Simulation News Europe (SNE) features a series on comparisons
of simulation software. Based on simple, easily comprehensible models
special
features of modelling and experimentation within simulation languages,
also with respect to an application area, are compared.
Features are, for instance: modelling technique, event handling,
numerical
integration, steady-state calculation, distribution fitting, parameter
sweep, output analysis, animation, complex logic strategies, submodels,
macros, statistical features etc.
Ten comparisons have been defined in Simulation News Europe, the series
will be continued. Furthermore,
a special comparison of parallel simulation techniques has been
defined.
- Comparison 1 (Lithium-Cluster
Dynamics
under Electron Bombardment, November 1990)
- deals with a stiff system of 3rd order. This comparison tests
features
for integration of stiff systems, for parameter variation, and for
steady
state calculation. A preliminary summary can be found in SNE 6,
November
1992.
- Comparison 2 (Flexible Assembly
System,
March 1991, comments July 1991)
- for discrete simulation languages compares features for submodel
structures,
control strategies, and optimization of process parameters. A
preliminary
evaluation can be found in SNE 4.
- Comparison 3 (Analysis of a
Generalized
Class-E Amplifier, July 1991)
- focusses on simulation of electronic circuits and requires
features
for table functions, eigenvalue analysis, and complex experiments.
- Comparison 4 (Dining Philosophers
I,
November 1991)
- is a more general task involving not only simulation but also
different
modelling techniques like Petri nets.
- Comparison 5 (Two State Model,
March
1992, revised July 1992)
- primarily addresses simulation tools with very high accurracy. It
checks
integration and state event handling with high accurracy.
- Comparison 6 (Emergency
Department
- Follow-up Treatment, November 1992)
- addresses discrete simulation languages and tests features for
modelling,
concepts of availability, and complex control strategies.
- Comparison 7 (Constrained
Pendulum,
March 1993)
- for continuous simulation languages, checks features for model
comparison,
state events, and boundary value problems.
- Comparison CP1 (Parallel
Comparison,
March 1994)
- This type of comparison diffenrent to the others deals with the
benefits
of distributed and parallel computation for simulation tasks. Three
test
examples have been chosen to investigate the types of parallelisation
techniques
best suited to particular types of simulation tasks.
- Comparison 8 (Canal-and-Lock
System,
March 1996)
- for discrete simulators checks features for modeling complex
logic,
which has to be verified by deterministic datasets. Also variance
reduction
capabilities are checked.
- Comparison 9 (Fuzzy Control of a
Two
Tank System, July 1996)
- asks for modules for fuzzy control or how such modules can be
implemented
efficiently.
- Comparison 10 (Dining
Philosophers
II, November 1996)
- reviews discrete simulators with respect to simultaneous
(concurrent)
access to resources and with respect to deadlocks.
- Comparison 11 (SCARA Robot,
March
1998)
- is the 6th comparison of continuous type and deals with the
handling
of implicit systems.
- Comparison 12 (Collision
Processes in Rows of Spheres, November 1999)
- This comparison deals with a model of the mechanics. The features
to be compared represent a large number of events, the numerical
accuracy, the iteration of a boundary value, and stochastic parameter
variations. Piecewise, constant velocities permit both a continuous and
a discrete treatment
- Comparison
13 (Crane Crab with Embedded Control, April 2001)
- This comparison checks techniques and features for embedded
digital control with sensors and with observersystems. The special
problems are discrete control coupled with sensor diagnosis and
observers and complex experiments.
- Comparison
14 (Supply Chain, November 2001)
- 4 factories produce 12 different products which are supplied to a
distributor. This comparison addresses discrete simulator - features
for supply chain systems (messages, strategies). Special problems that
occur are distinction between material flow and order flow and
distance-dependent control strategies.
- Comparison
15 (Clearance Identification, December 2002)
- This comparison checks identification features (based on measured
data) and influences of noise. Problems that occur are the appropriate
identification algorithms, short-term input functions (Dirac-like) and
the support of statistics.
- Comparison
16 (Restaurant Business Dynamics, May 2004)
- addresses agent-based simulation as well as DEVS approach and
classical programming. Occuring special problems are the renaissance of
activity scanning, coordination of run samples and optimisation.
- Comparison
17 (Spatial Dynamics of SIR-Type Epidemic, December 2004)
- A comparison which analyses temporal and spatial behaviour of a
disease spreading process by cellular automata models. Difficulties
that can be encountered are proper features for cellular automata in
simulation systems and the comparison of spatial/temporal results with
pure temporal results.
- Comparison
18 (Neural Networks versus Transfer Functions - Identification
of Nonlinear Systems, July 2005)
- compares transfer function modelling and neural net modelling for
given data of a nonlinear process. Special problems are proper features
for neural net modelling in the simulation system and combination of
transfer functions with neural nets for parameter tuning
- Comparison
19 (Pollution in Groundwater Flow, December 2005)
- studies the flow of contamination in the ground water in 2D-space
and time and allowing different modelling approaches for the spatial
behaviour (numerical PDE solution, discretisation to ODEs, cellular
automata and so on). Problems that have to be considered are features
for description of spatial dynamics and combination of spatial/temporal
behaviour with temporal behaviour of control inputs
We invite all readers to participate in these comparisons. Please,
simulate
the model(s) with a tool of your choice and send a report to the
editors
in the following form (on diskette, any word processing format, or per
email or transfer to our ftp-server):
- short description of the language,
- model description (part of source code, diagram, ...),
- results of the tasks with experimentation comments, max. 1 page.
(For
publication in Simulation News Europe all contributions
that exceed one page will be modified by the editors to fit into one
page.)
Reports of solutions of the Parallel Comparison should not be more than
one and a half page in length.
We offer to place the full model (source code, graphics, etc.) and
additional
information on our WWW server. We also invite you to prepare
animations.
Please send files and additional information in HTML-format.
Summaries of the comparisons are presented at conferences and
published
from time to time in SNE. Overviews on Comparison 1 and Comparison 2
(all
solutions, partly extended solutions, model descriptions, source codes
and all published summaries) have been published in the series ARGESIM
Report.