Finite Energy Sum Rule (FESR)

• FESRs are nonperturbative analytic tools that relate experimental spectral sums to theoretical predictions via weighted integrations.
• They leverage Cauchy’s theorem and specially designed integration kernels, such as pinched and Legendre types, to reduce duality violations and continuum uncertainties.
• Applications include quark mass extractions, resonance analyses, and SMEFT studies, providing a robust framework to match low- and high-energy phenomena.

A finite energy sum rule (FESR) is a nonperturbative analytic tool that exploits the analyticity and asymptotic behavior of correlation functions or scattering amplitudes to relate weighted sums of experimental spectral data up to a finite cutoff to first-principles theoretical predictions, typically rendered via the operator product expansion (OPE) or Regge asymptotics. FESRs are foundational for precision extractions of Standard Model parameters and for constraining models of resonance dynamics and UV physics across particle and nuclear theory. They allow for critical control over duality violations, continuum modeling, and operator expansion truncation through choice of integration kernels and moment strategies.

1. Analytic Foundations and General Formalism

FESRs rest on the analytic properties of two-point or four-point Green’s functions and the application of Cauchy's theorem to suitable analytic weightings. For a generic correlator $\Pi(s)$ analytic in the complex $s$-plane except for a cut $s > 0$, the master FESR is

$$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$

where $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$ is the spectral function, $w(s)$ is an analytic weight (integration kernel), and $s_0$ is the finite energy cutoff---chosen such that perturbative or effective field theory is reliable on the circle $|s|=s_0$ (Bodenstein et al., 2010, Bodenstein et al., 2013, Boito et al., 2021).

The LHS is evaluated from hadronic or experimental input up to $s_0$, while the RHS is theoretically accessible via the OPE or other QFT expansions. FESRs also generalize to scattering amplitudes, e.g., forward elastic amplitudes, using subtracted or unsubtracted dispersion relations and suitable moments (Gu et al., 2020).

2. Choice and Role of Integration Kernels

The choice of the weight $w(s)$ is central to the utility and reliability of FESRs. Two main classes are prominent:

• Pinched kernels: e.g., $s$0; these vanish at $s$1, greatly suppressing duality-violation contributions from the physical axis and end-point regions where the OPE may be unreliable (Bodenstein et al., 2010, Bodenstein et al., 2013). Higher-order pinched weights (Legendre-type, polynomials) can be constructed to guarantee vanishing moments over specific intervals and project out poorly constrained resonance windows.
• Legendre kernels: $s$2 with $s$3 mapping the $s$4 interval to $s$5; these kernels have orthogonality properties that allow the cancellation of contributions from unknown continuum regions (Bodenstein et al., 2010). The growth of Legendre kernels for $s$6 enhances resonance sensitivity, whereas their vanishing moments suppress systematic continuum effects.

Optimal kernels are devised to minimize both hadronic and perturbative uncertainties, exemplified in modern quark-mass FESR determinations, where they allow the selection of stable “duality windows” in $s$7 and specific suppression of resonance or continuum uncertainties (Bodenstein et al., 2013, Dominguez et al., 2018).

3. Operator Product Expansion and Perturbative Schemes

On the theoretical contour, the correlator is expanded using the OPE:

$s$8

where $s$9 are QCD Wilson coefficients, $s > 0$0 is the relevant quark mass, and $s > 0$1 encodes nonperturbative condensates (e.g., $s > 0$2) (Bodenstein et al., 2010, Dominguez et al., 2018). High-loop results up to four or five loops are available for several channels.

Three principal perturbative evaluation prescriptions are employed:

• Fixed Order Perturbation Theory (FOPT): $s > 0$3 and $s > 0$4 are expanded at a fixed scale $s > 0$5 with term-wise contour integration.
• Contour Improved Perturbation Theory (CIPT): full renormalization group evolution of $s > 0$6 and $s > 0$7 along the contour, resumming logarithms of $s > 0$8 and reducing higher-order sensitivity (Bodenstein et al., 2010, Dominguez et al., 2018).
• Fixed-$s > 0$9 PT (FMUPT): keeping the renormalization scale fixed, common in low-precision or exploratory studies.

The OPE also determines the scaling and suppression of nonperturbative contributions. Notably, weights with positive powers of $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$0 suppress condensate corrections, while inverse-moment (negative-power) weights amplify low-energy nonperturbative terms (Bodenstein et al., 2010, Bodenstein et al., 2011).

4. Applications and Numerical Implementations

FESRs underpin a spectrum of phenomenological determinations and theoretical probes:

• Quark masses: Precision determinations of $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$1, $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$2, $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$3 (and $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$4, $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$5) in the $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$6 scheme rely on FESRs with judicious kernel choice and high-order PQCD, stabilizing against systematic uncertainties in both experimental input and higher OPE truncation (Bodenstein et al., 2010, Bodenstein et al., 2011, Dominguez et al., 2018, Bodenstein et al., 2013).
• Testing duality and resonance parameters: The interplay of FESR and Laplace/Borel sum rules enables cross-validation of resonance predictions (e.g., $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$7-like exotic states) and provides direct handles on continuum threshold choices, OPE convergence, and the impact of higher-dimensional condensates (Albuquerque et al., 2022, Fu et al., 2018).
• Scattering amplitudes in Chiral and Effective Field Theory: FESRs formulated for scattering amplitudes (e.g., meson-meson, $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$8 photoproduction) bridge hadronic partial-wave input and Regge asymptotics, serving to map residue functions, enforce analyticity constraints, and test local duality (Collaboration et al., 2016, Guo et al., 2012).
• Electroweak and SMEFT: FESR methodology generalizes to the Standard Model Effective Field Theory (SMEFT), connecting forward elastic 4-point amplitude Taylor coefficients (Wilson coefficients of dimension-6 operators) to integrals over high-energy cross sections, elucidating UV-IR relationships and custodial-symmetry protections (Gu et al., 2020).
• Hadronic corrections in atomic physics: Finite-energy sum rules for the subtraction function in doubly virtual Compton scattering allow rigorous, data-constrained computation of structure-dependent shifts (e.g., the $$\int_0^{s_0} ds \, w(s) \, \rho(s) = -\frac{1}{2\pi i} \oint_{|s|=s_0} ds \, w(s) \, \Pi(s)$$9H Lamb shift), with explicit Regge+resonance input and controlled uncertainties (Gorchtein et al., 2013).
• Spectral sum rules in condensed matter: FESR formalism also applies to optical conductivity integrals in multiband solids (e.g., pnictides), quantifying redistribution of spectral weight between coherent and incoherent processes when only partial spectral data is experimentally accessible (Benfatto et al., 2010).

5. Systematics, Duality Violations, and Convergence Issues

The justification of OPE truncation and the size of duality-violation (DV) effects are central for FESR reliability:

• Duality violations: Even with pinched weights, residual DVs from non-OPE resonance physics are present, especially near the timelike axis. DV effects can be modeled, e.g., by exponential oscillatory functions, and their neglect in precision $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$0 extractions has led to demonstrable systematic biases (Boito et al., 2021).
• OPE truncation: The asymptotic, non-convergent nature of the OPE means that arbitrary setting of higher-dimension condensates to zero is unfounded. For high-degree weights, condensates up to large $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$1 contribute unsuppressed, mandating either direct modeling or conservative error inflation (Boito et al., 2021).
• Weight and window optimization: Numerical stability under variation of $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$2 and kernel class (e.g., degree and pinching) serves as an intrinsic diagnostic for systematic uncertainties (Bodenstein et al., 2010, Dominguez et al., 2018, Albuquerque et al., 2022).
• Comparison with Laplace/Borel sum rules: FESR and LSR often yield compatible results in their mutual stability windows when OPE convergence is well-behaved; in other channels, only FESR proves sufficiently robust (Fu et al., 2018).

6. Extensions, Physical Insights, and Theoretical Generalizations

Finite energy sum rules provide structural insights into symmetries and high-energy limits:

• Custodial symmetry and UV positivity: FESR analysis in SMEFT reveals exact connections between Wilson coefficients and symmetry-imposed sum rules, with UV positivity constraints and custodial symmetry protections directly visible in FESR structure (Gu et al., 2020).
• Chiral symmetry and isospin breaking: Generalized pion FESRs, when extended beyond the chiral limit to incorporate linear quark-mass effects, achieve correspondence with per-mille accuracy between data-based and chiral-theory-based predictions for isospin-breaking observables (Bijnens et al., 5 Feb 2026).
• Duality and large-$\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$3: FESR formalism allows scrutiny of local and semi-local duality at physical $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$4 and under extrapolation to large $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$5, identifying the dynamical correlations required for Regge-hadron cancellation and duality preservation (Guo et al., 2012).

7. Representative Table: FESR Schemes and Applications

Application Area	Typical Kernel Forms	Theoretical Inputs
Quark Mass Extraction (QCD)	Pinched, Legendre polynomials	PQCD to 4-5 loops, condensates
Exotic Hadron Spectroscopy	Power-law moments $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$6	OPE to $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$7 or $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$8
Scattering Amplitudes (ChPT, EFT)	Moments in $\rho(s) = (1/\pi) \operatorname{Im} \Pi(s)$9, $w(s)$0	Partial waves, Regge asymptotics
SMEFT Amplitudes	Forward amplitude derivatives	Wilson coefficients, cross sections
Condensed Matter: Optical Sum Rule	Cutoff-integrated $w(s)$1	Band structure, self-energies

This table illustrates that FESR methodology adapts to the specific structure of the physical observable through kernel choice and theoretical expansion, but the unifying principle remains the controlled matching between finite energy data and field-theoretic predictions.

---

Finite energy sum rules provide an essential analytic bridge between low- and high-energy physics, blending rigorous QFT analytic constraints, systematic experimental input, and advanced perturbative and non-perturbative expansions. Their reliability is enabled by careful kernel design, meticulous accounting for OPE truncation and duality-violation effects, and robust matching to phenomenological resonance and continuum models. This systematic machinery underpins major advances in parameter extractions, strong-interaction phenomenology, and hadronic structure determinations in the Standard Model and beyond. (Bodenstein et al., 2010, Bodenstein et al., 2013, Boito et al., 2021, Gu et al., 2020, Bodenstein et al., 2011, Dominguez et al., 2018, Albuquerque et al., 2022, Collaboration et al., 2016, Fu et al., 2018, Gorchtein et al., 2013, Benfatto et al., 2010, Guo et al., 2012, Bijnens et al., 5 Feb 2026)