The Feynman Diagram

Feynman diagrams are not pictures. They are schematics — shorthand made of lines and vertices that tells you how to write down a number. That number is the amplitude for some process to happen. You draw lines, you follow rules, you get a formula, you square it, you get a probability. That's the whole game in a nutshell.

Richard Feynman invented them in 1948, and the trick was so good that physicists stopped writing equations on paper and started drawing pictures. This sounds like a joke. It isn't.

Lines and What They Mean

A straight line with an arrow is a fermion — an electron, a quark, whatever. A wavy line is a photon. A coiled line might be a gluon. These aren't literal shapes. The wavy line doesn't mean the photon is waving around. It means "this is a boson, specifically a vector boson." The straight line with an arrow means "this is a spin-half particle carrying charge." The drawing conventions are a language, not a photograph.

A line that starts on the outside of the diagram and goes inward is an incoming particle. A line that goes outward is outgoing. Lines that connect two vertices inside the diagram are internal — they represent particles that exist only transiently, in the quantum sense.

Vertices

A vertex is where lines meet. In quantum electrodynamics, every vertex connects exactly three lines: two fermion lines and one photon line. That's it. One electron emits a photon. One electron absorbs a photon. An electron-positron pair annihilates into a photon. A photon splits into an electron-positron pair. All of these processes share the same vertex, because they're all described by the same term in the Lagrangian.

The vertex carries a coupling constant — in QED, it's the fine structure constant α ≈ 1/137, or more precisely the charge e. Every vertex in a diagram contributes a factor of e. So a diagram with two vertices gives you e². Four vertices gives e⁴. The expansion in powers of the coupling constant is called perturbation theory, and Feynman diagrams are its natural language.

Reading a Diagram

Let's read a simple diagram: electron-electron scattering. Two incoming electron lines come in from the left. They exchange a photon. Two outgoing electron lines leave to the right. That's one vertex on the left, one on the right, a photon line connecting them. This is called the "tree-level" diagram — it's the simplest possible contribution.

To turn this into a number, you follow Feynman rules:

• Every internal fermion line → a propagator (iγ·p - m)⁻¹
• Every internal photon line → a photon propagator
• Every vertex → factor of -ieγ^μ
• Integrate over all internal momenta
• Include combinatorial factors

The result is an amplitude M. The cross section is proportional to |M|². This is how you go from a stick-figure drawing to a prediction you can compare with a detector.

Why They Work

The deep reason Feynman diagrams work is that the path integral — the sum over all possible field configurations — can be expanded as a Taylor series in the coupling constant. Each term in that series corresponds to a specific topology of diagram. The diagrams aren't approximations of something more fundamental; they are the perturbation series, organized by the number of vertices.

There are limitations. When the coupling is large — like in QCD at low energies — the series converges slowly or not at all, and Feynman diagrams become useless. Lattice gauge theory takes over in that regime. But for QED, where α ≈ 1/137, the expansion is spectacularly accurate. The anomalous magnetic moment of the electron has been calculated using diagrams with up to five loops, and it agrees with experiment to better than one part in a billion.

The Picture Problem

The biggest mistake people make is taking Feynman diagrams literally. The "virtual particles" on internal lines are not little balls flying between vertices. They are terms in a mathematical expansion. The lines don't have trajectories. The vertices don't happen at specific points in spacetime in any classical sense.

But the diagrams are not just mnemonic devices either. They capture real structure: the topological classification of contributions, the symmetry factors, the divergence structure. A diagram with a loop integral tells you there's an ultraviolet divergence. A diagram with an internal photon tells you about radiative corrections. The pictures encode deep physics.

Feynman diagrams are what physicists call a "computational visualization" — the picture is the calculation, once you know the language.