[comp.simulation] SIMULATION DIGEST V14 N4

simulation@uflorida.cis.ufl.edu (Moderator: Paul Fishwick) (02/27/90)

Volume: 14, Issue: 4, Mon Feb 26 16:32:34 EST 1990

+----------------+
| TODAY'S TOPICS |
+----------------+

(1) RE: High Level Design
(2) TR: Time Warp Method
(3) SmartModel on HP Workstations

* Moderator: Paul Fishwick, Univ. of Florida
* Send topical mail to: simulation@bikini.cis.ufl.edu OR
  post to comp.simulation via USENET
* Archives available via FTP to bikini.cis.ufl.edu (128.227.224.1).
  Login as 'ftp', use your last name as the password, change
  directory to pub/simdigest.
* Simulation Tools available by doing above and changing the
  directory to pub/simdigest/tools.



-----------------------------------------------------------------------------

Date: Thu, 22 Feb 90 15:18 MST
From: ROZENBLIT%EVAX2@Arizona.EDU
Subject: IN reply to Wolfgang Mueller----High Level Design
To: simulation@bikini.cis.ufl.EDU
X-Envelope-To: simulation@bikini.cis.ufl.EDU
X-Vms-To: IN::"simulation@bikini.cis.ufl.edu"

             SUMMARY OF RESEARCH IN KNOWLEDGE-BASED 
                  SYSTEM DESIGN AND SIMULATION

     Our    research   employs   Artificial   Intelligence    and 
Multifacetted  Simulation Modelling to unify  engineering  design 
activities  and develop a methodology for  systematic  simulation 
model  construction and evaluation. The methodology is  based  on 
codifying  appropriate  decompositions, taxonomic,  and  coupling 
relationships.  This  constitutes  declarative  design  knowledge 
base.  Beyond this, we provide the procedural knowledge  base  in 
the  form of production rules used to process the elements  in  a 
design domain.

     As a step toward a complete knowledge representation scheme, 
we  have  combined  the decomposition,  taxonomic,  and  coupling 
relationships in a representation scheme called the system entity 
structure,  a declarative scheme related to  frame-theoretic  and 
object-based   representations.   The  entities  of  the   entity 
structure  refer  to conceptual components of reality  for  which 
models  may  reside  in the model  base.   Also  associated  with 
entities  are slots for attribute knowledge  representation.   An 
entity  may have several aspects, each denoting a  decomposition, 
and  therefore having several entities.  An entity may also  have 
several  specializations, each representing a  classification  of 
the  possible variants of the entity.  The generative  capability 
of  the  entity  structure  enables  convenient  generation   and 
representation   of  model  attributes  at  multiple  levels   of 
aggregation and abstraction. 

     A primary application of the above knowledge  representation 
scheme is the objectives-driven development of simulation models.  
In  this  approach,  a  model  is  synthesized  from   components 
identified through the system entity structure and stored in  the 
model  base.   The  synthesis  process  is  guided  by  project's 
objectives, requirements, and constraints. The objectives guide a 
pruning process which reduces the entity structure to one or more 
composition  trees from which models may be hierarchically  built 
up from atomic components.  Constraints, expressed in the form of 
production  rules  and  placed  on  the  aspects  of  the  entity 
structure, restrict the family of possible pruned structures  for 
more informed search.  

     Performance  of design models is evaluated through  computer 
simulation in DEVS-Scheme environment. DEVS-Scheme is an  object-
oriented  simulation    environment  for  modeling   and   design  
that   facilitates  construction  of  families  of  models  in  a 
form easily reusable by  retrieval from a model base.  Models are 
evaluated  in  respective experimental  frames.  An  experimental 
frame  defines  a  set of input,  control,  output,  and  summary 
variables.  Those objects specify conditions under which a  model 
is simulated and observed. The environment supports  construction  
of   distributed,  hierarchical  discrete  event  models  and  is 
written in the  PC-Scheme  language which runs on IBM  compatible 
microcomputers and  AI Workstations. 

     We  have  been  validating the  above  methodology  in  case 
studies  involving design and simulation of distributed  computer 
architectures,  local  area  networks, and  more  recently,  VLSI 
packages.  


For more information, please contact:

Jerzy W. Rozenblit or Bernard P. Zeigler
Dept. of ECE, Bldg #4
University of Arizona
Tucson, AZ 85721

rozenblit@arizevax.bitnet
zeigler@arizevax.bitnet


------------------------------

Date: Fri, 23 Feb 90 17:59:41 -0800
From: liny@cs.washington.edu (Yi-Bing Lin)
Return-Path: <liny@cs.washington.edu>
To: fishwick@bikini.cis.ufl.edu

The following  technical report can be requested via
e-mail: liny@cs.washington.edu

 Reducing the State Saving Overhead For Time Warp Parallel Simulation

		Yi-Bing Lin and Edward D. Lazowska
	  Department of Computer Science and Engineering
		    University of Washington
		       Seattle, WA 98195

			    Abstract

The Time Warp mechanism is the most common ``optimistic'' parallel
simulation protocol.  A process executes every message as soon as it
arrives.  If a message with a smaller timestamp subsequently arrives,
the process rolls back its state to the time of the earlier message
and re-executes from that point.

Clearly, the state of each process must be saved (checkpointed)
regularly in case rollback is necessary.  Although most existing Time
Warp implementations checkpoint after every state transition, this is
not necessary, and the checkpoint interval is in reality a tuning
parameter of the simulation.

In a previous paper, we derived the optimal frequency of checkpointing
in Time Warp simulation based on a specific assumption concerning the
rollback distance distribution.  This paper derives distribution-free
bounds for the state saving overhead.  Using these bounds, we are able
to select a checkpoint interval which minimizes the state saving
overhead.  High accuracy for our approach is shown in an experimental
study.



------------------------------

Date: Thu, 22 Feb 90 16:32:29 EST
From: mitel!spock!tsuia@uunet.UU.NET (Alan Tsui)
To: uunet!bikini.cis.ufl.edu!simulation@uunet.UU.NET
Subject: SmartModel


	I would like to know anyone has used SmartModel with System HILO
	(on HP workstation).  How easy is it to use?  Are they significant
	better than HILO models?  Why?

	Thanks.
	Alan



------------------------------




END OF SIMULATION DIGEST
************************