populations
It is shown how population-based models that make use of a continuous approximation of the discrete behaviour can be used to efficiently analyse the temporal behaviour of very large systems via their collective dynamics.
Fluid computation of passage-time distributions in large Markov models
Markovian Agents : A New Quantitative Analytical Framework for Large-Scale Distributed Interacting Systems
A Markovian Agent Model (MAM) is a stochastic model that provides a flexible, powerful and scalable way for analyzing complex systems of distributed interacting objects. The constituting bricks of a
From Symmetric Nets to Differential Equations exploiting Model Symmetries
This paper proposes a new approach based on fluid approximation to automatically derive from an SSN model a set of Ordinary Differential Equations which mimic the system behavior, and shows how the SSN formalism allows to define an efficient translation method which reduces the size of the corresponding ODE system with an automatic exploitation of system symmetries.
Differential Equivalence Yields Network Centrality
This paper investigates the relationship between centrality measures and model reduction techniques, such as lumpability of Markov chains, which seek to reduce a model into a smaller one that can be processed more efficiently, while preserving information of interest.
The Mean Drift: Tailoring the Mean Field Theory of Markov Processes for Real-World Applications
The concept of population processes and the conditions under which the approximation theorems apply are presented, and it is shown how the mean drift can be linked to observations which follow from the propagation of chaos.
Analysis of Large Populations of Interacting Objects with Mean Field and Markovian Agents
This work presents Mean Field analysis and Markovian Agents models, which produce exact results if the number of considered objects goes to the infinity, but provide reasonable approximations even for finite quantities.


A Generic Mean Field Convergence Result for Systems of Interacting Objects
We consider a model for interacting objects, where the evolution of each object is given by a finite state Markov chain, whose transition matrix depends on the present and the past of the
Advanced mean field methods: theory and practice
The theoretical foundations of advanced mean field methods are covered, the relation between the different approaches are explored, the quality of the approximation obtained is examined, and their application to various areas of probabilistic modeling is demonstrated.
Superposition of Interacting Aggregated Continuous-Time Markov Chains
Consider a system of interacting finite Markov chains in continuous time, where each subsystem is aggregated by a common partitioning of the state space. The interaction is assumed to arise from
A class of mean field interaction models for computer and communication systems
Comparison of Multilevel Methods for Kronecker-based Markovian Representations
Numerical experiments demonstrate that so far ML methods with successive over-relaxation as smoother provide the most effective solvers for considerably large Markov chains modeled as HMMs with multiple macrostates.
Hierarchical Markovian Models: Symmetries and Reduction
A Methodology for Solving Markov Models of Parallel Systems
Fluid flow approximation of PEPA models
  • J. Hillston
  • Mathematics, Computer Science
    Second International Conference on the Quantitative Evaluation of Systems (QEST'05)
  • 2005
A novel performance analysis technique for large scale systems modelled in the stochastic process algebra PEPA, whose underlying mathematical representation is a set of coupled ordinary differential equations (ODEs).
Stochastic Automata Network For Modeling Parallel Systems
An approach based on a modular state-transition representation of a parallel system called the stochastic automata network (SAN) is developed, which is automatically derived using tensor algebra operators, under a format which involves a very limited storage cost.
Hierarchical structuring of superposed GSPNs
  • P. Buchholz
  • Computer Science
    Proceedings of the Seventh International Workshop on Petri Nets and Performance Models
  • 1997
A pre-processing step is introduced to generate a hierarchical structure which defines a block structure of the generator matrix, where every block can be represented in a compact form similar to the representation of generator matrices originally proposed for SGSPNs or SANs.