Skip to content

Loop Integrals and Reduction

A loop integral is specified by its measure, denominators, numerator, regulator, and analytic prescription. This chapter connects parameterization and reduction with asymptotic, differential-equation, and numerical methods, keeping ultraviolet and infrared origins separate.

Begin with Anatomy of a Loop Integral to define the object before selecting an evaluation method.

  1. Anatomy of a Loop Integral — Records the complete regulated integral and its singular regions.
  2. Feynman and Schwinger Parameters — Combines propagators into useful parameter integrals.
  3. Dimensional Regularization as an Amplitude Tool — Continues integrals in spacetime dimension without identifying regulation with renormalization.
  4. UV/IR Poles and the Renormalized-Amplitude Interface — Classifies singular poles and the required renormalized inputs.
  5. One-Loop Integral Families and Analytic Functions — Organizes basic one-loop topologies and their branches.
  6. Tensor Reduction — Converts Lorentz tensor numerators into scalar integral coefficients.
  7. Expansion by Regions — Builds asymptotic expansions from the relevant momentum scalings.
  8. Integration-by-Parts Identities and Master Integrals — Reduces integral families to a set of independent masters.
  9. Differential Equations for Master Integrals — Evolves masters in kinematics once boundary constants are fixed.
  10. Numerical Evaluation and Validation of Loop Integrals — Tests numerical values against precision, branch, and independent-representation checks.

Integration-by-parts identities relate integrals within a regulated family; see Weinzierl 2022, § 6.1. Differentiating with respect to a mass or invariant gives further members of that family, so reduction produces a closed differential system for the masters. Solving it still requires boundary constants and analytic-continuation data; an equation alone does not select the physical solution.

A vanishing scaleless integral in dimensional regularization does not show that no singular region exists. Ultraviolet and infrared contributions can cancel under the common prescription, while a physical renormalization or factorization problem needs their origins distinguished.

Why should a numerical value of a master integral be checked at more than one precision or representation?

Solution

A small integration error estimate may not detect a wrong branch, contour, normalization, or missing cancellation. Precision changes test numerical stability; an independent representation or known limit tests whether the intended mathematical integral was evaluated.

Continue to Singularities, Cuts, and Integrand Reconstruction to connect the integral’s analytic structure with on-shell data.

  • Stefan Weinzierl, Feynman Integrals, UNITEXT for Physics, Springer, 2022. DOI. Open PDF.