Finite-volume effects on smeared spectral densities

arXiv:2606.14349 · hep-lat, hep-ph, hep-th · Submitted 2026-06-12 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: High Energy Physics: Lattice. The day in lattice field theory: what the simulations put on the table.

Kai: We're looking at the paper: "Finite-volume effects on smeared spectral densities".

Mira: Here's the gist of the paper. Using two distinct approaches, we derive a universal expression for the leading finite-volume effects of the smeared vector-vector spectral density (proportional to the smeared hadronic R-ratio) in a periodic cubic spatial volume of side length L <ref:2606.14349#pg6>.

Kai: This result is useful for controlling the L → ∞ extrapolation of smeared spectral densities by defining a scaling regime where finite-volume effects are dominated by leading terms in a large L expansion <ref:2606.14349#pg4>.

Mira: Derivation of Finite-Volume Corrections The authors derive the leading finite-volume effects on the Euclidean two-point function, GL(τ), using two distinct methods: one based on previous work for Euclidean two-point functions and another starting from the Lellouch-L¨uscher-Meyer expression for spectral decomposition <ref:2606.14349#pg2>.

Kai: They show that the L dependence is exponentially suppressed for a certain class of smearing kernels, and leading effects can be expressed universally in terms of the pion form factor <ref:2606.14349#pg6>.

Paper summary: Kai: So, looking at this paper "Finite-volume effects on smeared spectral densities," the authors successfully used two distinct approaches—the kernel reconstruction method and the Lellouch-L¨uscher-Meyer formalism—to derive a universal expression for the leading finite-volume effects on smeared vector-vector spectral densities.

Mira: The core of their argument is that they showed that for certain smearing kernels, the dependence on the volume L is exponentially suppressed, and this leading effect can be universally written in terms of the pion form factor.

Lev: They also proved this consistency by showing how the two different approaches align, which gives confidence in their final result.

Kai: For someone listening who only cares about what this means for experimental data or simulations, it basically provides a tool to handle those finite-volume corrections systematically when they try to take the limit to infinite volume.

Mira: They are also setting up a framework where you can relate Euclidean correlators directly to smeared spectral densities using coefficients designed for reconstruction, which helps manage the statistical variance and systematic uncertainty involved in that approximation.

Lev: The paper states one limitation, which is that they didn't specify the regularization procedure they used when defining those coefficients c sub i alpha.

Kai: That means while they've shown how to control the L to infinity for a specific class of kernels, the actual choice of smearing kernel still matters because it determines how well that approximation works.

Mira: Overall, this work gives us a strong theoretical understanding of these corrections by linking them through two different mathematical lenses, which should be useful for future lattice calculations involving smeared spectral densities.

Conclusion: Kai: So, this paper is "Finite-volume effects on smeared spectral densities," by Kai and Mira. It lays out how you can actually take those smeared measurements and get them to look like they come from an infinite volume system, even when you're using a finite box.

Mira: Right. The authors are showing that they found two different mathematical ways to describe these corrections—one based on reconstructing the spectral density directly, and another using that Lellouch-L¨uscher-Meyer formalism.

Kai: And both of those methods ended up agreeing on the leading finite-volume effect being tied to some kind of pion form factor. It’s not a huge leap, but it gives us a way to control the error when we extrapolate our finite volume calculations to infinity.

Lev: From my side, if we were running this on actual hardware, I'd be looking at how fast that exponential suppression happens with respect to the box size L. If it’s really strong enough, then the leading correction is just a function of some basic particle properties and not a complicated lattice artifact.

Mira: Exactly. The implication is that for certain types of smearing kernels, we can define a clear scaling regime where we know exactly how to handle those volume effects in our final results.

Kai: It means we can move past just hoping the corrections are small and start having a systematic way to check if our extrapolation method actually works for different setups.

Mira: And it opens the door for applying this same framework to other types of spectral densities that might be hard to handle otherwise, as long as we have a known way to connect those finite volume calculations back to infinite volume scattering amplitudes.

Lev: So the next thing we need is seeing how this plays out when you plug in real lattice data, not just these idealized mathematical models.

Dipartimento di Fisica, University of Milano-Bicocca · INFN Sezione di Milano-Bicocca · Higgs Centre for Theoretical Physics, The University of Edinburgh

hep-lat, hep-ph, hep-th

Submitted: 2026-06-12

Updated: 2026-11-04

Comments: 19 pages, 8 figures. v2: improved discussion of subleading finite-volume terms, including a new appendix table; minor corrections

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 91/100

The gist: Using two distinct approaches, we derive a universal expression for the leading finite-volume effects of the smeared vector-vector spectral density (proportional to the smeared hadronic R-ratio) in a…

Key concepts

Smeared Spectral Density
This is a representation of the spectral density that incorporates smearing, which smooths out the data. The paper defines it as $ ho_{bL, ilde{oldsymbol{ au}}}(oldsymbol{ au}) = ilde{Z} rac{1}{L} rac{doldsymbol{ au}}{d heta} ho_L( heta)$, linking the smeared density to the infinite-volume spectral density $ ho_L$. This allows for a more controlled analysis.
Lellouch-Lüscher-Meyer Formalism
This approach relates finite-volume Euclidean two-point functions, $GL( au)$, to energy eigenstates within a finite box of size L. It uses the quantization condition $Q(E_n, L) = rac{ ext{constant}}{L}$ to connect these discrete volume energies ($E_n$) to the infinite-volume scattering amplitudes.
Finite-Volume Corrections
These are the leading corrections that arise when calculations are performed in a finite spatial volume (side length L). The paper shows these effects can be expressed universally in terms of quantities like the pion form factor, allowing researchers to check convergence and identify when the infinite-volume limit is well-approximated.
Spectral Reconstruction Kernel
This framework defines coefficients $c_i(oldsymbol{ au})$ that approximate the full Euclidean correlator $GL( au)$ by using a finite number of sampled values. This method provides a statistical variance formula and a systematic uncertainty formula related to how well the approximation $ ilde{ ho}_b$ matches the true smeared density $ ho_b$.

Terminology

Summary

Using two distinct approaches, we derive a universal expression for the leading finite-volume effects of the smeared vector-vector spectral density (proportional to the smeared hadronic R-ratio) in a periodic cubic spatial volume of side length L <ref:2606. p. 0

Where the paper says it

Bresciani , 1, 2, ∗ Mattia Bruno , 1, 2, † and Maxwell T. Hansen 3, ‡ 1Dipartimento di Fisica, Universit`a degli Studi di Milano-Bicocca, Piazza della Scienza 3, I-20126 Milano, Italy 2 INFN, Sezione di Milano-Bicocca, Piazza della Scienza 3, I-20126 Milano, Italy 3Higgs Centre for Theoretical Physics, School of Physics and Astronomy, The University of Edinburgh, Edinburgh EH9 3FD, UK (Dated: June 15, 2026) Using two distinct approaches, we derive a universal expression for the leading finite-volume effects of the smeared vector-vector spectral density (proportional to the smeared hadronic R-ratio) in a periodic cubic spatial volume of side length L.

14349#pg6>. This result is useful for controlling the L → ∞ extrapolation of smeared spectral densities by defining a scaling regime where finite-volume effects are dominated by leading terms in a large L expansion <ref:2606. p. 0
Where the paper says it

The same representation is then derived starting from the Lellouch-L¨uscher-Meyer expression for the spectral decomposition of the correlator. The results may prove useful for controlling the L → ∞ extrapolation of smeared spectral densities, in particular by defining a scaling regime in which the finite-volume effects are dominated by the leading terms in a large L expansion and thus can be reliably estimated.

14349#pg4>.

Derivation of Finite-Volume Corrections

The authors derive the leading finite-volume effects on the Euclidean two-point function, GL(τ), using two distinct methods: one based on previous work for Euclidean two-point functions and another starting from the Lellouch-L¨uscher-Meyer expression for spectral decomposition <ref:2606.14349#pg2>. They show that the L dependence is exponentially suppressed for a certain class of smearing kernels, and leading effects can be expressed universally in terms of the pion form factor <ref:2606. p. 0

Where the paper says it

Hansen 3, ‡ 1Dipartimento di Fisica, Universit`a degli Studi di Milano-Bicocca, Piazza della Scienza 3, I-20126 Milano, Italy 2 INFN, Sezione di Milano-Bicocca, Piazza della Scienza 3, I-20126 Milano, Italy 3Higgs Centre for Theoretical Physics, School of Physics and Astronomy, The University of Edinburgh, Edinburgh EH9 3FD, UK (Dated: June 15, 2026) Using two distinct approaches, we derive a universal expression for the leading finite-volume effects of the smeared vector-vector spectral density (proportional to the smeared hadronic R-ratio) in a periodic cubic spatial volume of side length L. First, building on the results of previous work for finite-volume effects on Euclidean two-point functions, we show that the L dependence is exponentially suppressed for a certain class of smearing kernels, and that the leading effects can be expressed universally in terms of the pion form factor.

14349#pg6>. The main result, Eq. p. 1
Where the paper says it

IV, such a representation is only useful when the leading terms in the series dominate. Our main result, Eq.

(25), allows one to check convergence explicitly for a given choice of smearing kernel and thereby identify the onset of the asymptotic regime where L → ∞ extrapolation is under control <ref:2606. p. 1
Where the paper says it

Our main result, Eq. (25) below, allows one to check convergence explicitly for a given choice of smearing kernel and thereby identify the onset of the asymptotic regime, in which the L → ∞ extrapolation is under control.

14349#pg6>.

Relation to Spectral Reconstruction

The work introduces a framework for relating Euclidean correlators to smeared spectral densities, defining the smeared spectral density as ρbL,κ(α) = Z ∞ 0 dω κb(ω, α) ρL(ω) <ref:2606.14349#pg3>. They define coefficients ci(α) designed so that κbr(ωc(α)) approximates κb by approximating GL using sampled values of GL at τi <ref:2606. p. 1

Where the paper says it

2 Suppose that GL(τ ) is sampled at values τi for i ∈ {1, 2, · · · , N}. One can then determine a set of coefficients ci(α), designed so that κbr(ω|c(α)) = ω 2X N i=1 ci(α) e −ωτi , (6) approximates κb(ω, α), where the subscript r denotes the reconstructed version.

14349#pg8>. This leads to a statistical variance formula for the smeared spectral function and a systematic uncertainty formula related to the approximation of κb by κbr <ref:2606. p. 1
Where the paper says it

The goal is instead to quantify the systematic uncertainty associated with the approximation of κb by κbr , and to choose ci(α) such that this is subdominant to the statistical uncertainty on ρbL,κ,r . Denoting the covariance of GL(τi) by Σij , the statistical variance on the smeared spectral function is σ 2 stat = X N i,j=1 ci(α)Σij cj (α) ; (8) and a possible definition of the systematic uncertainty due to the approximation of κb by κbr is σsys = Z ∞ 0 dω h κb(ω, α) − κbr(ω|c(α))i ρL(ω) , = ρbL,κ(α) − ρbL,κ,r(c(α)) .

14349#pg9>.

Lellouch-L¨uscher-Meyer Formalism

An alternative approach relates finite-volume effects to the Lellouch and L¨uscher formalism, inserting a complete set of finite-volume energy eigenstates between current operators to yield GL(τ) = 1/3 X n cn(L) e−En(L)τ <ref:2606.14349#pg10>. This representation immediately yields a smeared density ρbL,κ(α) = 1/3 X∞ n=1 cn(L) En(L) 2 κb(En(L), α). p. 5

Where the paper says it

The states are normalized to unity. An advantage of this representation is that one can immediately write down the corresponding smeared spectral

The Lellouch-L¨uscher-Meyer approach relates finite-volume energies to infinite-volume scattering amplitudes through a quantization condition Q(En, L) = πn. p. 6
Where the paper says it

(31) To make progress we now use L¨uscher’s quantization condition [51, 52] which relates the energies En(L) to infinite-volume scattering amplitudes. More precisely, the finite-volume states that appear here transform as an irreducible representation, denoted by T − 1 , of the 48-element octahedral group (the appropriate symmetry group for states with zero total momentum in a cubic finite volume).

Numerical Evaluation and Asymptotic Scaling

Numerical results are presented for different form factor models (noninteracting, scattering volume, Gounaris-Sakurai) and smearing kernels (Cauchy kernel). The effective mass analysis highlights the common asymptotic behavior shared among the three form-factor parametrizations and across different values of τ, consistent with the expected 1/L e−mL scaling. For fixed ω⋆ and over the studied σ range, finite-volume effects are mainly governed by Cauchy-kernel poles, leading to an approximate exponential decay of the form ∼ exp(−meffL), with meff = 1/2 Imp(ω∗ + iσ) 2 − 4m2. p. 9

Where the paper says it

Cumulative finite-volume effects from shells with |n| ≤ 6 on the Cauchy-smeared spectral density, evaluated using the Gounaris-Sakurai form-factor parametrization (gs). The Cauchy smearing kernel is centred at ω ∗ = 3m, and results are shown for several values of the smearing width σ.

Conclusion and Future Directions

The work successfully derives a representation for the leading finite-volume effects in smeared spectral densities, showing agreement between the spectral reconstruction kernel approach and the Lellouch-L¨uscher-Meyer formalism. The framework suggests a broad range of future applications, such as deriving analogous finite-volume representations for any spectral density whose corresponding finite-volume matrix elements can be connected to infinite-volume amplitudes by a known Lellouch-L¨uscher-type relation. p. 6

Where the paper says it

The L¨uscher zeta function, Z00(1, q2 ) can be defined in a theoretically useful (though not numerically efficient) way via the regulated sum √ 4πZ00(1, q2 ) = lim Λ→∞ |nX |<Λ n 1 n2 − q 2 − 4πΛ , (36) where the sum runs over all triplets of integers n ∈ Z 3 . In addition to providing a relation between the scattering phase shift and the finite-volume energies, the same function Q(E, L) also relates finite- and infinite-volume matrix elements [49, 50, 53, 54].

The main utility will be in applications to lattice Monte Carlo data where explicit finite-volume formulae can be combined with spectral reconstruction methods to improve the control of the L → ∞ limit for phenomenologically relevant smeared spectral densities. p. 5
Where the paper says it

However, as the smeared spectral density is, in and of itself, a perfectly well-defined quantity, the expression for its finite-volume effects should be independent of the method used to reconstruct it. Therefore, the main result must hold for non-linear reconstruction methods as well.

Appendix A: Subdominance of the δGB L (τ) Term

The contribution from the δGB L(τ) term is shown to be subdominant to the leading finite-volume correction given by δGA L(τ). In the non-interacting limit, δGB L(τ) vanishes in the sense of a distribution as follows from R dzeikz = 2πδ(k). p. 0

Where the paper says it

INTRODUCTION Quantum field theory describes a rich range of physical systems, from the interactions of fundamental particles to various condensed matter and other material science applications. Interesting phenomenology is particularly visible in the non-perturbative regime, in which a series expansion about the non-interacting limit does not provide a useful description of the true dynamics.

The large-volume comparison in the interacting case shows that δGB L(τ) is subdominant to δGA L(τ) at large L, with the ratio scaling as 1/L as predicted. p. 0
Where the paper says it

The same representation is then derived starting from the Lellouch-L¨uscher-Meyer expression for the spectral decomposition of the correlator. The results may prove useful for controlling the L → ∞ extrapolation of smeared spectral densities, in particular by defining a scaling regime in which the finite-volume effects are dominated by the leading terms in a large L expansion and thus can be reliably estimated.

Appendix B: Proof of Symmetry

The proof in Appendix B demonstrates that the negative branch of the Wick-rotated contour gives an identical contribution to the positive branch, i.e., δGA−L(τ) = δGA+ L(τ). p. 2

Where the paper says it

Restricting to the pole contribution and following Refs. [40, 41], we note that it can be decomposed into two terms, δGpole L (τ ) = δGA L(τ ) + δGB L (τ ), (15) where δGA L(τ ) = 1 6 X n̸=0 Im Z R+iµ dk3 2π e −|n|L √ m2+k 2 3 /4 4πk3|n|L × e ik3|τ| (4m2 + k 2 3 )F(−k 2 3 ) 2 , (16) and δGB L (τ ) is again subdominant for realistic volumes and for a realistic functional form of the form factor, as discussed in Appendix A.

This symmetry is guaranteed by treating square-root functions consistently and using the principal branch of the square root. p. 2
Where the paper says it

(16) is shifted into the complex plane by an amount µ, with 0 < µ ≤ 2m. The lower bound is required to avoid the 1/k3 pole on the real axis,4 while the upper bound ensures that one does not cross the branch point of the square root function.

Appendix C: Properties of ϕ(q)

The large-L expansion of e 2iϕ(q) is derived, leading to Eq. (43) which shows that the Lellouch kinematic function grows linearly with L at fixed energy. p. 6

Where the paper says it

But in practice, the p-wave channel dominates the low-energy scattering of two pions with isospin 1. In the case that only this dominant scattering channel is included, the finite-volume energies are related to the corresponding scattering phase shift δππ(p(E)) according to the following quantization condition: Q(En, L) = πn , (32) where Q(E, L) = δππ(p(E)) + ϕ(q(E, L)), (33) cot ϕ(q) = − Z00(1, q2 ) qπ3/2 , (34) and we have introduced the kinematic variables p(E) = p E2/4 − m2 , q(E, L) = p(E)L 2π .

The positivity of Q(E, L) ensures that the sum over n in Eq. p. 0
Where the paper says it

The associated finite-volume spectral density is defined as ρL(ω) = − L 3 3ω2 X 3 k=1 ⟨0|jk(0)δPb ,0 δ(ω − Hb)jk(0)|0⟩L , (2) where Hb and Pb are the Hamiltonian and momentum operators, respectively. Here the Kronecker delta L 3 δPb ,0 = Z L 0 d 3x e iPb ·x , (3) ensures that only zero-momentum states contribute to the spectral density.

(39), as the nonpositive-n terms do not contribute. p. 6
Where the paper says it

In the second step we have used the property of the Dirac delta function under a change of variables to rewrite the first factor of the integrand. In addition we have extended the sum over n to all integers, noting that the non-positive-n terms do not contribute since Q(E, L) > 0 for all E > 2m.

p. 6
Where the paper says it

In the second step we have used the property of the Dirac delta function under a change of variables to rewrite the first factor of the integrand. In addition we have extended the sum over n to all integers, noting that the non-positive-n terms do not contribute since Q(E, L) > 0 for all E > 2m.

p. 6
Where the paper says it

In the second step we have used the property of the Dirac delta function under a change of variables to rewrite the first factor of the integrand. In addition we have extended the sum over n to all integers, noting that the non-positive-n terms do not contribute since Q(E, L) > 0 for all E > 2m.

The correspondence between the ground state and n = 1 corresponds to a near-threshold energy where δππ(p) = a1p cubed + O(p 5). p. 0
Where the paper says it

uk an observable of interest, e. by fitting to a single exponential at asymptotic separations in order to extract the ground state of the system for a particular set of quantum numbers.

Appendix C: Large Volume Expansion of e 2iϕ(q)

The large volume expansion of e 2iϕ(q) is given by Eq. (C10), which shows that the leading term at large L is captured by ρb(1) L,κ(α) = fκ(L, α) + ϵ(1) L,κ(α). p. 7

Where the paper says it

(42), we find that the leading behaviour at large L is captured by ρb (1) L,κ(α) = 1 3 1 16π 2 Z R+iµ dp e 2iδππ(p) 2ipL X n̸=0 e i|n|pL |n| + O(e 2ipL/(Lp) 2 ) p 4 (p 2 + m2) 2 F 4(p 2 + m2 ) 2 κb(E(p), α). (44) It follows that the leading finite-volume correction predicted by the Lellouch-L¨uscher-Meyer approach exactly matches the leading term in our main result of the previous section, Eq.

The ground state level non-interacting level is E1(L) = 2p(2π/L) squared + m squared. p. 0
Where the paper says it

to evaluate the Euclidean quantum path integral in a finite discretised spacetime using Monte Carlo importance sampling. This has been particularly impactful in the context of numerical lattice quantum chromodynamics (QCD), where it is applied to the strongly interacting sector of the Standard Model of Particle Physics, allowing for percent-level, first-principles predictions of quantities such as the proton and neutron masses [1] and the leading hadronic effects on the magnetic moment of the muon [2].

Appendix C: Positivity of Q(E, L)

The positivity of Q(E, L) ensures that the sum over n in Eq. (39), as the nonpositive-n terms do not contribute. This justifies extending the sum over n in Eq. p. 5

Where the paper says it

Future work is needed to determine whether this form is particularly useful, e. as a means of spectrally reconstructing the finite-volume correction itself or as a step towards extending these results to more complicated systems, such as those involving three or more particles.

(39), as the nonpositive-n terms do not contribute.

Appendix C: Correspondence between Ground State and n = 1

The correspondence between the ground state and n = 1 corresponds to the point where δππ(p(E)) = π − ϕ(q(E, L)). For large L this corresponds to a near-threshold energy, for which δππ(p) = a1p cubed + O(p 5). p. 6

Where the paper says it

But in practice, the p-wave channel dominates the low-energy scattering of two pions with isospin 1. In the case that only this dominant scattering channel is included, the finite-volume energies are related to the corresponding scattering phase shift δππ(p(E)) according to the following quantization condition: Q(En, L) = πn , (32) where Q(E, L) = δππ(p(E)) + ϕ(q(E, L)), (33) cot ϕ(q) = − Z00(1, q2 ) qπ3/2 , (34) and we have introduced the kinematic variables p(E) = p E2/4 − m2 , q(E, L) = p(E)L 2π .

The solution to the quantization condition at leading order in the large-L expansion is given by p = 2π/L − 24π squared a1/L 4 + O(1/L 6). p. 6
Where the paper says it

But in practice, the p-wave channel dominates the low-energy scattering of two pions with isospin 1. In the case that only this dominant scattering channel is included, the finite-volume energies are related to the corresponding scattering phase shift δππ(p(E)) according to the following quantization condition: Q(En, L) = πn , (32) where Q(E, L) = δππ(p(E)) + ϕ(q(E, L)), (33) cot ϕ(q) = − Z00(1, q2 ) qπ3/2 , (34) and we have introduced the kinematic variables p(E) = p E2/4 − m2 , q(E, L) = p(E)L 2π .

The correspondence between the ground state and n = 1 corresponds to the point where δππ(p(E)) = π − ϕ(q(E, L)) <ref:2606.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. More accurate calculation of finite-volume corrections for lattice QCD observables by using Eq (25) to check convergence explicitly for a given smearing kernel and identify the onset of the asymptotic regime where the L → ∞ extrapolation is under control.

  2. Improved control over long-distance physics in spectral density analysis by using the relationship between Euclidean correlators and smeared spectral densities, specifically defining a scaling regime where the leading finite-volume effects are dominated by the leading terms in a large L expansion.

  3. Enhanced reliability of lattice QCD predictions for quantities like the anomalous magnetic moment of the muon by combining explicit finite-volume formulae with spectral reconstruction methods to improve the control of the L → ∞ limit for phenomenologically relevant smeared spectral densities.

  4. Derivation of finite-volume representations for multi-hadron D decays by applying this framework, allowing researchers to derive analogous finite-volume representations for any spectral density whose corresponding finite-volume matrix elements can be connected to infinite-volume amplitudes by a known Lellouch–Lüscher–type relation.

Abstract

Using two distinct approaches, we derive a universal expression for the leading finite-volume effects of the smeared vector-vector spectral density (proportional to the smeared hadronic R-ratio) in a periodic cubic spatial volume of side length L. First, building on the results of previous work for finite-volume effects on Euclidean two-point functions, we show that the L dependence is exponentially suppressed for a certain class of smearing kernels, and that the leading effects can be expressed universally in terms of the pion form factor. The same representation is then derived starting from the Lellouch-Lüscher-Meyer expression for the spectral decomposition of the correlator. The results may prove useful for controlling the L to infinity extrapolation of smeared spectral densities, in particular by defining a scaling regime in which the finite-volume effects are dominated by the leading terms in a large L expansion and thus can be reliably estimated. To illustrate this point, we also present numerical estimates based on various kernels and models of particle interactions. Despite focusing on the vector channel, our derivation defines a general framework applicable to other cases as well.

Sources

Related papers