I have decent quantitative modeling skills gained from a MSc in Quantative Finance and some subsequent modelling work (synthetic credit derivatives, VaR models for the most part, but some interesting stuff with asian options as well). I have a undergrad degree in history, and can read french, spanish and english.

Where should I start in terms of trying to learn what kind of modeling is being done and what the state of the art is? Please assume that I have the necessary background to understand the math/OR algorithms and any polisci/history.

Some of the ones listed here should probably be on your list:
http://www.dupuyinstitute.org/booksfs.htm

Unfortunatly it looks like [u]Numbers, Predictions and War[/u] is out of print.

Are you intereste in computer models or abstract models?

Stochastic or deterministic or both? Especially in the derminstic case you should know a little chaos theory. In the stocastic case a fare amount of statistics (Keltaon and Law's book is the standard there).

Just military models or economic or something else or all of the above?

I haven't really thought about doing this kind of modeling as work...is there work in the field? I currently create Monte Carlo simulations of the probability of bad and good things happening in the world of finance. I am currently building a model for emerging markets instability and thought that there might be some interesting things in the toolcase of people doing military history simulations...

As this is a discussion forum affiliated with the Dupuy Institute, yes, I am interested in military models. While the denizens might have interesting perpectives in value at risk and modeling asian options, I think i'll continue having most of those discussions here in New York.

As for the math. When you are doing repeated runs of a stochastic model it turns out you can control a number of things that you can't in conventional statistics. If you take advantage of this you can greatly reduce the variance of you results for a given number of runs.

There is at least one group run by Mr Dunnigen:
milgames@yahoogroups.com
that you might want to look at.

Georgia Tech used to have an annual class/seminar. There are several other seminars on modeling througout the country and overseas. Their proceddings would be worth investigating.

Part of the problem is a lot of the work is specific to a particular sim. Or the other extreme is quick and dirty one off sim/model. The military tends to use models for several very distinct purposes (ie. training, mission rehearsal, system effectiveness, and engineering level models) and people tend to work in one area.

I guess I would suggest looking up Dunnigan's work and seeing if any of his books sound useful. For the math part like I said Kelton and Law (http://www.mhhe.com/engcs/industrial/lawkelton/). Also look for proceedings from various seminars and organizations such as:
http://www.scs.org/
http://www.acims.arizona.edu/
http://www.msiac.dmso.mil/
http://www.modelingandsimulation.org/

Here's an online text I haven't checked it out yet though:
http://home.ubalt.edu/ntsbarsh/simulation/sim.htm

A deterministic model doesn't have any stochastic processes in it. IE same inputs result in identical output. Such models can expibit chaotic behavior however in that a minor change in input can produce a large and unexpected (at least until examined in detail)change in output.

My background is in modeling. I'm not really interested in 'learning' about modeling; i already do it for work. What i am specifically interested in is the specific field knowledge.

That online text is really really basic, i think i'd recommend it--based on a five minute skim--to potential master's level operations research students...

Its interesting to me that the models haven't changed much in twenty years. Actually, its astounding, flabbergasting and befuddling.

If you are curious why this is so, simply check out the differences between the vasicek model and the libor model of term structure, or what's happened with Value at Risk, or the growth of complexity in credit derivatives.

Are there journals dedicated to this field?

Data is where i anticipated problems. Most of the work of developing finance models is the data, the math and the programming are always the easy part. Find the data, clean the data, ...

It is interesting to me, the level of simplicity at work here. We have found (20 year old model) robust mathematics for warfare because the inputs are very small compared to economics.

Relatively few actors and relatively few inputs with relatively simple outcomes. Tactical victory ends the simulation and results are applied to the next closed system, engagement level scenario.

I was going to ask about your processes because I am extremely interested in providing a similar simulation service, but the reverse.

Taking all of the other data, the "opposite" of your direction of focus, the relative factors leading to a peace prediction service. Hope is to provide strategies and relative success determination of different methods of peace negotiation and examining reasons for the breakdown of peaceful existence.

I believe it can be quantified in simple terms based upon readily available data in the pre and post war periods that you have already researched.

As a start, I would like to build a tool for the assessment of systems confined solutions that prevented or reduced the probability of unpeaceful activity.

Judging the avoidance of combat as a 'victory'.

What are your thoughts about the volume of database inputs required and structure of the project?

This would ultimately become integrated into a world visualising tool (NASA World Wind?) and made available to policy makers.

Sorry, but I have not been monitoring the board because of work. I can answer this question for you....and look forward to doing so, but the answer will be dribbling in over the course of several days. Please be patient.

quote:

---

Originally posted by cf:
[Snips]
Where should I start in terms of trying to learn what kind of modeling is being done and what the state of the art is?

---

I apologise in advance for the fact that this post isn't going to be very helpful, but that's because the published literature of combat modelling is horribly, horribly thin.

There is (and I would dearly love to be proven wrong on this) at the moment no book available to act as a guide on "how to go combat modelling". A friend of mine who has worked at Fort Halstead for years has long felt the lack of such a book for training new Operations Analysts, and has started writing an introduction to direct-fire combat modelling; whether he will be able to find a publisher is another matter.

Having said that, there a a few books that will give you part of the picture which may be worth trying to track down.

A general (non-technical) history of the military use of simulations can be obtained from Andrew Wilson's "War Gaming" (originally "The Bomb and the Computer"), now dated but well-written and interesting.

More recent (and for my money less well-written) such general views can also be obtained from Peter Perla's "The Art of War Gaming" or Thomas Allen's "War games".

None of these will tell you anything about the maths involved, though. For that, the best thing I know is "Applied operations research: Examples from defence assessment", by R W Shephard, D A Hartley, P J Haysman, L Thorpe and M R Bathe (Plenum Press, 1988). This is intended as a course textbook, and consists of a number of "potted" problems with model answers and instructors' notes. It covers a good breadth of typical applications of OA, some recognisably modelled on real life studies.

There is an excellent book on Naval Operations Research whose title and author currently evade me, as my copy is on loan to the aforementioned pal from Fort Halstead.

A final book worth a look, and which I think might appeal to a quant, is Stephen Biddle's "Military Power: Explaining Victory and Defeat". Excellent though the book is, I wouldn't say it is at all typical of what most combat modellers get up to.

I doubt that it is possible to give a sensible picture of "the state of the art" of combat modelling, because it is very varied and very fragmented. My impression is that there a lot of fundamental questions left unanswered, especially in the areas of suppression, fatigue, the direction of human attention, and human situational assessment and decision-making. Worse, there seems to be little effort being made to tackle these basic questions, either because the people who fund the research are interested in equipment programmes rather than human factors, or because a very great deal of modelling effort these days is wasted on feeding the monster of DIS/HLA and 3-D visualisation instead of concentrating on fundamental modelling questions.

All the best,

The starting point for the operations research community is Morse and Kimball, "Methods of Operations Research".

This book is usually kept in publication and dates back to 1950 (original claissfied report was from 1946). It is the starting point for describing American Operations Research, which of course, was learned from the British (who call it Operational Research).

While this book does not contain any specifics on combat modeling (because this was not being done yet), it does include a description of basic methods, including probability of hit and Lanchester equations.

The first campaign model developed was ATLAS, in the early 1960s. This was basically a force ratio/firepower score model developed from primarily the Leavenworth wargaming rules.

Therefore, the next important book on the modeling is the Leavenworth Wargaming rules, say the 1964 or 1967 edition. These are FM 105-5 "Manuever Control". It includes many concepts that were first used in various wargames, including ATLAS and as such is probably the single most influential document on combat modeling. The fact that it appears that none of the relationships or figures used in these manuals were established by any identifiable analytical study did not stop people from using them.

By the early 1970s, all basic forms of combat models were in use

1) Monte Carlo Simulations
2) Force Ratio/Firepower Score models
3) Lanchester Equations
4) Hierarchy of Model structures

At this point, there is a book out that general describes these various combat models and includes some descriptions. This is:

Martin Shubik and Garry D. Brewer, Models, Simulations, and Games – A Survey (R-1060-ARPA/RC, RAND, Santa Monica, CA, May 1972).

Try:

This is:

Catalog of War Games, Training Games, and Combat Simulations (Deputy Under Secretary of the Army (Operations Research), Washington, DC, 1983).

If you find a copy, please make me one. We do not have a copy of this on our shelves.

Also of interest would be:

For example Catalog of Simulation Models and Wargames Used for Unit and Leader Training (Training and Performance Data Center, Orlando, FL., 2nd Ed., 1987)

Paul K. Davis, Donald Blumenthal, The Base of Sand Problem: A White Paper on the Sate of Military Combat Modeling (N-3148-OSD/DARPA, RAND, Santa Monica, CA., 1991).

Background Paper on Distributed Interactive Simulation of Combat (Office of Technology Assessment, 1995).

You will see me listed as a researcher on that one.

It may be available on-line (as is the case for several RAND reports).

Trevor N. Dupuy, Attrition: Forecasting Battle Casualties and Equipment Losses in Modern War (HERO Books, Fairfax, VA, 1990).

This is available by order from us.

Paul K. Davis “The Influence of Trevor Dupuy’s Research on the Treatment of Ground Combat in RANDs RSAS and JICM Models” in The International TNDM Newsletter (Volume II, Number 4, December 1998)

This is available on-line on this website.

Patrick Allen, Situational Force Scoring: Accounting for Combined Arms Effects in Aggregate Combat Models (RAND Note N-3423-NA, RAND, Santa Monica, CA, 1992).

Col. T. N. Dupuy, “Criticism of Combat Models Cite Unreliability of Results,” (Army, March 1985, page 16) and George Miller & Seth Bonder, “Writers of ARI Study Respond to Critique of Approach” (Army, June 1985, page 5) and Col. T. N. Dupuy, “Rebuttal Rebutted,” (Army, September 1985, page 6).

This is a public exchange of letters addressing modeling methods and validation.

See James Ong and Michael F. Ling, Using the Joint Integrated Contingency Model for Campaign Analysis (DSTO-TR-1307, DSTO Electronics and Surveillance Research Laboratory, Edinburgh, Australia, May 2002).

This is an Australian description of RAND's JICM and is available on-line.

F. W. Lanchester, Aircraft in Warfare: The Dawn of the Fourth Arm (Lanchester Press Incorporated, Sunnyvale, CA. 94086, 1995). Originally published in 1916 by Constable & Co., London, England.

These are the original Lanchester equations. Note that he states up front that he does not think that they apply to ground combat (not that this stopped them being used for such).

J. H. Engel, “A Verification of Lanchester’s Law,” (Operations Research, Vol. 2, May 1954).

This allegedly verified Lanchester equations.

There have been more than a half-dozen efforts that have refuted all or part of Lanchester for use as a combat modeling method, including:

1)D. Willard, Lanchester as a Force in History: An Analysis of Land Battles of the Years 1618-1905 (RAC-TD-74, Research Analysis Corporation, 1962).

2) H. K. Weiss “Lanchester-Type Models of Warfare,” from Proceedings from the First International Conference of Operational Research, pages 82-98, ORSA (1957);

3) H. K. Weiss, “Combat Models and Historical Data: The U.S. Civil War,” Operations Research, volume 14, pages 750-790 (1966);

4) R. L. Helmbold “Some Observations on the Use of Lanchester’s Theory for Prediction” (Operations Research, Volume 12, pages 778-781, 1964);

5) W. Fain, J. B. Fain. L. Feldman and S. Simon, Validation of Combat Models Against Historical Data (Professional Paper No. 27, Center for Naval Analyses, Arlington, VA, 1970).

6) Janice B. Fain, “The Lanchester Equations and Historical Warfare: An Analysis of Sixty World War II Land Engagements.” Combat Data Subscription Service (HERO, Arlington, VA, Spring 1975);

7) D. S. Hartley and R. L. Helmbold, "Validating Lanchester's Square Law and Other Attrition Models," in Warfare Modeling, J. Bracken, M. Kress, and R. E. Rosenthal, ed., (New York: John Wiley & Sons, 1995) and originally published in 1993;

8) Jerome Bracken, “Lanchester Models of the Ardennes Campaign in Warfare Modeling (John Wiley & sons, Danvers, MA, 1995);

9) R. D. Fricker, "Attrition Models of the Ardennes Campaign," Naval Research Logistics, vol. 45, no. 1, January 1997;

10) S. C. Clemens, “The Application of Lanchester Models to the Battle of Kursk” (unpublished manuscript, May 1997);

11) 1LT Turker Turkes, Turkish Army, "Fitting Lanchester and Other Equations to the Battle of Kursk Data," Dissertation for MS in Operations Research, March 2000;

12) Captain John Dinges, U.S. Army, “Exploring the Validation of Lanchester Equations for the Battle of Kursk,” MS in Operations Research, June 2001;

13) Tom Lucas and Turker Turkes, “Fitting Lanchester Equations to the Battles of Kursk and Ardennes," Naval Research Logistics, 51, February 2004, pp. 95-116;

14) and Thomas W. Lucas and John A. Dinges, “The Effect of Battle Circumstances on Fitting Lanchester Equations to the Battle of Kursk,,” forthcoming in Military Operations Research.

Needless to say, we do not think that Lanchester has much validity except for certian tactical modeling situations.

Francis P. Hoeber, Military Applications of Modeling: Selected Case Studies (Gordan and Breach Science Publishers, New York, 1981)

Dorothy K. Clark, Casualties as a Measure of the Loss of Combat Effectiveness of an Infantry Battalion (Operations Research Office, Johns Hopkins University, 1954).

This paper is the first and most important "Breakpoints" study.

HERO report number 117, Forced Changes of Combat Posture (Breakpoints)

This is the only other major "Breakpoints" study

HERO Report number 36, Opposed Rates of Advance of Large Forces in Europe (ORALFORE) (1972)

Robert Helmbold, “Rates of Advance in Historical Land Combat Operations,” CAA-RP-90-1 (US Army Concepts Analysis Agency, Bethesda, MD, 1990).

Trevor N. Dupuy, Understanding War: History and Theory of Combat (Paragon House Publishers, New York, 1987)

Below is a list of various criticisms of modeling efforts:
1) Robert McQuie “Military History: a Mathematical Analysis,” Military Review (May 1970),
2) John Honig, et al., Report of the Army Models Committee (US Army, 1971),
3) Franz Uhle-Wettler “Computer Supported Studies and Military Experience,” Soldat und Technik (April 1975)
4) Lawrence J. Low, Theater-Level Gaming and Analysis Workshop for Force Planning, September 1977 (Report to Office of Naval Research, 1980),
5) Brian R. McEnany “Uncertainties and Inadequacies in Theater Level Combat Analysis” (paper delivered at 16th U.S. Army Operations Research Symposium, 12-14 October 1977)
6) David C. Hardison, et al., Review of Army Analysis (U.S. Army, April 1979),
7) Appendix A: “Some Critiques of Military Modeling” in John Battilega and Judith Grange. The Military Applications of Modeling. (Wright-Patterson AFB, OH: Air Force Institute of Technology Press, 1979) 8) Report to the Congress: Models, Data and War: A Critique of the Foundation for Defense Analysis (Comptroller General of the U.S., 1980),
9) J. P. P. Wood “Very Grave Suspicion” Royal United Services Institute Journal, (March 1982),
10) Anonymous “In Pursuit of the Essence of War,” Army, January 1984
11) Col. T. N. Dupuy, “Criticism of Combat Models Cite Unreliability of Results” (Army, March 1985, page 16)
12) Col. T. N. Dupuy, “Rebuttal Rebutted” (Army, September 1985, page 6)
13) Col. T. N. Dupuy, “Can We Rely Upon Computer Combat Simulations?” (Armed Forces Journal International, August 1987

Or in response to number 11 above:

Walt W. Hollis, “Yes We Can Rely on Computer Combat Simulations” (Armed Forces Journal, October 1987).

Report of the Model Input Data and Process Subcommittee of the Casualty Estimation Steering Committee, Prepared by Office, Deputy Under Secretary of the Army Operations Research for Personnel Plans and Systems Directorate, Department of the Army, Office, Deputy Chief of Staff for Personnel, Washington D.C., 20310, January 1986

Paul Berensen, Memorandum for Mike Bauman, Director, TRADCOC Analysis Center; Edgar Vandiver III, Director, USA Concepts Analysis Agency; John McCarthy, Director, US Army Material Systems Analysis Activity dated 21 January 1997.

This memorandum is presented on our website in our issue on Validation in the TNDM International Newsletter.

Robert L. Helmbold, "Personnel Attrition Rates in Historical Land Combat Operations: Some Empirical Relations Among Force Sizes, Battle Durations, Battle Dates, and Casualties," Bethesda MD: US Army Concepts Analysis Agency, 1995

Dr. Rob Alexander, SYST683 Lecture Notes #3. “Models, Gaming, and Simulation: Physical Models of Attrition.” Fairfax VA: George Mason University, 2004 (available at http://classweb.gmu.edu/ralexan3/SYST683/LectureNotes/CM-3.ppt).

John Battilega and Judith Grange. The Military Applications of Modeling (Wright-Patterson AFB, OH: Air Force Institute of Technology Press, 1979).

Dupuy, Trevor. Numbers, Predictions, and War: Using History to Evaluate Combat Factors and Predict the Outcome of Battles. New York: Bobbs-Merrill Company, Inc. 1979;

Testimony of Col. T.N. Dupuy, USA, Ret., Before the House Armed Services Committee, 13 Dec 1990.” The Dupuy Institute File I-30, “Iraqi Invasion of Kuwait.”

Brewer, Gary D., and Martin Shubik. The War Game. Cambridge MA: Harvard University Press. 1979

Defense Modeling and Simulation Office. Department of Defense Modeling and Simulation (M&S) Master Plan. Washington D.C.: Office of the Undersecretary of Defense for Acquisition and Technology, October 1995 and 1996.

Dr. James G. Taylor (author of Lanchester Models of Warfare in two volumes, published by the Operations Research Society of America, Arlington, Virginia, in 1983)

And probably a few others important reports that I do not have at my fingertips.

Hopefully, this helps.

quote:

---

Originally posted by Chris Lawrence:
The first combat model built was Carmonette in 1953.

---

The first computerised combat model, yes. But people had been using manual models (wargames) for a long time, including such memorable efforts as Admiral Hood's games using peach pits and plum stones to represent battlefleets (with the French flagship always represented by a lemon), and Fletcher Pratt's Naval Wargame, popular in the 1930s and resurrected at the annual Conference of Wargamers at Knuston Hall this year.

All the best,

John.

quote:

---

Originally posted by Chris Lawrence:
[Snips]
By the early 1970s, all basic forms of combat models were in use

1) Monte Carlo Simulations
2) Force Ratio/Firepower Score models
3) Lanchester Equations
4) Hierarchy of Model structures

---

Of course I would classify models on a completely different basis, but that might lead us into the sad necessity of devising some classification of model classification systems...

I think the category of game-theory models is probably important enough to be included among that lot, to include extensions of game theory such as the theory of moves and drama theory. Nigel Howard's book on the latter, "Confrontation Analysis: How to Win Operations Other Than War" might still be available as a freebie from the CCRP publication programme.

It might be worth noting that in the simulation modelling community, "Monte Carlo" is a term one hears rarely these days, "stochastic" being the usual word. Where "Monte Carlo" is used, if often has the restrictive sense of meaning a static model, rather than a proper entity-based discrete-event model.

Likewise, "Hierarchy of model structures" is not a term I hear used, but I take to mean much the same as "multi-modelling" (although hierarchy is not neessarily implied).

I'm wondering, Chris, what difference do you intend to be understood between catergories 2 and 3? Is cat 2 what we Brits would call Historical Analysis (HA) modelling, i.e. getting best fits to historical data?

All the best,

quote:

---

Originally posted by John D Salt:
It might be worth noting that in the simulation modelling community, "Monte Carlo" is a term one hears rarely these days, "stochastic" being the usual word. Where "Monte Carlo" is used, if often has the restrictive sense of meaning a static model, rather than a proper entity-based discrete-event model.

---

I've pretty much seen the word used interchangably.

quote:

---

I'm wondering, Chris, what difference do you intend to be understood between catergories 2 and 3? Is cat 2 what we Brits would call Historical Analysis (HA) modelling, i.e. getting best fits to historical data?

---

Category two are all those models that use some scoring system for weapons and usually then compares the ratio of firepower/combat power, etc. The QJM falls into this category, as does RAND's JICM (with SFS) and many other models.

Lanchester equations models are those that use the differential calculus to somehow or the other calculate attrition. This is certianly the entire Vector series of models, JTLS, etc.