Physics aims to describe the natural world in ways that help us understand what happens and predict what will happen.

Reality is extremely complex, so physics rarely tries to represent every detail. Instead, it builds models: simplified descriptions that keep the features relevant to a particular question and neglect others.

A ball, for example, may be treated as a point mass when studying its trajectory, as a rigid body when studying its rotation, or as a deformable material when studying an impact. The same system can require different models depending on what we want to understand.

A good model is not the most detailed possible model, but one that captures the relevant physics.

Physics makes these models precise using physical quantities and mathematics. Relationships between quantities allow us to derive predictions:

$$ \text{physical system} \longrightarrow \text{model} \longrightarrow \text{mathematical description} \longrightarrow \text{prediction}. $$

Those predictions can then be compared with observation or experiment.

Every model relies on assumptions, so every model has a domain of validity. A model can be extremely useful without being fundamental or universally accurate. Newtonian mechanics, for example, remains an excellent description of many everyday situations even though relativity provides a deeper theory.

Physical systems can therefore be described at many levels. Physics both develops useful models for particular scales and searches for increasingly fundamental descriptions of nature.

At present, our deepest successful descriptions are mainly organized around two frameworks: the Standard Model of particle physics, which describes elementary particles and the electromagnetic, weak, and strong interactions, and general relativity, which describes gravity and spacetime. These frameworks are highly successful but do not yet form a complete unified theory.

Much of physics is therefore about choosing the right level of description: deciding what matters, what can be neglected, and which model is appropriate for the phenomenon we want to understand.