NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/simulati/55

Internet Message Format | 1992-08-26 | 36.4 KB

Path: sparky!uunet!gatech!mailer.cc.fsu.edu!uflorida!simulation
From: [e-mail] (Moderator: Paul Fishwick)
Newsgroups: comp.simulation
Subject: SIMULATION DIGEST V28 N10
Message-ID: <[e-mail]>
Date: 26 Aug 92 19:01:24 GMT
Sender: [e-mail]
Reply-To: [e-mail]
Lines: 899
Approved: [e-mail]

Volume: 28, Issue: 10, Wed Aug 26 15:00:49 EDT 1992

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


[NEW QUESTIONS]
  DYNAMO Packages for MAC and UNIX
  Implementing the Pspice LAPLACE Option
  Dewdney's Planiverse
  Fluid Dynamics Simulation
  Generating Qualitative Descriptions
[REGARDING PAST QUESTIONS]
  Simulating Group Behavior
  N Body Simulation
[CALL FOR PAPERS/PARTICIPATION]
  Computer Aided Design
[DEPARTMENTS]
  Simulation in the Service of Society

* Moderator: Paul Fishwick, Univ. of Florida
* Send topical mail to: [e-mail] OR
  post to comp.simulation via USENET
* Archives available via FTP to bikini.cis.ufl.edu (128.227.224.1).
  Login as 'anonymous', use your e-mail address as the password, change
  directory to pub/simdigest. Do 'binary' before any file transfers.
* Simulation Tools available by doing above and changing the
  directory to pub/simdigest/tools.




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


Subject: [NEW QUESTIONS]


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


From [e-mail] Wed Aug 19 01:21:35 1992
    (5.65/UW-NDC Revision: 2.22 ) id AA09388; Tue, 18 Aug 92 22:21:34 -0700
Date: Tue, 18 Aug 92 22:21:34 -0700
From: Patrick Vye <[e-mail]>
Sender: [e-mail]
To: [e-mail]
Subject: System Dynamic Simulation package?

Hello,

I am trying to get some help finding a simulation package.

I am trying to find a continuous simulation package that is similar
to DYNAMO but works on the MAC or under UNIX.

I don't need a graphical capability, but would like the syntax
to be similar to DYNAMO so that I can set up an an input file that
specifies rate, level and auxiliary equations.  It can solve the
equations using Euler's method or Runge-Kutta.  Again, there is
no need for a graphical capability for either input or output.

I plan to use the simulation in batch with another application.

Any pointers to a package would be greatly appreciated.

Pat




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

To: [e-mail]
Path: corton!geocub!gonzales!bardo
From: [e-mail] (Bardo Muller)
Newsgroups: comp.lsi.cad,comp.lsi,comp.simulation,alt.cad,sci.comp-aided,sci.math.num-analysis
Subject: PSPICE, option LAPLACE - algorithm ?
Keywords: spice,laplace transform,transient analysis,convolution
Date: 19 Aug 92 14:00:00 GMT
Sender: [e-mail] (le facteur fnet)
Reply-To: [e-mail]
Followup-To: comp.lsi.cad
Organization: Universite de Paris


Hi there!

Does anyone have any pointers, code, literature or other comments on how to
implement something similar to the PSPICE LAPLACE option ?

[ you describe an (arbitrary ?) function F(s) in the frequency domain which is then used in a (nonlinear, variable timestep) transient analysis ]

Convolution with the inverse transform of the impulse response should be
one viable method. Is there any more analytical (even approximative) approach known ?

At the moment I experiment with a backward-euler for rational functions N(s)/D(s).

Ideas ?  Please e-mail. I will post a summary of the proposals.

Thank you !

Bardo

___________________________________________________________________________
| Bardo MULLER                          | phone  : [33] [1] 69 41 78 50    |
| Institut d'Electronique Fondamentale  | fax    : [33] [1] 60 19 25 93    |
| Bat. 220 Universite Paris Sud         | e-mail : [e-mail]  |
| 91405 ORSAY CEDEX FRANCE              |          THE NET WORKS !         |
|_______________________________________|__________________________________|



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

Posted-Date: Mon, 24 Aug 1992 16:37:27 GMT
Newsgroups: comp.simulation
Path: wlh04953
From: [e-mail] (William L. Hayden)
Subject: planiverse
Relay-Version: version Trumpet 1.03; site elp.aero.org
Summary: Is there any life on this group ?
Sender: [e-mail]
Organization: The Aerospace Corporation
Posting-Version: version Trumpet 1.03; site elp.aero.org
Date: Mon, 24 Aug 1992 16:37:27 GMT
Apparently-To: [e-mail]

Just read DEWDNEY's Planiverse.
Does the class project program 2DWORLD or equivalent exist ?
If so is it available for download somewhere (FTP) ?

Thanks in advance for any response.



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

Newsgroups: sci.areonautics,comp.simulation
Path: johng
From: infmx![e-mail] (John Galloway)
Subject: Fluid dynaics sim (50mph air) on a Quadra 700, is it possible?
Originator: johng@godzilla
Sender: infmx![e-mail] (Usenet News)
Organization: Galloway Research
Distribution: ca
Date: Tue, 25 Aug 92 09:30:14 GMT
Apparently-To: uunet!comp-simulation

I have been tinkering around with various edge treatments for the
windshield on my motorcycle in an attempt to reduce/move/affect
the turblance layer that comes off the trailing/top edge (right into
my helmeted head (helmet skirts and/or ear plugs are the existing
answer).

I would like to get more scientific about this and am wondering if
my mostly idle but up all day Quadra 700 A/UX system could be useful
doing fluid dynamics simulations of proposed configurations.

I would appreciate some guesses as to the complexity of such calculations
(navier-stokes or something like that?) needed to carry out such simulations
(even assuming the sw is available and not toooo expensive).

Just wild guessing I get:

3.00            width (ft.)
3.00            length
3.00            height

46,656.00       cubic inches
8.00            points/inch
23,887,872.00   points/data set

3.00            floats per point in dataset (3D pressure vector?)
4.00            bytes per float
15.00           flops/point/itteration (vector to components and back and?)
1,025.16        MFlops/itteration
50            itterations (gives ?? 1/4 second at 50 mph?)
500.00          MFLOPS of sys
30              % of run MFLOPing
341.72          hours for run
14.24           days for run

0.01            MB/sec (reading/writing compressed data for run)
273.38          MB/raw data set
2.73            compression ratio to achive:
100.00          MB/compress data

Which says that even on a 500MFLOP machine I would need 1/2 month, with 1/8
inch grid spacing, on a 5(?)MFLOP Quadra => no way.  Is this problem really
not this bad, or have I guessed close enough to be clear that its totally not
feasable to get any meaningfull results (for design assistance) within a few
days on a Quadra.  The data set size is actually viable, I could make good
use of a new 1.5GB disk, and 68MB of memory, but its the crunchyness, or
lack of it, that kills the idea.  right?  thanks.
        -jrg

internet    [e-mail]  John R. Galloway, Jr  795 Beaver Creek Way
applelink   D3413                  CEO...receptionist    San Jose, CA   95133
                                   Galloway Research     [phone]
 --
internet    [e-mail]  John R. Galloway, Jr  795 Beaver Creek Way
applelink   D3413                  CEO...receptionist    San Jose, CA   95133
                                   Galloway Research     [phone]



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

Newsgroups: comp.ai,comp.simulation
Path: cbl6rjw
From: [e-mail] (R.J. Williams)
Subject: qualitative description of the output of a quantitative simulation
Sender: [e-mail] (Usenet poster)
Nntp-Posting-Host: gps1.leeds.ac.uk
Organization: University of Leeds, England
Date: Wed, 26 Aug 1992 16:35:20 GMT
Apparently-To: [e-mail]

Hello,

I am trying to find information on any systems that can generate
qualitative descriptions of output from a quantitative simulation.  We
are utilising a quantitative modelling system which uses a flow metaphor;
instrumentation can be placed anywhere on the model.  Each piece of
instrumentation produces a value every dt.

[[ED: You might consider using qualitative analysis for systems of
differential equations. Consider "Practical Numerical Algorithms for
Chaotic Systems" by Parker and Chua for algorithms yielding qualitative
insights (Lyapunov exponents, fixed points, etc.). Also, there is work
in AI that has focused on this area -- see the past few AAAI/IJCAI
proceedings -PAF]]

We have a knowledge based version of the quantitative model with
dependencies represented (but not the actual formulas).  We would
like to produce a qualitative description using this together with the
quantitative output.  Recognising maxima, minima, decreasing and
increasing values is ok, but we would also like to be able to notice
periodicity, possible relationships between features in different
output streams and trend analysis.

Thanks for any pointers,

rod
 ----
Roderick Williams          [e-mail]



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


Subject: [REGARDING PAST QUESTIONS]


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


Newsgroups: comp.simulation
Path: news
From: [e-mail] (Douglas W. Jones,201H MLH,3193350740,3193382879)
Subject: Re: Application to the Softer Sciences
Sender: [e-mail] (News)
Date: Fri, 21 Aug 1992 16:25:15 GMT
References: <[e-mail]>
Nntp-Posting-Host: pyrite.cs.uiowa.edu
Organization: University of Iowa, Iowa City, IA, USA
Apparently-To: [e-mail]

>From article <[e-mail]> in sci.cognitive,
by [e-mail] (James G Spice):

> I am a graduate student in Political Science, a field not well known
> for its mathamatical or modeling sophistication (though exceptions
> certainly exist).  Does anyone know of any applications of cognitive
> science methods in the softer social sciences?

I can't help in the area of political science, but I'm involved in
developing a system for simulating the behavior of small groups of
people confonting crisis situations, known as Icarus because it is
a bit of a high-risk venture.  Icarus uses a discrete-event simulation
framework (as commonly used in operations research) with small expert
systems used to model each human participant in the group.

One aim of Icarus is to allow simulation models to express the contents
of the "rule books" for such contexts as nuclear reactor control rooms
in order to ask the question "if people played by these particular
rules through this scenario, what is the likelyhood of a successful
outcome."  The cost of such experiments using human subjects can be
outrageous, but cased like Three Mile Island and the Stark incident
make such experiments quite valuable.

A second aim is to allow the formulation of more nearly formal theories
about small group behavior in the form of simulation models.  The use
of simulation methodology can give such a theory a predictive value
comparable to the predictive value of traditional theories in the hard
sciences, and this can significantly reduce the number of experiments
needed to explore the behavior of such a group.

The Icarus simulator is extremely limited in what it can accomplish!
It cannot model learning, metaknowledge (that is, my knowledge about
what I or others know), complex concepts, or convoluted reasoning, and
we cannot handle more than a handfull of people in one of our models.

Without learning, Icarus can be used to model crisis situations where
the feedback on which learning could be based happens too late to
influence the outcome.  We can deal with simple quantitative knowledge,
including, eventually, fuzzy quantities, and this should be quite
adequate for many purposes, for example, the reactor control example
where things like water pressure, temperature, and level are the key
bits of knowledge exchanged by the participants.

Currently, our work on Icarus is bogged down in some interesting
technical problems that arise when you try to combine expert systems
with discrete-event simulation.  We now know two or three ways to fix
these problems, but we don't know which is right way to go.  When we've
solved these problems, we'll start to try to attract a user community
from among small-group sociologists and the management science
community.
                Doug Jones
                [e-mail]





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

Newsgroups: comp.simulation
Path: uad1184
From: [e-mail] (Dino Dini)
Subject: Re: Gravitational motion errors
Date: Tue, 25 Aug 1992 22:53:47 GMT
References: <[e-mail]>

[e-mail] writes:

>I am going to be writing a little routine to do N-body gravitational
>motion simulations in 2 dimensions.
>
>The first time I tried this I had some sort of error (perhaps a rounding
>error) that caused free objects to continuously gain momentum down
>the screen.  When I modified it to allow objects that drop off the
>bottom of the grid/screen to reappear at the top, they quickly gained
>significant momentum and zipped around in an interesting, if not
>useful, way.
>
>Does it sound like I just need to be more careful about rounding
>off numerical values?
>
>Sorry if this is lame... I haven't much simulation experience.
>
>Michael McGinnis                Internet: [e-mail]
>Computer Center                   Bitnet: mcginnis@ukanvax
>University of Kansas               Voice: [phone]
>Lawrence, KS  66045                  FAX: [phone]
>
>Rust never sleeps.

It is difficult to say without seeing your listing. I have done
an N-body simulation crudely in X using floating point. Rounding
errors create inaccuaracies, but the general behavior is consistent
with what you expect. From the sound of it you had a fairly
major bug in there.



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


Subject: [CALL FOR PAPERS/PARTICIPATION]


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


To: [e-mail]
Subject: ICCAD-92 Registration (CAD Conference)
Path: roundup.crhc.uiuc.edu!news
From: [e-mail] (W. Kent Fuchs)
Newsgroups: comp.lsi.cad,comp.lsi,comp.simulation,comp.lang.vhdl,news.announce.conferences,comp.lsi.testing,comp.cad.cadence,alt.cad,comp.sys.mentor,comp.lang.verilog
Subject: ICCAD-92 Registration (CAD Conference)
Date: 24 Aug 1992 22:08:32 -0500
Organization: Center for Reliable and High-Performance Computing, University of Illinois at Urbana-Champaign
Nntp-Posting-Host: roundup.crhc.uiuc.edu
Summary: Registration Information for ICCAD-92
Keywords: CAD Conference Registration Tutorials



       ACM/IEEE  International Conference on Computer-Aided Design
        November 8-12, 1992,  Santa Clara, CA  USA

                    ICCAD-92

          Preregistration deadline: October 12, 1992

    Contents of This Post:

    1) Conference Overview
    2) Schedule of Conference Sessions
    3) Registration Forms (print, clip and mail)

 ---------------------------------------------------------------------------
   CONFERENCE OVERVIEW      CONFERENCE OVERVIEW      CONFERENCE OVERVIEW
 ---------------------------------------------------------------------------


    The Tenth  International  Conference  on  Computer-Aided  Design,
    ICCAD-92,  offers an exceptionally strong schedule of papers, tu-
    torials, and vendor suites. The program promises to challenge ex-
    perienced   CAD  developers,  hardware  designers,  and  academic
    researchers who want to learn from the latest in CAD research and
    development.

    ICCAD-92 will be held November 8-12, 1992  in  the  Santa  Clara
    Convention  Center, Santa Clara, California.  The conference con-
    centrates on CAD for electronic design and features three days of
    technical  sessions,  one day of  four different tutorials, and a
    lively evening panel discussion.    To  complement  the technical
    program,   numerous  CAD  vendors  will  have suites at ICCAD-92.
    This offers the conference attendees a unique opportunity to hold
    in-depth  technical discussions with vendors on their latest  pro-
    ducts.

    An advance  program  is  available and  can be requested from the
    conference managers, MP associates, by phone, fax or mail:

    MP Associates
    7490 Clubhouse Road, Suite 102
    Boulder, CO  80301

    Telephone: [phone]

    FAX:       [phone]

 --------------------------------------------------------------------------
 ICCAD-92 AT A GLANCE      ICCAD-92 AT A GLANCE      ICCAD-92 AT A GLANCE
 --------------------------------------------------------------------------

                      Monday, Nov. 9

     8:30                       PLENARY SESSION
    10:00

    10:30   2A: Design for       2B: Technology and  2C: FPGA Synthesis
    12:30       Test                 Layout

     2:00   3A: Waveform         3B: Topics in       3C: Asynchronous
     3:30       Evaluation           Simulation          Circuit Synthesis

     4:00   4A: Clocking and     4B: High Density    4C: Formal
     6:00       Level Sensitive      Module              Hardware
                Latches              Assembly            Verification

     6:30       PANEL  "University Faculty: Visionaries or Mercenaries?"
     8:00

                     Tuesday, Nov. 10

     8:30   5A: CMOS Power      5B: Sequential      5C: High-Level
    10:00       and Timing          Circuit             Design
                Estimation          Test Generation

    10:30   6A: Classical       6B: Testing &       6C: Synthesis for
    12:30       Simulation          Diagnosis           DSP Applications


     2:00   7A: Analog          7B: Multi-View      7C: Timing in High-
     3:30       CAD                 Design              Level Synthesis
                                    Representations

     4:00   8A: High            8B: Detailed        8C: Topics in
     6:00       Performance         Routing             Logic
                Simulation                              Synthesis

     7:00       Banquet
    10:00

                     Wednesday, Nov. 11

     8:30   9A: Partitioning   9B: Interconnect    9C: Engineering and
    10:00       & Clustering       Analysis            Education:
                                                       Trends and Needs

    10:30  10A: High           10B: Hardware/     10C: Retiming and
    12:30       Performance         Software           Sensitization
                Routing             Co-Design          Conditions

     2:00  11A: Design         11B: Delay          11C: Asynchronous
     3:30       Management          Testing             Synthesis
                Styles

     4:00  12A: Placement &    12B: High-Level     12C: Hazards in
     6:00       Floorplan           Testing             Combinational
                Design                                  Synthesis

                     Thursday, Nov. 12

    Tutorials:

        - Multi-Level Logic Synthesis
        - Interconnect and Packaging Analysis
        - Introduction to Embedded System Design
        - Architectures for Software Systems

 -----------------------------------------------------------------------------
Please notice that registration and hotel reservation are two different forms
that go to two different places, and that only the hotel reservations can use
a credit card.
 -----------------------------------------------------------------------------

 -----------------------------------------------------------------------------
REGISTRATION REGISTRATION REGISTRATION REGISTRATION REGISTRATION REGISTRATION
 -----------------------------------------------------------------------------
            Advance Registration Form
                ICCAD-92
        November 8-12, 1992,  Santa Clara, CA  USA

              Advance Registration Deadline: October 12, 1992

    Send completed registration form and check payable to ICCAD-92 to:

    MP Associates
    Registration Desk
    7490 Clubhouse Rd. #102
    Boulder, CO 80301

    Information Only: [phone]

    All advance registrations  must  be  postmarked  by  October  12,
    1992.  On-site conference payments must be made by check or cash.
    NO CREDIT CARDS!  No refunds unless cancellation is received  be-
    fore October 12, 1992.  All refunds are subject to a $15 fee.

    Name _______________________________________________________________

    Company ____________________________________________________________

    Mailing Address ____________________________________________________

    Mail Stop __________________________________________________________

    City ________________________________ State __________ Zip _________

    Country _____________________________ Telephone (_____)_____________

    Member No. ACM/IEEE ________________________________________________

    Conference Registration Fees:

                   Before    After
                       Oct. 12   Oct. 12   Amount

        IEEE/ACM Members    $200      $260      ___________
        Non-Members         $260      $325      ___________
        Students            $100      $100      ___________
        One Day Only        $120      $140      ___________
        select day:    ____ Mon   ____ Tues   ____ Wed

    Tutorial Fees:
                   Before    After
                       Oct. 12   Oct. 12   Amount

    IEE/ACM Member Full Day    $280      $325      ___________
    Non-Member Full Day    $350      $435      ___________
    Students                $100      $100      ___________

    Total Amount (Conference and Tutorial)        ___________

    TUTORIAL REGISTRATION: To participate in the tutorials there must
    be at least a one-day conference registration.   If you have paid
    the proper amount to  include a tutorial,  please  indicate first
    (1)  and  second (2) choices.   Tutorials are limited to a first-
    come, first-served basis.

    _______ Multi-Level Logic Synthesis
    _______ Interconnect and Packaging Analysis
    _______ Introduction to Embedded System Design
    _______ Architectures for Software Systems

 -----------------------------------------------------------------------------
Please notice that registration and hotel reservation are two different forms
that go to two different places, and that only the hotel reservations can use
a credit card.
 -----------------------------------------------------------------------------

 -----------------------------------------------------------------------------
HOTEL HOTEL HOTEL HOTEL HOTEL HOTEL HOTEL HOTEL HOTEL HOTEL HOTEL HOTEL HOTEL
 -----------------------------------------------------------------------------

Hotel Reservations for ICCAD: 10th Intl. Conference on Computer Aided Design
        November 8-12, 1992,  Santa Clara, CA  USA

               Hotel Reservation Deadline: October 12, 1992

    Hotel reservations will be handled on a first-come,  first-served
    basis  by  the  Santa Clara Housing Bureau.  Reservations must be
    postmarked by OCTOBER 12, 1992 to guarantee availability of rooms
    and  rates  at  the  conference hotels.  The following hotels are
    available:

    Please indicate first (1) and second (2) choices.

    HOTEL            Single        Double
    ____ Westin Hotel    $ 95.00        $105.00
    ____ Holiday Inn    $ 53.00        $ 53.00
    ____ Days Inn        $ 53.00        $ 53.00

    The Days Inn and the Holiday Inn will  provide shuttle service to
    the Santa Clara Convention Center.   The Westin Hotel is attached
    to the Conference Center.

    Indicate hotel preference above and send completed hotel form to:

    Santa Clara Housing Bureau I
    C/O Santa Clara Convention and Visitors Bureau
    PO Box 387
    Santa Clara, CA 95052                  Caution: Hotel Reservations
                        sent to MPA office will be
    Information Only: [phone]        discarded!  No telephone
    Fax Reservations: [phone]         reservations!

    Name _____________________________________________________________

    Company ___________________________________ Mail Stop ____________

    Mailing Address __________________________________________________

    City __________________________________ State ______ Zip _________

    Country __________________________ Phone (_____) _________________

    FAX  (____) ______________________________

    Arrival Date _____________________________ Time __________ AM / PM

    Departure Date ___________________________ Time __________ AM / PM

    A deposit of $85.00, regardless of hotel room rate,  is  required
    by check or credit card.  Make check payable to Santa Clara Hous-
    ing Bureau.  Your reservation will be confirmed by both the Santa
    Clara  Housing  Bureau and the hotel at which you have a reserva-
    tion.  The $85.00 deposit will be applied to your  room  bill  at
    check-in.

          ____ VISA  ____ MasterCard  ____ American Express

    Number_________________________________ Expiration Date __________

    Signature_________________________________________________________

 -----------------------------------------------------------------------------
Please notice that registration and hotel reservation are two different forms
that go to two different places, and that only the hotel reservations can use
a credit card.
 -----------------------------------------------------------------------------



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


Subject: [DEPARTMENTS]


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


From [e-mail] Fri Aug 21 17:39:01 1992
Date:    Fri, 21 Aug 92 21:20:51 GMT
From: [e-mail]
Subject: e-s3-8-1

The following sample issue of our electronic magazine, "E-S3",
covering selected topics about computer modeling and simulation,
is sent to you with the compliments of the publisher of the
technical journal SIMULATION, the Society for Computer
Simulation, and John and Suzette McLeod, the Editors of
Simulation in the Service of Society (S3), a special section of
that journal.  If you do not care to receive future issues please
type REPLY -- cancel.  Let's not clutter up our E-mail with "Junk
Mail"!
 -----------------------------------------------------------------

                  E-S3  Vol. 1, No. 8, Part One

                based on the August 1992 issue of
              "Simulation in the Service of Society"
John McLeod, Technical Editor     Suzette McLeod, Managing Editor
          8484 La Jolla Shores Dr., La Jolla, CA 92037.
                   E-mail: [e-mail]
                                *
                   S3 is a special section of
                           SIMULATION
                   the monthly journal of the
                 SOCIETY for COMPUTER SIMULATION
             P.O.Box 17900, San Diego, CA 92177-7900
            Phone: [phone] FAX: [phone]
                                *
[Copyright Notice: E-S3 is the electronically delivered version
of "Simulation in the Service of Society" which is a special
section of SIMULATION, a monthly technical journal of the
Society for Computer Simulation International. It may be
reproduced only for personal use or for the use of students. In
any case full credit must be given to the original source of
publication: SIMULATION 59:2, August 1992.
  All rights  reserved, (c) 1992, Simulation Councils, Inc.]
 -----------------------------------------------------------------

Oil on Troubled Waters

Herewith we present edited excerpts from an article originally
published in Water Resources Research, v. 26, No. 9, September
1990.

     Multidimensional Simulation Applied to Water Resources
                          Management

             A.S. Camara, F.C. Ferreira, M.J.Seixas
              Environmental Systems Analysis Group
               New University of Lisbon, Portugal

                           D.P. Loucks
          School of Civil and Environmental Engineering
                       Cornell University
                        Ithaca, New York

Introduction

Mathematical modeling has been traditionally dominated by
quantitative formulations which can be easily manipulated.  Thus
reality, which is usually described qualitatively by pictures and
words, has been mapped into numerical representations.  In this
paper we explore how spatially related objects can be represented
and manipulated visually and how abstract concepts can be
represented using natural language.

Today models using linguistic and pictorial entities and
operators, in addition to the traditional numerical formulations,
are clearly lacking.  Modelers tend to limit the use of pictures
or graphics to the input-output stages of the modeling process
and use natural language as a medium of interaction between
humans and machines.  Theoretical concepts underlying an
integrated decision aiding simulator (IDEAS), which considers
numerical, linguistic, and pictorial entities and operations, are
applied to the impact assessment of an oil spill in the sea.
This example illustrates the potential applications of IDEAS for
environmental and water resources management.

Application

The Valdez accident in Alaska contributed to the increasing
worldwide interest in the environmental, social, and economic
impacts of oil spills.  Although many methods have been developed
over the years to assess those impacts, IDEAS may provide a rapid
modeling alternative using more intuitive representations for
simulation model variables and operations.  A generic oil spill
model implemented on a microcomputer, following the IDEAS
methodology, is described in the following paragraphs.

Problem Description

The assessment of the fate of an oil spill and its effects is a
complex process.  Typically the oil spilled is dispersed
immediately after discharge because of advection and spreading.
The area and thickness of the oil is then affected by
evaporation, sedimentation, and decay phenomena.  Advection is
influenced by dominant winds and currents.  Spreading results
from a dynamic equilibrium between the forces of gravity,
inertia, friction, viscosity, and surface tension.

Evaporation accounts for the loss of one third to two thirds of
the oil mass in a period of a few hours after the spillage.
Calculation of evaporation rate is difficult because it depends
on a number of factors, all of which may change with time.
However, one may say that the rate of evaporation from a thick,
cold slick under calm conditions will be orders of magnitude
slower than from a thin, warm slick under stormy conditions.
Oil may be transported to the bottom sediments through
hydrodynamic processes.  Oil consumed by the zooplankton also
reaches the sediments in the form of fecal pellets.

Photochemical oxidation and microbial action are the two most
important decay processes, depending on the amount of light and
temperature in the area, respectively.  Cold water with reduced
light, as in the Valdez case, can slow decay.  Warm water with
plenty of light can accelerate it.

Figure 1 shows the causal diagram extracted from this problem
description, while Table 1 lists the entities included in the
model.  The dictionaries for the pictorial and linguistic
entities are shown in Figure 2 and Table 2 respectively.  Note
that current and wind direction are treated as linguistic
entities with north, south, west, and east directions.

Model Results

The oil spill model was applied to a generic case, and a sample
of the model outputs using a monochromatic monitor is presented
in Figure 3.  This figure illustrates the wealth and
complementarity of information provided by models using
numerical, linguistic, and pictorial entities.  The latter
include plant and profile views of the oil spill and iconic
representations of the sea (fish) and shoreline (salt marshes)
wildlife.  In this example the oil spread is starting to reach
the shoreline, damaging the wildlife.  After 18 days, two thirds
of the total amount of oil spilled still remains.  Note that the
oil spread contour seems to have been randomly drawn because of
the wind direction and speed changes.

Future Developments

Further development of IDEAS will be directed toward the
improvement of pictorial entities representations and the
numerical, linguistic, and pictorial modeling methods.  For
example, other pictorial variables such as orientation, texture,
and value will be considered.  Efforts will be made to handle
additions or deletions of entities and relationships during the
modeling process.  Modeling of a contaminated estuary changing
from aerobic to anaerobic conditions and back to aerobic is an
example of the need to develop such rules.

IDEAS may be useful in combination with a geographic information
system (GIS).  Pictorial and linguistic methods included in the
proposed approach may be applied to solve problems requiring
geographical information.  The environmental impact analysis of
water resource infrastructures is a potential application of an
IDEAS-GIS system.

More substantial modeling developments are tied to the adoption
of parallel processing.  The application of cellular automata
formulations, which are ideally implemented on parallel
computers, instead of differential equations, is one of the
topics to be studied in this context.

In brief, a cellular automaton considers a discrete lattice of
sites, evolution in discrete time steps, each site taking a
finite set of possible values, the value of each site evolving
according to deterministic or nondeterministic rules, and the
rules for the evolution of a site depending only on a local
neighborhood of sites around it.  This alternative numerical
modeling approach may consider millions of cells.  Cellular
automata applied to water quality modeling thus may enable
particle tracking (each cell with a nonzero value could represent
a particle.)  This feature may be relevant if there is a need to
obtain spatial patterns of the pollutants.

Parallel processing will certainly benefit the use of linguistic
and pictorial models.  These models divide the continuum into a
number of discrete categories (i.e., high, medium, and low for a
linguistic variable; black, grey, and white for the color of a
pictorial object) and the operations are also scaled on a discrete
basis (i.e., absorption may be none, half, or total).

It is obvious that an aggregate discretization of a continuum,
which is the feasible option for standard computers, could lead
to significant error.  However, disaggregate discretization,
which is feasible with parallel processing, can certainly
minimize the errors of the discrete linguistic and pictorial
models, making them comparable to approximate numerical models,
with the advantage of being more intuitive.

Another significant development relates to the creation of
pictorial explanation tools which allow one to determine
pictorially the control options required to obtain a frame B from
a frame A.  The tuning of a television image is a quite similar
problem.  In this case there are few control variables (color,
sound, brightness, contrast) and they are independent.  The search
for the perfect image is thus relatively simple.  Similarly,
if one represents control variables as pictures with color, size,
shape, and position, to solve a simple problem one just changes
those properties until one goes from frame A to frame B.
Backtracking those changes allows one to define the required
strategy for turning A into B.  However, if there are many
control variables and their pictorial properties are not
independent, the problem may involve combinatorial optimization.

[The original article contained 36 references, 11 figures, and 5
tables, and notes that the work resulted in part from a Visiting
Fulbright Scholarship and a NATO Fellowship granted to the first
author for research on this topic at Cornell University.
Note that this e-mail version of S3 cannot duplicate figures
and tables as they appear in the version published in the
journal SIMULATION. JM]




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




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