Intuition Beginner

The Feynman propagator is the answer to a basic question: if a free quantum field is disturbed at one spacetime point, what amplitude reaches another point? In perturbative quantum field theory, every internal line in a Feynman diagram carries this object. It is the line's mathematical weight.

The word "propagator" suggests motion, but it is more precise than a particle path. A quantum field does not choose one route between two events. The propagator is the time-ordered two-point amplitude of the field. If the later event is listed first, it measures one ordering. If the earlier event is listed first, time ordering swaps the fields before taking the vacuum average.

The contour-integral representation explains the small $iϵ$ in the denominator of Feynman rules. The poles of the energy variable sit just above or below the real axis. That tiny displacement tells the contour which pole to enclose for positive time and which pole to enclose for negative time. The rule encodes causal time ordering in one compact formula.

Visual Beginner

Imagine the real energy axis as a horizontal road with two marked poles near it. One pole sits just below the road on the positive-energy side. The other sits just above the road on the negative-energy side. For positive time, the contour closes downward and catches the lower pole. For negative time, it closes upward and catches the upper pole.

That picture is the $iϵ$ prescription. It is a bookkeeping device with physical content: positive-frequency modes travel forward in time, and negative-frequency modes are handled by the opposite pole.

Worked example Beginner

Take a single momentum mode of a free scalar field. Its frequency is 5 in some units. The propagator's time part uses the factor "oscillation by frequency times elapsed time" and a normalization from the mode energy.

Step 1. Use a time gap of 3. The absolute time gap is also 3.

Step 2. Multiply frequency by the absolute time gap: 5 times 3 gives 15.

Step 3. The oscillatory part is the complex wave with phase -15. The normalization is one divided by twice the frequency, so it is 1 over 10, with an extra factor of $i$ in the physics convention.

Step 4. If the time gap were -3 instead, the absolute time gap would still be 3. The same compact expression handles both time orderings.

What this tells us: the Feynman propagator remembers the order of time through the absolute time gap, while the contour formula remembers it through which pole is selected.

Check your understanding Beginner

Formal definition Intermediate+

For a real scalar Klein-Gordon field $\varphi (t,\mathbf{x})$ of mass $m\ge 0$, set $$ \omega_{\mathbf{p}}=(|\mathbf{p}|^2+m^2)^{1/2}. $$ The time-ordering operator is defined by $$ T{\phi(x)\phi(y)}= \begin{cases} \phi(x)\phi(y), & x^0\ge y^0,\ \phi(y)\phi(x), & y^0>x^0. \end{cases} $$ The Feynman propagator is the vacuum two-point time-ordered distribution $$ \Delta_F(x-y)=\langle\Omega,T{\phi(x)\phi(y)}\Omega\rangle. $$

With metric convention $p^2=(p^0{)}^2-|\mathbf{p}{|}^2$, Chatterjee's scalar convention writes the propagator as $$ \Delta_F(t,\mathbf{x})

i\int_{\mathbb{R}^3} \frac{d^3\mathbf{p}}{(2\pi)^3,2\omega_{\mathbf{p}}} e^{-i|t|\omega_{\mathbf{p}}+i\mathbf{p}\cdot\mathbf{x}} $$ and equivalently as the boundary-value contour integral $$ \Delta_F(x)

\lim_{\epsilon\to 0^+} \int_{\mathbb{R}^4} \frac{d^4p}{(2\pi)^4} \frac{e^{-ip\cdot x}}{-p^2+m^2-i\epsilon}. $$ The small negative imaginary part in the denominator moves the pole at $p^0=+{\omega }_{\mathbf{p}}$ below the real axis and the pole at $p^0=-{\omega }_{\mathbf{p}}$ above it.

Key theorem with proof Intermediate+

Theorem (contour representation of the scalar Feynman propagator). For $x=(t,\mathbf{x})$ and ${\omega }_{\mathbf{p}}=(|\mathbf{p}{|}^2+m^2{)}^{1/2}$, $$ \lim_{\epsilon\to0^+} \int_{\mathbb{R}^4} \frac{d^4p}{(2\pi)^4} \frac{e^{-ip\cdot x}}{-p^2+m^2-i\epsilon}

i\int_{\mathbb{R}^3} \frac{d^3\mathbf{p}}{(2\pi)^3,2\omega_{\mathbf{p}}} e^{-i|t|\omega_{\mathbf{p}}+i\mathbf{p}\cdot\mathbf{x}}. $$

Proof. Fix $\mathbf{p}$ and integrate first in the complex variable $p^0$. The denominator is $$ -(p^0)^2+\omega_{\mathbf{p}}^2-i\epsilon, $$ so its two poles approach $+{\omega }_{\mathbf{p}}$ from below and $-{\omega }_{\mathbf{p}}$ from above as $ϵ\to 0^{+}$. The exponential factor is $e^{-i{p}^{0}t+i\mathbf{p}\cdot \mathbf{x}}$.

When $t>0$, close the $p^0$ contour in the lower half-plane. The arc contribution vanishes in the distributional regularised sense because $e^{-i{p}^{0}t}$ decays there after the usual damping argument. The contour orientation is clockwise, so the residue theorem gives the contribution of the pole at $p^0={\omega }_{\mathbf{p}}-i0$. The residue equals $-e^{-i{\omega }_{\mathbf{p}}t}/(2{\omega }_{\mathbf{p}})$ before the clockwise sign is applied, producing $i{e}^{-i{\omega }_{\mathbf{p}}t}/(2{\omega }_{\mathbf{p}})$ after the factor $1/(2\pi )$ from the energy integral is included.

When $t<0$, close the contour in the upper half-plane. The selected pole is $p^0=-{\omega }_{\mathbf{p}}+i0$, and the same residue computation gives $i{e}^{i{\omega }_{\mathbf{p}}t}/(2{\omega }_{\mathbf{p}})$. Since $t<0$, this is $i{e}^{-i{\omega }_{\mathbf{p}}|t|}/(2{\omega }_{\mathbf{p}})$. Multiplying by $e^{i\mathbf{p}\cdot \mathbf{x}}$ and integrating over $\mathbf{p}$ gives the displayed three-momentum formula. The value at $t=0$ is understood as the corresponding distributional boundary value.

Bridge. The contour formula builds toward the loop integrals in 08.10.06 and appears again inside every propagator line of the Dyson expansion in 08.10.03. The foundational reason is that the $iϵ$ displacement identifies causal time ordering with a complex-analytic boundary value; putting these together, the bridge is a residue calculation that generalises the two-point Wick contraction from 08.10.04.

Exercises Intermediate+

Advanced results Master

The Feynman propagator is a Green distribution with a specified boundary value. Algebraically, the denominator $-p^2+m^2-i0$ chooses the inverse of the Klein-Gordon symbol by approaching the mass shell from a prescribed side in complex energy. This is stronger than saying "solve the differential equation": retarded, advanced, Wightman, and Feynman distributions solve closely related equations but differ by their support or boundary-value prescriptions ^{[Folland 2008]}.

In spectral language, the free scalar field has positive-energy mass-shell support. The Wightman function is supported on the positive sheet of the mass hyperboloid, while the Feynman distribution is assembled by time ordering those positive-frequency pieces. The contour representation hides that assembly inside a single four-momentum integral. That is why Feynman rules can attach one algebraic denominator to a line rather than carrying separate cases for the sign of time.

Under Wick rotation, the Feynman prescription becomes the Euclidean covariance $$ C_E(p_E)=\frac{1}{p_E^2+m^2}. $$ This object is positive as a Gaussian covariance after imposing the correct test-function framework, which is the analytic heart of the Euclidean constructive approach. The reconstruction problem then asks when Euclidean Schwinger functions recover Wightman functions with relativistic spectrum and locality properties ^{[Glimm 1987]}.

Synthesis. The contour representation builds toward renormalised loop integration because every loop is assembled from these denominators; it appears again in 08.10.06 when ultraviolet divergences are isolated. The central insight identifies time ordering with a pole prescription, and this is exactly the bridge between Wick contractions and momentum-space Feynman rules. This pattern recurs in causal perturbation theory, Euclidean reconstruction, and scattering amplitudes.

Full proof set Master

Proposition (Green-distribution identity). With the convention $$ \Delta_F(x)= \lim_{\epsilon\to0^+} \int\frac{d^4p}{(2\pi)^4} \frac{e^{-ip\cdot x}}{-p^2+m^2-i\epsilon}, $$ the distribution $(-\square -m^2){\Delta }_F$ equals the boundary-value distribution obtained from $$ \lim_{\epsilon\to0^+} \int\frac{d^4p}{(2\pi)^4} e^{-ip\cdot x}\left(1+\frac{i\epsilon}{-p^2+m^2-i\epsilon}\right). $$ In the limit this is the delta distribution, with the sign fixed by the Fourier convention.

Proof. Under the Fourier transform convention used above, applying $-\square -m^2$ to $e^{-ip\cdot x}$ multiplies the integrand by $-p^2+m^2$. Therefore $$ (-\Box-m^2)\Delta_F(x)

\lim_{\epsilon\to0^+} \int\frac{d^4p}{(2\pi)^4} e^{-ip\cdot x} \frac{-p^2+m^2}{-p^2+m^2-i\epsilon}. $$ The fraction is $$ 1+\frac{i\epsilon}{-p^2+m^2-i\epsilon}. $$ The Fourier transform of $1$ is ${\delta }^{(4)}(x)$. The second term is the regulator-dependent boundary contribution that vanishes away from the light cone and is absorbed into the standard distributional definition of the Feynman inverse. This proves the stated identity and records exactly where the convention-dependent sign enters.

Proposition (time-ordering from Wightman functions). Let $$ \Delta^+(t,\mathbf{x})

\int\frac{d^3\mathbf{p}}{(2\pi)^3,2\omega_{\mathbf{p}}} e^{-i\omega_{\mathbf{p}}t+i\mathbf{p}\cdot\mathbf{x}} $$ be the positive-frequency two-point function. Then $$ \Delta_F(t,\mathbf{x}) =i\theta(t)\Delta^+(t,\mathbf{x}) +i\theta(-t)\Delta^+(-t,-\mathbf{x}) $$ matches the contour-evaluated propagator above.

Proof. If $t>0$, time ordering leaves $\varphi (t,\mathbf{x})\varphi (0,0)$ in that order, so the vacuum expectation is the positive-frequency two-point function. The convention in this unit includes the overall factor $i$, giving $i{\Delta }^{+}(t,\mathbf{x})$. If $t<0$, time ordering reverses the two fields, producing the two-point function with displacement $(-t,-\mathbf{x})$ and again the same overall factor. Substituting the integral formula for ${\Delta }^{+}$ gives exactly the three-momentum expression with $|t|$: $$ i\int\frac{d^3\mathbf{p}}{(2\pi)^3,2\omega_{\mathbf{p}}} e^{-i\omega_{\mathbf{p}}|t|+i\mathbf{p}\cdot\mathbf{x}}, $$ after the change of variable $\mathbf{p}\mapsto -\mathbf{p}$ in the $t<0$ term.

Connections Master

• 06.01.03 supplies the residue theorem used to evaluate the energy contour and select the shifted pole.

• 08.10.04 defines the Wick contraction; for two time-ordered scalar fields that contraction is precisely the Feynman propagator.

• 08.10.03 uses this propagator as the line factor in the Dyson expansion and in the Feynman diagrams of ${\varphi }^4$ theory.

• 08.09.01 explains the analytic continuation that changes the Lorentzian Feynman denominator into the Euclidean covariance.

• 08.07.01 packages the same covariance as the Gaussian kernel in the free path integral.

Historical & philosophical context Master

Stueckelberg's 1941 work anticipated the interpretation of antiparticles through reversed time direction ^{[Stueckelberg 1941]}. Feynman's 1949 papers turned that interpretive move into a computational calculus: a line in a diagram carried a propagator, and the pole prescription told the line how to pass singular energies ^{[Feynman 1949]}.

The philosophical point is that causality is not imposed by adding a verbal rule after the calculation. It is encoded analytically in the boundary value of a distribution. The $iϵ$ is tiny in notation but large in meaning: it chooses a causal inverse among several possible Green distributions.

For mathematicians, the propagator is also a warning about formal inverses. The Klein-Gordon operator has many distributional inverses, and the physically relevant one depends on time ordering, spectrum, and analytic continuation. That is why the same denominator appears differently in Wightman theory, Euclidean field theory, and perturbative scattering.

Bibliography Master

@article{Stueckelberg1941PairCreation,
  author = {Stueckelberg, E. C. G.},
  title = {Remarque {\`a} propos de la cr{\'e}ation de paires de particules en th{\'e}orie de relativit{\'e}},
  journal = {Helvetica Physica Acta},
  volume = {14},
  pages = {322--323},
  year = {1941}
}

@article{Feynman1949Positrons,
  author = {Feynman, Richard P.},
  title = {The Theory of Positrons},
  journal = {Physical Review},
  volume = {76},
  pages = {749--759},
  year = {1949}
}

@article{Feynman1949SpaceTime,
  author = {Feynman, Richard P.},
  title = {Space-Time Approach to Quantum Electrodynamics},
  journal = {Physical Review},
  volume = {76},
  pages = {769--789},
  year = {1949}
}

@misc{Chatterjee2022QFT,
  author = {Chatterjee, Sourav},
  title = {Introduction to Quantum Field Theory for Mathematicians},
  year = {2022},
  note = {Stanford lecture notes}
}

@book{PeskinSchroeder1995QFT,
  author = {Peskin, Michael E. and Schroeder, Daniel V.},
  title = {An Introduction to Quantum Field Theory},
  publisher = {Addison-Wesley},
  year = {1995}
}

@book{Folland2008TouristGuide,
  author = {Folland, Gerald B.},
  title = {Quantum Field Theory: A Tourist Guide for Mathematicians},
  publisher = {American Mathematical Society},
  year = {2008}
}

@book{GlimmJaffe1987QuantumPhysics,
  author = {Glimm, James and Jaffe, Arthur},
  title = {Quantum Physics: A Functional Integral Point of View},
  edition = {2},
  publisher = {Springer},
  year = {1987}
}

@book{StreaterWightman2000PCT,
  author = {Streater, R. F. and Wightman, A. S.},
  title = {PCT, Spin and Statistics, and All That},
  publisher = {Princeton University Press},
  year = {2000}
}