Combat models

Fig. 8.26 Three sample rules in Woodcock, et. al. s CA combat model.

The payoff of an agent-based approach is a radically new way of looking at fundamental issues of land combat. Models such as these are emphatically not to be used for prediction rather they are best used to enhance understanding. Specifically, agent-based models of combat are designed to help military theorists... 

Chapter 8 describes a number of generalized CA models, including reversible CA, coupled-map lattices, quantum CA, reaction-diffusion models, immunologically motivated CA models, random Boolean networks, sandpile models (in the context of self-organized criticality), structurally dynamic CA (in which the temporal evolution of the value of individual sites of a lattice are dynamically linked to an evolving lattice structure), and simple CA models of combat. 

In Woodcock, et al. s model, each combatant - or automaton - is endowed with a set of rules with which it can perform certain tasks. Rules are of four basic varieties ... 

In this section we introduce a mobile CA model of land combat called EINSTein, developed at the Center for Naviil Analyses for the US Marine Corps. We include a discussion of this model here becau.se it is an interesting blend of CA-like local dynamics and agent-based modeling techniques. 

In 1914, F. W. Lanchester introduced a set of coupled ordinary differential equations-now commonly called the Lanchester Equationsl (LEs)-as models of attrition in modern warfare. Similar ideas were proposed around that time by [chaseSS] and [osip95]. These equations are formally equivalent to the Lotka-Volterra equations used for modeling the dynamics of interacting predator-prey populations [hof98]. The LEs have since served as the fundamental mathematical models upon which most modern theories of combat attrition are based, and are to this day embedded in many state-of-the-art military models of combat. [Taylor] provides a thorough mathematical discussion. 

While the LEs are particularly relevant for the kind of static trench warfare and artillery duels that characterized most of World War I, they are too simple and lack the spatial degrees of freedom to realistically model modern combat. The fundamental problem is that they idealize combat much in the same way as Newton s laws idealize physics. 

The dynamic,s underlying EINSTein is patterned after mobile CA rules, and are somewhat reminiscent of Braitenberg s Vehicles [brait84]. Specifically, EINSTein takes a artificial-life-like bottom-up, synthesist approach to the modeling of combat, rather than the more traditional top-down, or reductionist approach,... 

Fig. 11.13 Illustration of how mobile CA (here used to model elementary combat ) can be used to explore the relationship between primitive rules governing behavior on the micro-scale and emergent behavior on the macro-scale see text.

Payoffs to Using Agent-Based Models of Combat... 

General van Riper, during the time when he was Commanding General at the Marine Corps Concept Development Command at Quantico, VA, (1994-1996), approached CNA with an idea for a study that asked. What do all of these new ideas I ve been hearing about - nonlinearity, complexity, complex adaptive systems - have to do with combat Since many at CNA knew of my deep interest in the subject, I was soon asked to direct a project addressing this question a project that has since spawned a follow-on multi-year effort sponsored by the Office of Naval Research to develop an artificial-life model of combat. To say that Lt.Gen van Riper s appear-... 

---