Examples of Applications
Modeling and simulation are not bound to a single discipline. The same methodology — strip a system down to a formal description, then experiment on the description instead of the system — shows up wherever a question is worth asking and the original is too large, too dangerous, too slow, or too expensive to probe directly. The breadth of those settings is itself a reason the field is worth studying as a methodology in its own right, separately from any one of its host disciplines.
A non-exhaustive tour
Rather than enumerate every corner of science and industry where models live, it helps to group representative examples into a handful of broad areas — enough to give a sense of the range without pretending the list is complete.
- Natural sciences. Astrophysics and geophysics in physics, protein science and drug design in chemistry, bioinformatics and bioengineering in biology, smart materials and nanostructures in materials science. Many of these fields are mathematical at the foundations — Newton’s laws, Maxwell’s equations, the Schrödinger equation — so modeling is more the native language than an imported tool.
- Climate and weather. Tomorrow’s storm, the next century’s average temperature, the gulf stream, the ozone hole. Same machinery, very different time horizons.
- Engineering and industry. The automotive industry alone covers crash mechanics, wind-tunnel aerodynamics, combustion in fuel injection, microsystems coupling for airbags, vehicle dynamics, and aeroacoustics for cabin noise. Semiconductor work spans individual transistor devices, the chemical processes that produce purified crystals, circuit simulation, and full chip layout. Computer graphics models how light interacts with surfaces, locally and globally.
- Economics and finance. Business cycles, fiscal and monetary policy, pricing, option valuation. These are also the fields where the appropriateness of mathematical modeling is most contested — the models themselves are real, and so are the disagreements about what they predict.
- Logistics, traffic, and operations. Routing, scheduling, fleet management, congestion control, signal timing. Often combinatorial rather than continuous, but no less model-driven.
- Strategy and society. Military, political, and economic scenario analysis; electoral and opinion research; game-theoretic models of competition. Tends to lean on plausible structural assumptions rather than first-principles equations.
- Information systems. Coding theory and information models, load and reliability models for service providers, fail-safe network design, controlling models for complex systems, workflow models in software engineering. Computer science is not only a tool for modeling — it is regularly itself the target of modeling.
The catalog could be extended indefinitely. The point is simply that the methodology travels.
Hard models and soft models
Looking across that range, two broad styles of model stand out.
- Hard models are equation-based, grounded in physical laws or quantitative theory. The model takes the form of a system of equations one then has to solve, and its credibility rests on the validity of those equations. Example: a planetary orbit derived from Newton’s law of gravitation.
- Soft models are phenomenological or descriptive. They capture observed regularities, structural relationships, or qualitative behavior without committing to a first-principles derivation underneath. Many models in economics, sociology, and decision-making live here, as do most data-driven models that fit structure to observations. Example: a software-engineering workflow model that describes how tasks move from intake to release.
Most realistic applications mix the two. A weather forecast has hard physics at its core — fluid dynamics, thermodynamics — wrapped in soft parameterizations for everything that lives below the model’s spatial resolution. The classification chapter later in this part of the notes returns to this distinction more carefully.
The questions this raises
Two questions follow naturally from such a wide application range, and they shape the rest of this chapter.
- How does one arrive at a suitable model? Out of the infinitely many possible simplifications of a real system, which one captures enough to be useful and little enough to be tractable? This is what the derivation section addresses.
- Which descriptive tools should be used? Once a model is in hand, what mathematical machinery — equations, graphs, automata, distributions, optimization problems — is the right vocabulary to express it in? This is what the classification section addresses.
The remaining sections of the chapter walk through these questions in turn.