Posted by ""Stefan.Grösser"" <stefan.groesser@web.de>
Dear Colleagues,
I am very interested in the grey zone questions about System Dynamics and other modeling technologies. The questions raised and partially answered in this thread deviated from the initial question about advantages and drawback of SD on other modeling approaches; it turned to the question about the low acceptance of SD in the scientific community and the question about when to apply different modeling approaches.
For me, the deviation from the original question about advantages and drawbacks of SD on other modeling approaches towards the question when to use SD indicates the strong connection between the two issues. Only when SD is applied in the correct problem situation, it can be used to the most benefit and thus SD can unfold its advantages.
However, in order to have a clear idea about when the SD-method is applicable, it is useful, especially for junior researchers who have not much experience in modeling, to learn also basics about other modeling approaches. By this comparison approach, it is possible to draw sharper distinctions for the fields of applications for each modeling approach. In other words, the fussy and implicit criteria when to apply SD will become more tangible. Within this thread, several scholars and practitioners have stated the same idea, e.g. Coyle, Laublé, and Gill. Lane shows with his ‘magnetic hexagon’-approach a possibility to reduce biases which arise due to method pre-selection during a consulting project (Lane, 1992, The road not taken, System Dynamics Review).
The problem about ‘How to select the appropriate research method’ is not unique to the simulation modeling science. I would argue that the question about method selection is important in every branch of sciences and practice. For the social science, I know it with a high degree of certainty. Andreas Diekmann, not a SD-modeler, but a famous Sociologist in the German speaking countries briefly addresses the question about method selection in his book ‘Empirical Research in the Social Science’ (Diekmann, 2005, Empirische Sozialforschung). According to him (my translation), ‘if it is the task of the Social Sciences to solve scientific or practical problems, then the problem should guide the choice of the method and not vice versa. This attitude assumes that the social researcher does not concentrate on one method only, but has the competence to handle the most important and most common instruments in his field of science (p. 18).’
At least for me, it is obvious why it is important for simulation modelers (like SD-modelers) to have a certain knowledge about several simulation approaches and not only about SD. The argument of Tony Gill (‘It takes time to be proficient in the use of tools. So most modelers settle for a tool set that works for them most of the time…’ (Tony Gill 24. Dec 2005, this thread) is also convincing, but can be weakened by the following: when the selection criteria for the available simulation approaches would be elicited from the vast amount of literature available and would be represented in a condensed way, the search and information costs could be reduced significantly. Thereby, it would be possible to create a battery of decision criteria for each simulation approach and thus a modeler could enlarge his or her toolbox within a relatively small amount of time.
I have to admit that my statement represents an academic point of view. The client in the business world normally does not care about the method used (this was stated by some members of this list). Here I am a bit critical and hope to trigger some feedback.
* Is it not the consultant’s task to explain, not in detail but in
principle, the clients the methods used, especially when logic and accessible white-box SD models have been used, in order to foster the understanding and implementation of the research results?
* Will this not build a connection of trust between client and
consultant?
* Will this not lead to follow-up projects?
* And finally, would this not lead to a better recognition of System
Dynamics in the business world with the overall effect of increasing demand for SD-projects?
In my opinion due to my limited but existing experience with top level managers, it is possible to convey the fundamentals of System Dynamics and its use within several hours, especially when they participate in a group model building session. However, the aforementioned is only successful, if the situation at hand fits to the field of application of the method, e.g. SD. Hence, it is evident why a sound understanding about the question ‘When shall I use SD and when not?’ is important.
Questions:
Does anyone know about publications which develop selection criteria, i.e., when to apply which method?
(One is, for instance, Flood, 1991, Creative Problem Solving, but I would argue this could be more detailed).
Does anyone know about a summary about simulation approaches, their fields of application, and their advantages and disadvantages?
(One is Meadows et al., 1985, The Electronic Oracle, is it very well written, but by no means comprehensive.)
I want to pick up the question Lorenz asked several weeks ago. (‘I was thinking, might one not propose a session upon these topics?, Tobias Lorenz, 6. Jan 2006, this thread). The System Dynamics website does not indicate if a workshop about multi-methodology or related simulation approaches (or whatever the label will be) will take place. Therefore the following question is posed:
* Is someone experienced, with regards to several simulation
approaches or other simulation approaches than SD, willing to moderate, or participate and contribute to such a multi-methodology workshop? Is this topic of interest for more members of the society (this list-server) or just for a small quantity?
Thanks for your time and I am hoping for feedback,
Best regards,
Stefan Groesser
University of St. Gallen, Switzerland
University of Berne, Switzerland
Posted by ""Stefan.Grösser"" <stefan.groesser@web.de>
posting date Thu, 6 Apr 2006 08:22:30 +0200
SD and other modelling approaches
-
Stefan.Grösser stefan.groesser w
- Junior Member
- Posts: 3
- Joined: Fri Mar 29, 2002 3:39 am
-
John Gunkler jgunkler sprintmail
- Member
- Posts: 30
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by ""John Gunkler"" <jgunkler@sprintmail.com>
Allow me to be both a little practical and a little cynical about this issue.
Before doing so, however, I will say that it makes sense for the simulation community to work on a clean set of criteria for choosing ""best"" approaches to types of problems. This could be helpful.
Having said that ...
Most problems I've seen solved (or ameliorated) in the real world (i.e., not
academia) were solved when someone looked at it, thought ""I know a way to attack this"" and then did something about it.
Notice, the someone did NOT say, ""I know THE BEST WAY to attack this.""
This, I believe, is in the best tradition of what Herbert Simon called finding ""satisficing"" solutions to problems. He argued that organizations were so complex that looking for ""optimizing"" solutions often just didn't make sense. I agree.
So, if we happen to have a toolkit (such as SD, or some other methodology) that can help people deal practically with the complexity they face in day-to-day work and living -- why not use it? Who cares if it's the ""best"" way of dealing with things? If our approach provides good bang for the buck, it's justified. It gets us another step down the road -- where, perhaps, other approaches can be employed.
This is NOT to say that we ought to use SD in situations where we know it is clearly not suitable. That's a somewhat different issue, I think. But to spend a lot of belly-button-contemplation time trying to choose the one best methodology is just not practical. You academics can do this -- and some day, maybe something useful will come from it (I hope so) -- but those of us working in the real world need to get on with it.
[I warned you about the cynicism.]
Posted by ""John Gunkler"" <jgunkler@sprintmail.com>
posting date Thu, 6 Apr 2006 10:28:48 -0500
Allow me to be both a little practical and a little cynical about this issue.
Before doing so, however, I will say that it makes sense for the simulation community to work on a clean set of criteria for choosing ""best"" approaches to types of problems. This could be helpful.
Having said that ...
Most problems I've seen solved (or ameliorated) in the real world (i.e., not
academia) were solved when someone looked at it, thought ""I know a way to attack this"" and then did something about it.
Notice, the someone did NOT say, ""I know THE BEST WAY to attack this.""
This, I believe, is in the best tradition of what Herbert Simon called finding ""satisficing"" solutions to problems. He argued that organizations were so complex that looking for ""optimizing"" solutions often just didn't make sense. I agree.
So, if we happen to have a toolkit (such as SD, or some other methodology) that can help people deal practically with the complexity they face in day-to-day work and living -- why not use it? Who cares if it's the ""best"" way of dealing with things? If our approach provides good bang for the buck, it's justified. It gets us another step down the road -- where, perhaps, other approaches can be employed.
This is NOT to say that we ought to use SD in situations where we know it is clearly not suitable. That's a somewhat different issue, I think. But to spend a lot of belly-button-contemplation time trying to choose the one best methodology is just not practical. You academics can do this -- and some day, maybe something useful will come from it (I hope so) -- but those of us working in the real world need to get on with it.
[I warned you about the cynicism.]
Posted by ""John Gunkler"" <jgunkler@sprintmail.com>
posting date Thu, 6 Apr 2006 10:28:48 -0500
-
Bill Braun bbraun hlthsys.com
- Member
- Posts: 29
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by Bill Braun <bbraun@hlthsys.com>
John, is it that dichotomous, academics versus practitioners? I straddle
both sides of the fence, and I find that one view of the world does a
nice job of informing the other - I'm more practical in the classroom
and more circumspect in the field.
If THE BEST WAY is synonymous with THE ONE TRUTH, then I'm quite on board
with you. On the other hand (so said the Fiddler on the Roof) if the
APPARENT BEST WAY is something that is teased out of a good, practical
discussion - itself a companion to dialogue (in the sense of dialogue
being a divergent conversation, searching out the broadest understanding
of the issue at hand, and discussion being a convergent conversation
searching the best option for action given the situation at hand) - then
I think there is an overlap that serves practitioners and academics
quite well, and indeed serves as a framework for being in conversation
with each other.
Related...does not satisficing feed into reductionistic thinking and
symptomatic solutions? I may be interpreting what you write way too
literally, but it seems more like ""ready, fire, aim"" than a systemic
look at things.
Bill Braun
Posted by Bill Braun <bbraun@hlthsys.com>
posting date Fri, 07 Apr 2006 09:52:33 -0400
John, is it that dichotomous, academics versus practitioners? I straddle
both sides of the fence, and I find that one view of the world does a
nice job of informing the other - I'm more practical in the classroom
and more circumspect in the field.
If THE BEST WAY is synonymous with THE ONE TRUTH, then I'm quite on board
with you. On the other hand (so said the Fiddler on the Roof) if the
APPARENT BEST WAY is something that is teased out of a good, practical
discussion - itself a companion to dialogue (in the sense of dialogue
being a divergent conversation, searching out the broadest understanding
of the issue at hand, and discussion being a convergent conversation
searching the best option for action given the situation at hand) - then
I think there is an overlap that serves practitioners and academics
quite well, and indeed serves as a framework for being in conversation
with each other.
Related...does not satisficing feed into reductionistic thinking and
symptomatic solutions? I may be interpreting what you write way too
literally, but it seems more like ""ready, fire, aim"" than a systemic
look at things.
Bill Braun
Posted by Bill Braun <bbraun@hlthsys.com>
posting date Fri, 07 Apr 2006 09:52:33 -0400
-
Justin Lyon simudyne gmail.com
- Junior Member
- Posts: 2
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by Justin Lyon <simudyne@gmail.com>
Stefan,
AnyLogic wrote an interesting paper on ABM, SD and DES. Check
it out on their web site.
It is also available for download off of the Simudyne web site.
http://www.simudyne.com/7.html
See my tipping point presentation on the simudyne site for more
thoughts on hybrid modeling and the future of systems modelling.
Warmly,
Justin Lyon
Posted by Justin Lyon <simudyne@gmail.com>
posting date Fri, 07 Apr 2006 12:08:36 +0100
Stefan,
AnyLogic wrote an interesting paper on ABM, SD and DES. Check
it out on their web site.
It is also available for download off of the Simudyne web site.
http://www.simudyne.com/7.html
See my tipping point presentation on the simudyne site for more
thoughts on hybrid modeling and the future of systems modelling.
Warmly,
Justin Lyon
Posted by Justin Lyon <simudyne@gmail.com>
posting date Fri, 07 Apr 2006 12:08:36 +0100
-
Alex Leus leusa tds.net
- Newbie
- Posts: 1
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by ""Alex Leus"" <leusa@tds.net>
I am in total agreement with these comments. I am a retired engineer and I remember those days of trying to optimize solutions. Good luck, the complexity of the real world does not afford that luxury. And the way things are going today with global competition etc. there is even less time to play around with optimization techniques.
Cheers,
Alex Leus
Posted by ""Alex Leus"" <leusa@tds.net>
posting date Fri, 7 Apr 2006 08:30:14 -0400
I am in total agreement with these comments. I am a retired engineer and I remember those days of trying to optimize solutions. Good luck, the complexity of the real world does not afford that luxury. And the way things are going today with global competition etc. there is even less time to play around with optimization techniques.
Cheers,
Alex Leus
Posted by ""Alex Leus"" <leusa@tds.net>
posting date Fri, 7 Apr 2006 08:30:14 -0400
-
Keith Linard klin4960 bigpond.ne
- Junior Member
- Posts: 7
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by ""Keith Linard"" <klin4960@bigpond.net.au>
If I was suffering a serious medical problem, I would not go to a medical specialist who knowledge was totally limited to one field. A oncologist hopefully knows a lot about about a range of cancers, radiology, pharmacology, etc, but also has sufficient skills in a broad range of medical diagnosis and treatment fields so s/he knows when to apply or call in other specialities. Similarly, were I hire a ""systems dynamics"" consultant, I would certainly expect them to be skilled in system dynamics and to be familiar with the broad range of systems disciplines. However, I would also expect them to have at least some knowledge of broader mathematical and operations research techniques, basic accounting and micro-economics.
I have seen many SDM solutions to problems which were more appropriately addressed by spreadsheets or econometric analysis; just as I have seen spreadsheet solutions which are more appropriately addressed by SDM.
Two very important Operations Research tools, which are now readily available to SD modellers through the Powersim software, are Monte Carlo simulation and optimisation.
Regarding optimisation: whilst I do not believe there is ever an 'optimum' solution to any problem, if only because there are inevitably multiple perspectives on the desirable outcomes, optimisation can be invaluable both in initialising the parameters for 'what-iffing' and in testing the implications of multiple shifts in those levers. In complex models (say, half a dozen interacting positive & negative feedback loops), as the number of user-available 'levers' rises above about 5 (& I have seen models with 30 or more levers!!!) one is faced with the situation of requiring an 'infinite number of monkeys' playing with the levers to understand desirable & undesirable combinations of lever positions. Alternatively, dispense with the monkeys and use Powersim.
The Monte Carlo simulation capability enables one to take into account uncertainty in parameters. Some years back I analysed the 'as constructed cost' of 90 civil engineering projects (from $500000 to $500m) compared with their 'pre-tender estimates' (based on final drawings and schedules of quantities). Whilst the mean was close to 1, the standard deviation was 30%+ ... and these estimates were much more accurate than most parameter estimates in our SD models (many of which seemed to be 'plucked out of the air' ... & are likely to be accurate to less than ±100%). Especially where interacting +ve/-ve feedback loops are finely balanced, differences in parameter values can lead to dramatic swings in outcomes. In my experience, explicitly taking uncertainty into account dramatically increases the credibility of the models with clients.
Keith Linard
134 Gisborne Road
Bacchus Marsh
Vic 3340
Posted by ""Keith Linard"" <klin4960@bigpond.net.au>
posting date Sun, 9 Apr 2006 10:04:06 +1000
If I was suffering a serious medical problem, I would not go to a medical specialist who knowledge was totally limited to one field. A oncologist hopefully knows a lot about about a range of cancers, radiology, pharmacology, etc, but also has sufficient skills in a broad range of medical diagnosis and treatment fields so s/he knows when to apply or call in other specialities. Similarly, were I hire a ""systems dynamics"" consultant, I would certainly expect them to be skilled in system dynamics and to be familiar with the broad range of systems disciplines. However, I would also expect them to have at least some knowledge of broader mathematical and operations research techniques, basic accounting and micro-economics.
I have seen many SDM solutions to problems which were more appropriately addressed by spreadsheets or econometric analysis; just as I have seen spreadsheet solutions which are more appropriately addressed by SDM.
Two very important Operations Research tools, which are now readily available to SD modellers through the Powersim software, are Monte Carlo simulation and optimisation.
Regarding optimisation: whilst I do not believe there is ever an 'optimum' solution to any problem, if only because there are inevitably multiple perspectives on the desirable outcomes, optimisation can be invaluable both in initialising the parameters for 'what-iffing' and in testing the implications of multiple shifts in those levers. In complex models (say, half a dozen interacting positive & negative feedback loops), as the number of user-available 'levers' rises above about 5 (& I have seen models with 30 or more levers!!!) one is faced with the situation of requiring an 'infinite number of monkeys' playing with the levers to understand desirable & undesirable combinations of lever positions. Alternatively, dispense with the monkeys and use Powersim.
The Monte Carlo simulation capability enables one to take into account uncertainty in parameters. Some years back I analysed the 'as constructed cost' of 90 civil engineering projects (from $500000 to $500m) compared with their 'pre-tender estimates' (based on final drawings and schedules of quantities). Whilst the mean was close to 1, the standard deviation was 30%+ ... and these estimates were much more accurate than most parameter estimates in our SD models (many of which seemed to be 'plucked out of the air' ... & are likely to be accurate to less than ±100%). Especially where interacting +ve/-ve feedback loops are finely balanced, differences in parameter values can lead to dramatic swings in outcomes. In my experience, explicitly taking uncertainty into account dramatically increases the credibility of the models with clients.
Keith Linard
134 Gisborne Road
Bacchus Marsh
Vic 3340
Posted by ""Keith Linard"" <klin4960@bigpond.net.au>
posting date Sun, 9 Apr 2006 10:04:06 +1000
-
wakeland pdx.edu
- Junior Member
- Posts: 3
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by wakeland@pdx.edu
To elaborate slightly on Keith's posting regarding Optimization and Monte Carlo (SD5894), one key benefit of Monte Carlo is to assess risk. What I describe below is by no means new information (it was call probabilistic SD or PSD many years ago), but perhaps it is worth repeating.
Assume that you are studying an inefficient business process. You have discovered a way to simplify the process that will cost you $X to implement.
Your SD model shows that the process improvements would save you $Y per year.
It is easy to use common financial measures (ROI, NPV, etc.) to determine if the improvement would be a good investment or not.
But what if your assumptions about key parameter or other aspects of the problem are wrong? How wrong do they have to be to turn the investment into a loser instead of a winner? Knowing that the NPV is $500K or that the ROI is 30% tells us nothing about the risk. What you might want to know is, ""What are odds that this process improvement will not provide sufficient benefit to justify the cost?""
By specifying probability functions for the major uncertain parameters, and running MANY, MANY cases (managed by the software), sampling from these probability functions for each case, we can find out what percentage of the cases end up with an ROI that is negative or below some minimum threshold.
This kind of knowledge complements what we can learn from the model without Monte Carlo (studying feedback loops, understanding leverage points, determining parameter sensitivity, etc.).
Risk analysis is not needed in all cases, but when it would help, it would be important for the SD analyst to know how to assess risk and for the SD software to facilitate this type of analysis.
Wayne Wakeland
Portland State Univ.
Portland, Oregon USA
Posted by wakeland@pdx.edu
posting date Sun, 09 Apr 2006 08:44:44 -0700
To elaborate slightly on Keith's posting regarding Optimization and Monte Carlo (SD5894), one key benefit of Monte Carlo is to assess risk. What I describe below is by no means new information (it was call probabilistic SD or PSD many years ago), but perhaps it is worth repeating.
Assume that you are studying an inefficient business process. You have discovered a way to simplify the process that will cost you $X to implement.
Your SD model shows that the process improvements would save you $Y per year.
It is easy to use common financial measures (ROI, NPV, etc.) to determine if the improvement would be a good investment or not.
But what if your assumptions about key parameter or other aspects of the problem are wrong? How wrong do they have to be to turn the investment into a loser instead of a winner? Knowing that the NPV is $500K or that the ROI is 30% tells us nothing about the risk. What you might want to know is, ""What are odds that this process improvement will not provide sufficient benefit to justify the cost?""
By specifying probability functions for the major uncertain parameters, and running MANY, MANY cases (managed by the software), sampling from these probability functions for each case, we can find out what percentage of the cases end up with an ROI that is negative or below some minimum threshold.
This kind of knowledge complements what we can learn from the model without Monte Carlo (studying feedback loops, understanding leverage points, determining parameter sensitivity, etc.).
Risk analysis is not needed in all cases, but when it would help, it would be important for the SD analyst to know how to assess risk and for the SD software to facilitate this type of analysis.
Wayne Wakeland
Portland State Univ.
Portland, Oregon USA
Posted by wakeland@pdx.edu
posting date Sun, 09 Apr 2006 08:44:44 -0700
-
Stefan Grösser stefan.groesser w
- Newbie
- Posts: 1
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by Stefan Grösser <stefan.groesser@web.de>
Dear all,
thank you folks for your replies to my recent post about SD
and other modelling approaches.
What did I obtain from your posts?
First, it could be that there is no (more or less comprehensive)
formal approach to answer the question in simulation/management
science about the selection of method for the subject under study.
Second, such a selection heuristic must firstly make sense in the
practical world and should be backed up by the scientific side. An
issue John Gukler raised (?You academics can do this -- and some day,
maybe something useful will come from it (I hope so) -- but those of
us working in the real world need to get on with it.?, Post No. SD5885)
seems to be a possibility to bridge the practice-science gap in the
field to obtain what Bill Braun described (?I find that one view of the
world does a nice job of informing the other - I'm more practical in
the classroom and more circumspect in the field?, Post No. SD5890). It
seems that a system dynamicists equipped with knowledge about several
simulation tools and skills in using some of them will be a more
accepted consultant since s/he could build confidence both in the chosen
method and the method?s results. For a SD-business consultant, I would
agree with the statement of Keith Linard (?I would ... expect them to be
skilled in system dynamics and to be familiar ? at least some knowledge of
broader mathematical and operations research techniques, basic accounting
and micro-economics, Post No. SD5894).
Third, almost no member of this list is interested in discussing the issue
about ?SD and other modelling approaches? in a separate workshop. Well,
perhaps only a few in the SD-society are interested in that topic at all.
Hopefully, I will run into some of them at the next conference in Njimegen,
perhaps in the Business Interest Group meeting.
Thanks again.
Best personal regards
Stefan Groesser
Research Associate, University of Berne, Switzerland
Ph.D. Candidate, University of St. Gallen, Switzerland
Posted by Stefan Grösser <stefan.groesser@web.de>
posting date Wed, 19 Apr 2006 13:15:24 +0200
Dear all,
thank you folks for your replies to my recent post about SD
and other modelling approaches.
What did I obtain from your posts?
First, it could be that there is no (more or less comprehensive)
formal approach to answer the question in simulation/management
science about the selection of method for the subject under study.
Second, such a selection heuristic must firstly make sense in the
practical world and should be backed up by the scientific side. An
issue John Gukler raised (?You academics can do this -- and some day,
maybe something useful will come from it (I hope so) -- but those of
us working in the real world need to get on with it.?, Post No. SD5885)
seems to be a possibility to bridge the practice-science gap in the
field to obtain what Bill Braun described (?I find that one view of the
world does a nice job of informing the other - I'm more practical in
the classroom and more circumspect in the field?, Post No. SD5890). It
seems that a system dynamicists equipped with knowledge about several
simulation tools and skills in using some of them will be a more
accepted consultant since s/he could build confidence both in the chosen
method and the method?s results. For a SD-business consultant, I would
agree with the statement of Keith Linard (?I would ... expect them to be
skilled in system dynamics and to be familiar ? at least some knowledge of
broader mathematical and operations research techniques, basic accounting
and micro-economics, Post No. SD5894).
Third, almost no member of this list is interested in discussing the issue
about ?SD and other modelling approaches? in a separate workshop. Well,
perhaps only a few in the SD-society are interested in that topic at all.
Hopefully, I will run into some of them at the next conference in Njimegen,
perhaps in the Business Interest Group meeting.
Thanks again.
Best personal regards
Stefan Groesser
Research Associate, University of Berne, Switzerland
Ph.D. Candidate, University of St. Gallen, Switzerland
Posted by Stefan Grösser <stefan.groesser@web.de>
posting date Wed, 19 Apr 2006 13:15:24 +0200
-
Colm Toolan subscriptions toolan
- Junior Member
- Posts: 9
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by ""Colm Toolan"" <subscriptions@toolan.de>
Stefan Grösser wrote >> Third, almost no member of this list is interested in discussing the issue about ""SD and other modelling approaches"" in a separate workshop. Well, perhaps only a few in the SD-society are interested in that topic at all.
I would agree with that evaluation - but that's probably much like asking a Mercedes driver to discuss the merits of driving a BMW
The guys at XJtek (creators of AnyLogic) have a beautifully agnostic view about all the flavours of simulation modelling - ABM, SD, DS, DES.
See
Andrei Borshchev
System Dynamics and Applied Agent Based Modeling (PDF: 1.2Mb) http://www.xjtek.com/files/papers/abmod ... er2005.pdf
Presentation for the Workshop “Agent Based Modeling: Why Bother?” held at the 23rd International Conference of the System Dynamics Society, July 17-21, 2004, Boston, MA, USA
Regards,
Colm Toolan, Business Architect
Germany
Posted by ""Colm Toolan"" <subscriptions@toolan.de>
posting date Fri, 21 Apr 2006 15:18:56 +0200
Stefan Grösser wrote >> Third, almost no member of this list is interested in discussing the issue about ""SD and other modelling approaches"" in a separate workshop. Well, perhaps only a few in the SD-society are interested in that topic at all.
I would agree with that evaluation - but that's probably much like asking a Mercedes driver to discuss the merits of driving a BMW
The guys at XJtek (creators of AnyLogic) have a beautifully agnostic view about all the flavours of simulation modelling - ABM, SD, DS, DES.
See
Andrei Borshchev
System Dynamics and Applied Agent Based Modeling (PDF: 1.2Mb) http://www.xjtek.com/files/papers/abmod ... er2005.pdf
Presentation for the Workshop “Agent Based Modeling: Why Bother?” held at the 23rd International Conference of the System Dynamics Society, July 17-21, 2004, Boston, MA, USA
Regards,
Colm Toolan, Business Architect
Germany
Posted by ""Colm Toolan"" <subscriptions@toolan.de>
posting date Fri, 21 Apr 2006 15:18:56 +0200
-
Bipin Chadha bchadha coensys.com
- Junior Member
- Posts: 2
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by Bipin Chadha <bchadha@coensys.com>
To continue along Colm's comments - What we are finding is that for system
level multi-disciplinary problems, no one modeling approach is sufficient.
I am very interested in the topics of SD and other modeling approaches and
hybrid modeling approaches, but won't be at the conference.
Thanks,
Bipin Chadha
Coensys, Inc.
Posted by Bipin Chadha <bchadha@coensys.com>
posting date Sat, 22 Apr 2006 09:48:08 -0400
To continue along Colm's comments - What we are finding is that for system
level multi-disciplinary problems, no one modeling approach is sufficient.
I am very interested in the topics of SD and other modeling approaches and
hybrid modeling approaches, but won't be at the conference.
Thanks,
Bipin Chadha
Coensys, Inc.
Posted by Bipin Chadha <bchadha@coensys.com>
posting date Sat, 22 Apr 2006 09:48:08 -0400
-
Geoff McDonnell gmcdonne bigpond
- Junior Member
- Posts: 10
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by ""Geoff McDonnell"" <gmcdonne@bigpond.net.au>
At the Health Policy SIG WIKI of the SD Society there are several
pages devoted to this area, particularly the use of both classicial
difference equation SD and agent based or ""emergent"" SD.
http://www.hpsig.com/index.php?title=Simulation_Methods
http://www.hpsig.com/index.php?title=Em ... m_Dynamics
http://www.hpsig.com/index.php?title=IS ... _Bother%3F
Feel free to contribute to this online collaboration if you cant
find enough people to hold a workshop.
BTW the Thursday workshop at the ISDC2005 Boston conference on Agent
Based Modeling Why bother? was very well attended.
regards
geoff
Dr Geoff McDonnell
Director Adaptive Care Systems
Posted by ""Geoff McDonnell"" <gmcdonne@bigpond.net.au>
posting date Sun, 23 Apr 2006 00:30:23 +1000
At the Health Policy SIG WIKI of the SD Society there are several
pages devoted to this area, particularly the use of both classicial
difference equation SD and agent based or ""emergent"" SD.
http://www.hpsig.com/index.php?title=Simulation_Methods
http://www.hpsig.com/index.php?title=Em ... m_Dynamics
http://www.hpsig.com/index.php?title=IS ... _Bother%3F
Feel free to contribute to this online collaboration if you cant
find enough people to hold a workshop.
BTW the Thursday workshop at the ISDC2005 Boston conference on Agent
Based Modeling Why bother? was very well attended.
regards
geoff
Dr Geoff McDonnell
Director Adaptive Care Systems
Posted by ""Geoff McDonnell"" <gmcdonne@bigpond.net.au>
posting date Sun, 23 Apr 2006 00:30:23 +1000
-
Joel Rahn jrahn sympatico.ca
- Junior Member
- Posts: 19
- Joined: Fri Mar 29, 2002 3:39 am
SD and other modelling approaches
Posted by Joel Rahn <jrahn@sympatico.ca>
As chairman of the 'Alternative Methodologies' thread at the last few
conferences, I beg to differ with Messrs. Grösser and Toolan. Sure, a
thread is not a workshop but it is an outlet for discussion about other
approaches to modelling dynamic systems. To avoid turning the conference
into an OR Fest, my main concern in the thread is to feature papers that
illuminate 'dynamic systems' i.e. those where 'behavior over time' is the
focus. Not just variations over spatial dimensions, not just stochastic
variations, just variations over time (with the other stuff so beloved of
ABM and DES mavens if it tells us something new about the dynamics, its
causes or consequences).
Joel Rahn
Posted by Joel Rahn <jrahn@sympatico.ca>
posting date Sat, 22 Apr 2006 08:58:05 -0400
As chairman of the 'Alternative Methodologies' thread at the last few
conferences, I beg to differ with Messrs. Grösser and Toolan. Sure, a
thread is not a workshop but it is an outlet for discussion about other
approaches to modelling dynamic systems. To avoid turning the conference
into an OR Fest, my main concern in the thread is to feature papers that
illuminate 'dynamic systems' i.e. those where 'behavior over time' is the
focus. Not just variations over spatial dimensions, not just stochastic
variations, just variations over time (with the other stuff so beloved of
ABM and DES mavens if it tells us something new about the dynamics, its
causes or consequences).
Joel Rahn
Posted by Joel Rahn <jrahn@sympatico.ca>
posting date Sat, 22 Apr 2006 08:58:05 -0400