Path: utzoo!mnetor!uunet!husc6!purdue!gatech!uflorida!simulation From: simulation@ufl.edu (SIMULATION DIGEST) Newsgroups: comp.simulation Subject: SIMULATION MODELING AND ANALYSIS Message-ID: <15088@uflorida.cis.ufl.EDU> Date: 21 Apr 88 21:01:36 GMT Sender: fishwick@uflorida.cis.ufl.EDU Reply-To: simulation@ufl.edu Lines: 194 Approved: fishwick@fish.cis.ufl.edu From fishwick Thu Apr 21 16:08:29 1988 To: fishwick < ~/sim/header1 Subject: testing Volume: 1, Issue: 2, Thu Apr 21 17:01:27 EDT 1988 +----------------+ | CONTENTS | +----------------+ (1) Simulation at Politechnical University of Catalonia (2) In the Boxing Ring: Simulation vs. Closed-Form ########################################################################## Date: Tue, 19 Apr 88 19:51 N From: Subject: SIMULATION DIGEST To: simulation@ufl.edu X-Original-To: sim_topic My best wishes to the "comp.simulation" subscribers and moderator. I just want to introduce a few ideas: 1) We think that it would be interesting to know the people or organizations subscribed to the simulation news group. i.e. to have a short 'curriculum' of them (name, ordinary mail address, e-mail addresses, research activities, topics of interest, simulation tools used or being developed,...). So, I'll introduce myself: *- Name : Antoni GUASCH O-mail : Cybernetics Institute Polithechnical University of Catalonia Diagonal 647 08028 Barcelona - SPAIN Phone : (343) 2492842 E-mail : earafael@ebrupc51.bitnet *- Sim. research activities : - Research in continuous or piecewise continuous, hierarchical simulation languages. - Object oriented languajes and expert systems in modelling and simulation. *- Topics of interest : - Real time simulation - Expert systems - Object oriented programming *- Simulation tools used : ICDSL (CSSL-like) and SIMSCRIPT *- Simulation tools being developed : MUSS (hierarchical simulation environment) *- Some recent activities: - Organisation of the IMACS International Symposium on "AI, Expert Systems and Languages in Modelling and Simulation". Barcelona, June 1987 - A study on simulation and modelling: "future satellite simulators" for the European Space Agency. *- Miscellaneous : The Institute has three main research areas: Automatic Control, Bioengineering and Robotics. Simulation is a master piece to deal with complex electrical networks and bioengineering models. The simulation group provides support and tools for the Institute modelling and system analysis. It also gives courses on Simulation in the Politechnical University of Catalonia. The simulation group is composed by 4 people: Prof. Rafael M. HUBER (in charge of the simulation group), two undergraduate students (Carles MUNTADA and Jose Manuel MARTIN) and myself (Associated Prof.). 2) Designing and coding a new robust simulation 'environment' is not an easy task. It involves aspects such as: - formal language definition - compiler-compiler techniques - numerical integration - state and time events management - sorting or event scheduling problems - model or experiment data bases - graphical or numerical postprocessors - drivers for different peripherals - documentation management For a small group its quit difficult to cover all of these aspects. Moreover, a lot of time is spent designing modules or routines which have been already coded by somebody else. Thus, I feel that it would be very interesting to create macro-groups (M-GROUPS) interested in joining efforts and developing specific tools. I wonder if this is possible. I will appreciate ideas and comments about this point. Bye, Toni --//-- ########################################################################## ########################################################################## Date: Thu, 21 Apr 88 16:52:03 EDT From: Paul Fishwick To: simulation@ufl.edu Subject: static analysis vs. simulation How about some controversy?! I just recently got back from the SCS Eastern Simulation Conference in Orlando and was interested in several of the topics that arose from panel discussions and talks. In particular, Ramana Reddy made a comment during his plenary talk that I found very interesting...he questioned the notion of "simulation being the *last* resort" in solving hard problems. I agree with Ramana that this notion of "last resort" is often incorrect. Perhaps many of you have your own thoughts about whether simulation as a problem solving method should be considered "a last resort." Why not a first resort! There are many sub-issues here...I'll list some that I can think of: (disadvantages are identified with '-' and advantages are identified with '+'): METHOD: CLOSED-FORM SOLUTION (i.e. static analysis) (+) often very fast since it involves equation solving. Good example is queueing models (with many assumptions!). (-) you obtain a steady state solution (which might be all that is required!). METHOD: SIMULATION (i.e. dynamic analysis) (+) extremely flexible. Can handle very large and complex scenarios. you obtain a transient solution. (-) slower than closed-form in some circumstances. Also, many claim that it "costs more." Here are some debatable notes: 1. Remember when calculators were larger than bookcases? At that time, people probably said...use calculators as a last resort----but first, see if you can solve the problem using your handy dandy slide rule. Perhaps this issue of "simulation as a last resort" falls into the same category. As hardware becomes cheaper and faster--there will be less and less problems that will necessitate closed form solutions. What is the percentage of problems now that require a closed-form approach as a first try in problem solving? 30%, 2% ? Are there certain domains or application areas that favor a closed-form approach? 2. As companies are coming out with visually (and object) oriented simulation products, will simulation always *cost* more than picking out formulas from someone's textbook on closed-form problem solving methods? What exactly are these "costs"? 3. With closed-form methods, many assumptions must often be made to make the problem solving process mathematically tractable. When we make these assumptions, are we diluting our analysis and results? 4. Perhaps programs with combinations of simulation and closed-form methods are best? How 'bout some lively discussion? Paul Fishwick University of Florida ########################################################################## ######################## END OF SIMULATION DIGEST ########################