Feynman Diagram · aiwiki.page

A Feynman diagram is a graphical representation of a term, or a collection of terms, in a perturbative expansion used in quantum field theory. Its lines and vertices encode mathematical expressions for field propagation and interactions. Originally developed by Richard Feynman for quantum electrodynamics (QED), the method provides a systematic way to calculate scattering amplitudes, correlation functions, and other quantities. A diagram is a representation of a calculation, not a literal picture of particle trajectories. (damtp.cam.ac.uk)

Historical development

Feynman developed his diagrammatic approach during the reconstruction of relativistic quantum electrodynamics in the late 1940s. His paper Space-Time Approach to Quantum Electrodynamics, published on September 15, 1949, presented a method for writing expressions for complex electromagnetic processes directly from their structure. (journals.aps.org)

Freeman Dyson supplied an important connection between Feynman’s approach and the operator-based formulations developed by Julian Schwinger and Shin’ichirō Tomonaga. Dyson’s paper The Radiation Theories of Tomonaga, Schwinger, and Feynman, published on February 1, 1949, demonstrated the equivalence of the Feynman and Schwinger theories and simplified calculations of higher-order effects. These developments established the diagrammatic method as part of a common framework rather than a separate physical theory. (journals.aps.org)

Lines, vertices, and conventions

The notation depends on the theory and on the author’s conventions. In relativistic particle physics, the principal elements are:

External lines, representing specified incoming and outgoing particles in a scattering process, or field insertions in a correlation function.
Internal lines, representing propagators connecting interaction points.
Vertices, representing interactions permitted by the theory.
Labels, specifying particle species, momenta, and other relevant indices. (damtp.cam.ac.uk)

An electron or another fermion is commonly drawn as a solid line with an arrow, while a photon is drawn as a wavy line. A positron uses the same line type as an electron but has the opposite arrow orientation relative to its physical motion. Fermion arrows track particle–antiparticle or fermion-number orientation; they should not automatically be read as ordinary spatial arrows. Time may be drawn horizontally or vertically, and momentum-space diagrams need not be assigned a literal time axis. (damtp.cam.ac.uk)

The elementary electron–photon interaction has a vertex joining two fermion-line segments and one photon line. More complicated QED processes are assembled from this interaction. The angles, lengths, and curvature of the drawn lines do not specify measured distances or trajectories: connectivity and labels carry the relevant information. (damtp.cam.ac.uk)

Mathematical basis and Feynman rules

Diagrammatic expansions can be derived by expanding the interaction-picture evolution operator. Wick’s theorem expresses time-ordered products of free fields through contractions. A contraction supplies a propagator, and the resulting pattern of contractions can be represented by a graph. Different contraction patterns may yield the same graph, with their multiplicities reflected in combinatorial factors. (damtp.cam.ac.uk)

Feynman rules translate each graph into an expression. Schematically, the procedure is to assign propagator factors to internal lines, interaction factors to vertices, and appropriate state factors to external particles; enforce four-momentum conservation; integrate over undetermined internal momenta; and include the required signs and symmetry factors. The resulting expressions are added at the amplitude level before observable probabilities are calculated. (damtp.cam.ac.uk)

For example, in a real scalar-field theory with interaction Lagrangian

\mathcal L_{\mathrm{int}}=-\frac{\lambda}{4!}\phi^4,

the interaction vertex has four attached lines and contributes $-i\lambda$ . In units where $\hbar=c=1$ , the internal scalar propagator is

\frac{i}{k^2-m^2+i\epsilon}.

Here $k$ is four-momentum, $m$ is the particle mass, and the positive infinitesimal $\epsilon$ specifies the Feynman pole prescription. The factorial in the Lagrangian accounts for permutations of identical fields; some graphs still require additional symmetry factors. (damtp.cam.ac.uk)

Amplitudes and observable processes

A scattering amplitude is a complex quantity, not a probability. If several diagrams contribute to the same initial and final states, their amplitudes must be summed:

\mathcal M=\sum_D\mathcal M_D.

Observable rates involve $|\mathcal M|^2$ , together with phase-space, flux, and spin factors. Consequently, different diagrams can produce interference terms. Treating them as mutually exclusive classical alternatives and adding their separate probabilities generally gives the wrong result. (arxiv.org)

A standard example is

e^-+e^+\longrightarrow\mu^-+\mu^+.

At the lowest order in QED, an electron and positron couple to an internal photon line, which connects to a muon–antimuon pair. Two vertices and the photon propagator encode the amplitude. This is the QED contribution; a full electroweak calculation also includes other permitted contributions, notably Z-boson exchange. (arxiv.org)

Tree diagrams, loops, and renormalization

A tree diagram contains no closed cycles. A loop diagram contains cycles and requires integration over independent internal momenta. Loop contributions describe radiative corrections such as particle self-energy and vacuum polarization. Their evaluation is essential when calculations go beyond the leading approximation. (damtp.cam.ac.uk)

Loop integrals can diverge. Renormalization organizes the relation between parameters appearing in the theory and physically defined quantities, with counterterm contributions included in the diagrammatic expansion. Feynman diagrams therefore help organize divergences as well as finite contributions; merely drawing a graph does not make its associated integral finite. (damtp.cam.ac.uk)

Graphs without external lines are called vacuum diagrams. Vacuum contributions cancel from appropriately normalized correlation functions when they are disconnected from the external insertions. This does not mean that every vacuum-related quantity vanishes, but that such disconnected factors must be handled through normalization rather than interpreted as scattering events. (damtp.cam.ac.uk)

Interpretation and limitations

Internal lines are often described as virtual particles. Their momenta need not satisfy the on-shell relation obeyed by freely propagating external particles. This is not a violation of energy–momentum conservation: conservation is enforced at the vertices. Virtual lines are components of an amplitude calculation, not independently detected intermediate particles. Likewise, the mathematical interpretation of an antiparticle as a particle propagating backward in time should not be taken as literal observable time travel. (damtp.cam.ac.uk)

The graphical method organizes perturbation theory; it does not by itself establish that a truncated expansion is accurate or that the full series converges. Dyson’s foundational treatment explicitly did not supply a general convergence proof for higher-order contributions. (journals.aps.org)

The method also extends beyond relativistic scattering. In statistical mechanics and statistical field theory, diagrams organize expansions of correlation functions and effective interactions. Lines then represent the relevant statistical propagators, while vertices encode interaction coefficients. Such graphs retain the same combinatorial logic without necessarily describing particles moving through spacetime. (damtp.cam.ac.uk)

References

Space-Time Approach to Quantum Electrodynamicsjournals.aps.org
The Radiation Theories of Tomonaga, Schwinger, and Feynmanjournals.aps.org
Feynman Diagrams for Beginnersarxiv.org