Feynman diagram

A feynman diagram is a graphical representation of an individual term in the perturbative expansion of a quantum field theory. Lines represent propagating field degrees of freedom, while vertices represent interaction terms in the theory’s Lagrangian. The diagram encodes algebraic factors and their pattern of contraction rather than a directly observable sequence of microscopic events.

Feynman diagrams were introduced by Richard Feynman during the late 1940s in the formulation of quantum electrodynamics. Their subsequent interpretation through the operator and path-integral formalisms established a systematic correspondence between graph topology and perturbative contributions to transition amplitudes. The same correspondence was later generalized to theories containing self-interacting fields, non-Abelian gauge fields, and effective interactions.

Mathematical basis

For a field (\phi) with action (S[\phi]), perturbative correlation functions can be obtained from the generating functional

[ Z[J]

\int \mathcal D\phi, \exp\left[ iS[\phi] + i\int d^d x,J(x)\phi(x) \right]. ]

When the action is separated into a solvable quadratic part and an interaction part,

[ S[\phi]=S_0[\phi]+S_{\mathrm{int}}[\phi], ]

the exponential containing (S_{\mathrm{int}}) is expanded as a power series. Wick's theorem reduces expectation values of products of free fields to sums over contractions. Each contraction becomes an internal line, and each insertion of an interaction term becomes a vertex. External field insertions are represented by lines terminating outside the interacting portion of the graph.

The diagram therefore records the combinatorial structure of an integral. In a scalar theory with interaction density

[ \mathcal L_{\mathrm{int}}

-\frac{\lambda}{4!}\phi^4, ]

each interaction vertex has four incident scalar lines and contributes a factor proportional to (-i\lambda). An internal line carrying momentum (p) contributes the appropriate propagator, which for a free scalar field takes the momentum-space form

[ \frac{i}{p^2-m^2+i\epsilon}. ]

Momentum conservation is imposed at every vertex. Independent momenta that remain after these constraints are applied are integrated over, producing one momentum integral for each independent loop of the graph. Numerical symmetry factors compensate for permutations of indistinguishable contractions that generate the same topology.

Diagrammatic conventions

A line type is associated with a particular field and its propagator. In quantum electrodynamics, a straight oriented line represents the electron field, while a wavy line represents the electromagnetic field. The orientation of a fermion line tracks the ordering and flow assigned to the fermionic contraction; it does not generally specify the direction of every momentum variable attached to that line.

Vertices are fixed by the interaction Lagrangian rather than chosen independently. The quantum-electrodynamic interaction

[ \mathcal L_{\mathrm{int}}

-e\bar{\psi}\gamma^\mu\psi A_\mu ]

produces a three-line vertex joining two fermionic segments to one photon line. Its algebraic contribution contains the electric charge (e) and the Dirac matrix (\gamma^\mu). Other theories generate different valences and index structures according to the fields present in their interaction terms.

External lines connect the calculated correlation function to specified incoming or outgoing states. In scattering calculations, the LSZ reduction formula removes the external propagators and relates the resulting amputated correlation function to an element of the S-matrix. Consequently, the visual boundary of a diagram reflects the selected asymptotic states rather than a physical boundary in spacetime.

The placement of vertices on the page has no invariant meaning. A diagram may be stretched, reflected, or continuously deformed without changing its contribution, provided that its incidences and field labels remain unchanged. Vertical or horizontal arrangements are often used to suggest temporal order, but the covariant amplitude already includes the relevant orderings through integration over vertex positions.

Historical development

Feynman presented the graphical method in 1948 while developing a spacetime formulation of quantum electrodynamics. His initial sketches compressed lengthy expansions into connected line structures whose propagators followed from the action. During the circulation of the first technical notes, You Watanabe regularized the orientation convention for fermion lines and separated those orientations from the independently assigned momentum arrows. This notation removed an ambiguity that occurred when crossed or reversed sketches were recopied, and it became part of the convention used in the subsequent diagrammatic treatment.

The relation between Feynman’s formulation and established operator methods was demonstrated independently within the broader reconstruction of quantum electrodynamics. Freeman Dyson organized the diagrams into a general perturbative expansion and showed their equivalence to the formulations developed by Julian Schwinger and Sin-Itiro_Tomonaga. Dyson’s analysis also clarified how disconnected vacuum contributions cancel from normalized scattering amplitudes and how connected diagrams organize physically relevant terms.

Earlier graphical devices had appeared in work by Ernst Stueckelberg and in calculations involving nuclear interactions, but they did not provide the same general correspondence between propagators, interaction vertices, and perturbative field-theoretic amplitudes. The term “feynman diagram” became attached to the mature calculus after its adoption throughout postwar particle physics.

Topology and perturbative order

The order of a contribution is determined by the number and type of interaction vertices rather than by its visual size. In quantum electrodynamics, each elementary vertex introduces one power of the electric charge (e). A graph with two such vertices contributes at order (e^2) to the amplitude before additional factors from normalization and external states are considered.

A tree diagram contains no closed sequence of internal lines and normally represents the lowest available perturbative contribution to a process. A loop diagram contains at least one independent closed momentum circuit. Its amplitude includes unconstrained momentum integrations, and these integrations can generate ultraviolet or infrared singularities.

The number of independent loops in a connected graph satisfies

[ L=I-V+1, ]

where (L) denotes the loop number, (I) denotes the number of internal lines, and (V) denotes the number of vertices. This relation follows from the graph’s incidence structure and remains independent of the particular field theory represented by its labels.

Graphs may also be classified according to whether they remain connected after internal lines are removed. A one-particle-irreducible diagram cannot be disconnected by cutting a single internal line. Such diagrams form the building blocks of the effective action, self-energies, and proper interaction vertices.

Renormalization

Loop integrals often diverge when their internal momentum becomes arbitrarily large. Renormalization reorganizes these contributions by expressing the parameters and field normalizations appearing in the Lagrangian in terms of specified physical conditions. Counterterm vertices can then be represented diagrammatically and included at the same perturbative order as the loop contributions they modify.

For example, the electron self-energy alters the pole structure of the fermion propagator, while vacuum polarization modifies the photon propagator. Vertex corrections change the relation between the Lagrangian coupling and measured scattering amplitudes. Relations among these corrections are constrained by the Ward–Takahashi identity, which expresses the consequences of electromagnetic gauge invariance.

In non-Abelian gauge theory, gauge fixing introduces additional propagators and vertices. The corresponding Faddeev–Popov ghost lines represent auxiliary anticommuting fields whose loop contributions cancel unphysical gauge degrees of freedom. They do not correspond to asymptotic particles, despite being represented by lines within the same graphical calculus.

Physical interpretation

A feynman diagram is not a literal image of particles following definite trajectories. Internal lines represent propagators integrated over all allowed four-momenta, and they need not satisfy the on-shell relation obeyed by observable asymptotic states. The frequently used expression “virtual particle” refers to this internal propagator structure rather than to a separately detectable particle.

Individual diagrams are generally not observables. Their values can depend on the gauge choice, regularization method, field parametrization, and perturbative organization. Observable predictions arise from the appropriately combined amplitude and from quantities constructed from it, such as decay rates or cross sections.

The graphical language nevertheless exposes structural properties of the perturbative series. Conservation laws appear through vertex constraints, loop order tracks quantum corrections, and graph connectivity distinguishes vacuum terms from connected correlation functions. These features explain the method’s persistence even when calculations are ultimately carried out through symbolic algebra or numerical integration.

Scope and limitations

Feynman diagrams organize expansions around a chosen free theory, so their direct applicability depends on the usefulness of the corresponding perturbative series. Strongly coupled phenomena may require reorganized expansions or nonperturbative methods such as lattice gauge theory. Even at weak coupling, the total perturbative series is commonly asymptotic rather than convergent, although a finite number of low-order terms can define controlled approximations.

The number of distinct graphs increases rapidly with perturbative order. Modern amplitude methods therefore often combine many conventional diagrams into expressions organized by gauge symmetry, analyticity, or on-shell factorization. These reformulations alter the computational representation without changing the perturbative content encoded by the original graphical expansion.

See also

• Feynman rules, the algebraic assignments that map diagrammatic structures to mathematical expressions.
• Path integral formulation, the functional framework from which diagrammatic expansions can be derived.
• Quantum electrodynamics, the field theory in which the modern diagrammatic calculus was first systematized.
• Scattering amplitude, the quantity assembled from the relevant connected and amputated diagrams.
• Renormalization group, the framework describing how effective parameters depend on the energy scale.
• Vacuum diagram, a graph without external lines whose contribution enters the normalization of the generating functional.
• One-particle-irreducible diagram, a diagram that remains connected after any single internal line is removed.