arxiv: 2604.19860 · v1 · submitted 2026-04-21 · ✦ hep-th

Recognition: unknown

Mutual Information from Modular Flow in General CFTs

C\'esar A. Ag\'on , Pablo Bueno , Adem Deniz Piskin , Guido van der Velde

Authors on Pith no claims yet

Pith reviewed 2026-05-10 01:27 UTC · model grok-4.3

classification ✦ hep-th

keywords mutual informationconformal field theorymodular flowtwist operatorsoperator product expansionreplica symmetryentanglement measures

0 comments

The pith

Modular flow and twist operator expansions yield a hierarchy of approximations to the vacuum mutual information in any CFT.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a systematic way to approximate the mutual information between subregions in a general conformal field theory by combining the geometric modular flow of ball-shaped regions with an operator product expansion for the twist operators that implement replica symmetry. Two-point functions of primary operators with arbitrary spin in the replicated theory are used to determine the twist operator contributions to the mutual information of arbitrarily boosted balls. When the lowest-scaling-dimension primary is included, this produces the most accurate long-distance expansion available. Combining the long-distance result with universal short-distance behavior then gives a new analytic formula that works for any separation. The formula is checked against exact two-dimensional results and three-dimensional lattice data before being applied to the four-dimensional Maxwell field.

Core claim

The vacuum mutual information of subregion algebras in general CFTs can be approximated through a hierarchy of increasingly refined expressions constructed from the modular flow of ball-shaped regions and the operator product expansion of the twist operators in the replica theory. The two-point functions of primaries of arbitrary spin constrain these twist operators and determine their contribution to the mutual information of arbitrarily boosted balls in any dimension. The leading contribution from the primary with the lowest scaling dimension gives the most accurate long-distance expansion, and combining this with universal short- and long-distance properties produces a high-precision 3d=2

What carries the argument

The operator product expansion of twist operators in the n-fold replicated CFT, constrained by two-point functions of primary operators of arbitrary spin, which determines their contribution to the mutual information via modular flow of ball-shaped regions.

If this is right

The result from the lowest-dimension primary supplies the most precise long-distance expansion for the mutual information in any CFT.
A single analytic expression now exists that approximates the mutual information to high precision for arbitrary separations in any dimension.
The approximation reproduces exact results in two dimensions and lattice results in three dimensions.
The same expression can be evaluated for a four-dimensional Maxwell field, a case with no prior analytic or numerical results.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Similar modular-flow expansions could be applied to other entanglement measures such as Rényi entropies in the same CFTs.
Because low-dimension primaries control the long-distance regime, the mutual information between distant regions may depend only on a few universal CFT data.
The method could be tested by comparing its predictions for boosted balls against direct replica computations in a solvable CFT at intermediate distances.
In holographic duals the approximation might translate into new constraints on bulk geometry or correlation functions.

Load-bearing premise

The contribution from the lowest scaling dimension primary dominates the long-distance mutual information, the hierarchy of modular flow and OPE approximations converges to the full value, and universal short- and long-distance properties are sufficient to interpolate accurately.

What would settle it

A lattice computation of mutual information at intermediate distances in a three-dimensional CFT that deviates substantially from the new analytic approximation at several points would show the interpolation is inaccurate.

Figures

Figures reproduced from arXiv: 2604.19860 by Adem Deniz Piskin, C\'esar A. Ag\'on, Guido van der Velde, Pablo Bueno.

**Figure 1.** Figure 1: FIG. 1. Approximations to the MI of pairs of parallel disjoint balls for various free theories. In all cases, [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗

read the original abstract

The vacuum mutual information (MI) of subregion algebras provides a universal window into the data of general conformal field theories (CFTs). Exploiting the geometric nature of the modular flow associated to ball-shaped regions and the operator product expansion of twist operators implementing the replica symmetry in an $n$-fold version of a CFT, it is possible to construct a hierarchy of increasingly refined approximations to the full MI. In this letter, we use the two-point functions of primaries of arbitrary spin in the replicated theory to constrain the twist operators, and find their contribution to the MI of arbitrarily boosted balls in any $d$-dimensional CFT. When the two-point functions involve the primary with the lowest scaling dimension, our result provides the most precise approximation for the long-distance behavior of the MI, superseding all previous expansions. Building upon this result and certain universal properties of the short- and long-distance regimes, we put forward a new high-precision analytic approximation to the MI for arbitrary separations. The accuracy of our approach is validated against exact $d=2$ and lattice $d=3$ results. We further apply it to characterize the MI of a $d=4$ Maxwell field, a case for which no prior results are available.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This paper gives a usable analytic approximation for mutual information between boosted balls in any d by combining modular flow with the replica OPE expansion fixed by two-point functions of arbitrary-spin primaries.

read the letter

The central advance is a hierarchy of approximations to the mutual information that starts from the geometric modular flow of ball regions and expands the twist operators in the n-fold replicated CFT. Two-point functions of primaries with any spin fix the coefficients, so the lightest primary supplies the leading long-distance term because its correlator decays slowest. They then combine that tail with the known short-distance expansion using universal features of the MI to produce a closed-form expression valid at arbitrary separations. The long-distance piece improves on earlier expansions when the lowest-dimension primary is used, and the checks against the exact d=2 result and d=3 lattice data are direct and reassuring. The application to the d=4 Maxwell field is also concrete, since no prior analytic handle existed there. The arbitrary-spin extension is new relative to previous scalar-only treatments. The interpolation step between short and long distances is the softest part. It assumes the OPE hierarchy converges fast enough and that the universal short- and long-distance properties are sufficient to control the intermediate regime without large uncontrolled errors. No proof of convergence is given, but the low-dimensional tests provide some empirical support. For the Maxwell case the numbers are therefore best viewed as high-quality estimates rather than exact values. This is for people who compute entanglement quantities in CFTs and holography and need something more practical than pure numerics or crude expansions. A reader who wants explicit formulas and checks against known cases will get immediate value. The work is grounded enough in standard CFT tools and produces falsifiable outputs, so it deserves a serious referee.

Referee Report

1 major / 2 minor

Summary. The manuscript develops a hierarchy of approximations to the vacuum mutual information (MI) of ball-shaped subregions in general d-dimensional CFTs. It combines the geometric modular flow of these regions with the OPE expansion of replica twist operators in the n-fold replicated theory, using two-point functions of primary operators of arbitrary spin to fix leading coefficients. This yields an improved long-distance expansion for the MI dominated by the lowest-dimension primary, which is then matched to the short-distance regime via universal MI properties (boost invariance and divergence structure) to produce a global analytic approximation for arbitrary separations. The result is validated against exact d=2 results and d=3 lattice data, and applied to the d=4 Maxwell field.

Significance. If the central approximation holds, the work supplies a parameter-free, high-precision analytic expression for MI in arbitrary CFTs that leverages only standard CFT data (two-point functions and modular geometry) without fitted parameters or additional dynamical assumptions. The explicit checks against solvable cases and the extension to a previously uncomputed theory (Maxwell) are strengths, as is the systematic use of the OPE hierarchy and universal short/long-distance properties. This could become a useful tool for entanglement studies in higher-dimensional CFTs.

major comments (1)

[The construction of the global analytic approximation (around the matching of short- and long-distance regimes)] The global analytic approximation for arbitrary separations is constructed by interpolating the long-distance tail (from the lowest-Δ primary two-point function) with the short-distance expansion using universal MI properties. However, the manuscript does not provide a rigorous demonstration that the chosen interpolating function is uniquely determined by these properties or that it controls the truncation error of the OPE hierarchy for general d; explicit error bounds or additional cross-checks beyond d=2 would be needed to support the claim of 'high-precision' for arbitrary separations.

minor comments (2)

[Abstract] The abstract states that the long-distance result 'supersedes all previous expansions'; a brief explicit comparison to the leading terms of prior long-distance MI expansions in the literature would strengthen this claim.
Notation for the replicated twist operators and their OPE coefficients could be introduced with one additional sentence for readers less familiar with the replica trick in higher-dimensional CFTs.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful reading of our manuscript and for the positive overall assessment of the work. We address the major comment below.

read point-by-point responses

Referee: The global analytic approximation for arbitrary separations is constructed by interpolating the long-distance tail (from the lowest-Δ primary two-point function) with the short-distance expansion using universal MI properties. However, the manuscript does not provide a rigorous demonstration that the chosen interpolating function is uniquely determined by these properties or that it controls the truncation error of the OPE hierarchy for general d; explicit error bounds or additional cross-checks beyond d=2 would be needed to support the claim of 'high-precision' for arbitrary separations.

Authors: We thank the referee for this observation. The interpolating function is chosen as the minimal analytic expression that exactly reproduces the short-distance expansion (including its universal divergence structure and boost invariance) together with the leading long-distance asymptotics fixed by the lowest-dimension primary. While the universal properties constrain the form without introducing free parameters, they do not furnish a rigorous uniqueness proof for intermediate separations. The OPE hierarchy is systematic, so the leading term controls the dominant error at large separations where higher contributions are exponentially suppressed. The manuscript already validates the approximation against exact d=2 results and lattice d=3 data; the latter constitutes an explicit cross-check beyond d=2. We have revised the text to clarify the motivation for the interpolant and to emphasize the d=3 validation. Deriving explicit error bounds for arbitrary d would require a non-perturbative estimate of the full OPE remainder, which lies outside the scope of this letter. revision: yes

Circularity Check

0 steps flagged

No significant circularity: derivation uses standard CFT kinematics and external validations

full rationale

The paper constructs approximations to mutual information by combining the geometric modular flow for ball regions with the OPE expansion of replica twist operators, fixing coefficients via two-point functions of primaries in the replicated theory. The leading long-distance term follows directly from the slowest power-law decay 1/r^{2Δ_min} of the lowest-dimension primary, which is a standard consequence of CFT operator dimensions and does not involve fitting or self-referential definitions. The global approximation then interpolates this tail to the short-distance regime using only universal MI properties (boost invariance and divergence structure). Explicit comparisons to exact d=2 results and independent d=3 lattice data, plus application to the d=4 Maxwell field, provide external checks. No step reduces by construction to the paper's own inputs, no load-bearing self-citation chain appears, and the central claims remain independent of any ansatz smuggled via prior work.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The paper builds on standard axioms of conformal field theory and replica trick methods without introducing new free parameters or entities. The inputs are the two-point functions of primaries, which are CFT data.

axioms (2)

standard math Conformal invariance and the existence of a well-defined OPE for twist operators in the replicated CFT.
Invoked throughout the construction of the hierarchy from modular flow and twist operator constraints.
domain assumption Universal properties of short- and long-distance regimes for MI in CFTs.
Used to build the high-precision approximation for arbitrary separations.

pith-pipeline@v0.9.0 · 5529 in / 1510 out tokens · 48331 ms · 2026-05-10T01:27:42.971392+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

42 extracted references · 39 canonical work pages

[1]

Casini,Entropy localization and extensivity in the semiclassical black hole evaporation,Phys

H. Casini,Entropy localization and extensivity in the semiclassical black hole evaporation,Phys. Rev. D79(2009) 024015 [0712.0403]

work page arXiv 2009
[2]

Casini and M

H. Casini and M. Huerta,Remarks on the entanglement entropy for disconnected regions, JHEP03(2009) 048 [0812.1773]

work page arXiv 2009
[3]

Casini, F

H. Casini, F. D. Mazzitelli and E. Test´ e,Area terms in entanglement entropy,Phys. Rev. D91 (2015), no. 10 104035 [1412.6522]

work page arXiv 2015
[4]

Casini, M

H. Casini, M. Huerta, R. C. Myers and A. Yale, Mutual information and the F-theorem,JHEP10 (2015) 003 [1506.06195]

work page arXiv 2015
[5]

Bueno, H

P. Bueno, H. Casini, O. L. Andino and J. Moreno, Disks globally maximize the entanglement entropy in 2 + 1 dimensions,JHEP10(2021) 179 [2107.12394]

work page arXiv 2021
[6]

Huerta and G

M. Huerta and G. van der Velde,Instability of universal terms in the entanglement entropy,Phys. Rev. D105(2022), no. 12 125021 [2204.09464]

work page arXiv 2022
[7]

Bueno, H

P. Bueno, H. Casini, O. L. Andino and J. Moreno, Conformal Bounds in Three Dimensions from Entanglement Entropy,Phys. Rev. Lett.131 (2023), no. 17 171601 [2307.05164]

work page arXiv 2023
[8]

Bueno, A

P. Bueno, A. F. Garc´ ıa, F. Gentile, O. Lasso Andino and J. Moreno,Entanglement entropy and conformal bounds ford= 5CFTs, 2604.01436

work page arXiv
[9]

Headrick,Entanglement Renyi entropies in holographic theories,Phys

M. Headrick,Entanglement Renyi entropies in holographic theories,Phys. Rev. D82(2010) 126010 [1006.0047]

work page arXiv 2010
[10]

Calabrese, J

P. Calabrese, J. Cardy and E. Tonni, Entanglement entropy of two disjoint intervals in conformal field theory II,J. Stat. Mech.1101 (2011) P01021 [1011.5482]

work page arXiv 2011
[11]

Long,On co-dimension two defect operators, 1611.02485

J. Long,On co-dimension two defect operators, 1611.02485

work page arXiv
[12]

Chen and J

B. Chen and J. Long,R´ enyi mutual information for a free scalar field in even dimensions,Phys. Rev. D96(2017), no. 4 045006 [1612.00114]

work page arXiv 2017
[13]

B. Chen, L. Chen, P.-x. Hao and J. Long,On the Mutual Information in Conformal Field Theory, JHEP06(2017) 096 [1704.03692]

work page arXiv 2017
[14]

Cardy,Some results on the mutual information of disjoint regions in higher dimensions,J

J. Cardy,Some results on the mutual information of disjoint regions in higher dimensions,J. Phys. A 46(2013) 285402 [1304.7985]

work page arXiv 2013
[15]

C. A. Ag´ on, I. Cohen-Abbo and H. J. Schnitzer, Large distance expansion of Mutual Information for disjoint disks in a free scalar theory,JHEP11 (2016) 073 [1505.03757]

work page arXiv 2016
[16]

Ag´ on and T

C. Ag´ on and T. Faulkner,Quantum Corrections to Holographic Mutual Information,JHEP08(2016) 118 [1511.07462]

work page arXiv 2016
[17]

C. A. Ag´ on, P. Bueno and H. Casini,Is the EMI model a QFT? An inquiry on the space of allowed entropy functions,JHEP08(2021) 084 [2105.11464]

work page arXiv 2021
[18]

Casini, E

H. Casini, E. Test´ e and G. Torroba,Mutual information superadditivity and unitarity bounds, JHEP09(2021) 046 [2103.15847]

work page arXiv 2021
[19]

C. A. Ag´ on, P. Bueno and H. Casini,Tripartite information at long distances,SciPost Phys.12 (2022), no. 5 153 [2109.09179]

work page arXiv 2022
[20]

C. A. Ag´ on, P. Bueno, O. Lasso Andino and A. Vilar L´ opez,Aspects of N-partite information in conformal field theories,JHEP03(2023) 246 [2209.14311]

work page arXiv 2023
[21]

C. A. Ag´ on, P. Bueno and G. van der Velde, Long-distance N-partite information for fermionic CFTs,JHEP12(2024) 178 [2409.03821]

work page arXiv 2024
[22]

C. A. Agon, H. Casini, U. G¨ ursoy and 12 G. Planella Planas,Mutual information from modular flow in CFTs,JHEP08(2025) 176 [2409.01406]

work page arXiv 2025
[23]

Towards a derivation of holographic entanglement entropy

H. Casini, M. Huerta and R. C. Myers,Towards a derivation of holographic entanglement entropy, JHEP05(2011) 036 [1102.0440]

work page Pith review arXiv 2011
[24]

P. D. Hislop and R. Longo,Modular Structure of the Local Algebras Associated With the Free Massless Scalar Field Theory,Commun. Math. Phys.84(1982) 71

1982
[25]

Calabrese and J.L

P. Calabrese and J. L. Cardy,Entanglement entropy and quantum field theory,J. Stat. Mech. 0406(2004) P06002 [hep-th/0405152]

work page arXiv 2004
[26]

Calabrese and J

P. Calabrese and J. L. Cardy,Entanglement entropy and quantum field theory: A Non-technical introduction,Int. J. Quant. Inf.4(2006) 429 [quant-ph/0505193]

work page arXiv 2006
[27]

L.-Y. Hung, R. C. Myers and M. Smolkin,Twist operators in higher dimensions,JHEP10(2014) 178 [1407.6429]

work page arXiv 2014
[28]

Casini, C

H. Casini, C. D. Fosco and M. Huerta, Entanglement and alpha entropies for a massive Dirac field in two dimensions,J. Stat. Mech.0507 (2005) P07007 [cond-mat/0505563]

work page arXiv 2005
[29]

Notes on Some Entanglement Properties of Quantum Field Theory

E. Witten,APS Medal for Exceptional Achievement in Research: Invited article on entanglement properties of quantum field theory, Rev. Mod. Phys.90(2018), no. 4 045003 [1803.04993]

work page Pith review arXiv 2018
[30]

Casini and M

H. Casini and M. Huerta,Lectures on entanglement in quantum field theory,PoS T ASI2021(2023) 002 [2201.13310]

work page arXiv 2023
[31]

(2), and s≡iτ lk

Note that this definition is in agreement with the KMS condition, which leads to the relation Gαβ(x1, x2,−s) = (−1) F Gαβ(x1, x2, s), where Gαβ(x1, x2, s) is the left hand side of Eq. (2), and s≡iτ lk. Moreover,F= 1 for fermions whileF= 0 for bosons. Sinces→ −scorresponds to the exchangel→k, this reflects the symmetric (antisymetric) nature of bosonic (fe...
[32]

Hirai,The characters of irreducible representations of the lorentz group ofn-th order, Proceedings of The Japan Academy41(01, 1965)

T. Hirai,The characters of irreducible representations of the lorentz group ofn-th order, Proceedings of The Japan Academy41(01, 1965)

1965
[33]

Casini and M

H. Casini and M. Huerta,Reduced density matrix and internal dynamics for multicomponent regions, Class. Quant. Grav.26(2009) 185005 [0903.5284]

work page arXiv 2009
[34]

R. E. Arias, H. Casini, M. Huerta and D. Pontello, Entropy and modular Hamiltonian for a free chiral scalar in two intervals,Phys. Rev. D98(2018), no. 12 125008 [1809.00026]

work page arXiv 2018
[35]

Czech, L

B. Czech, L. Lamprou, S. McCandlish, B. Mosk and J. Sully,A Stereoscopic Look into the Bulk, JHEP07(2016) 129 [1604.03110]

work page arXiv 2016
[36]

Simmons-Duffin,Projectors, Shadows, and Conformal Blocks,JHEP04(2014) 146 [1204.3894]

D. Simmons-Duffin,Projectors, Shadows, and Conformal Blocks,JHEP04(2014) 146 [1204.3894]

work page arXiv 2014
[37]

Casini and M

H. Casini and M. Huerta,Entanglement and alpha entropies for a massive scalar field in two dimensions,J. Stat. Mech.0512(2005) P12012 [cond-mat/0511014]

work page arXiv 2005
[38]

Abate, H

N. Abate, H. Casini, M. Huerta and L. Martinek, Exact Mutual Information Difference: Scalar vs. Maxwell Fields,2511.04742

work page arXiv
[39]

Benedetti, H

V. Benedetti, H. Casini and P. J. Martinez,Mutual information of generalized free fields,Phys. Rev. D 107(2023), no. 4 046003 [2210.00013]

work page arXiv 2023
[40]

F. A. Dolan and H. Osborn,Conformal four point functions and the operator product expansion,Nucl. Phys. B599(2001) 459–496 [hep-th/0011040]

work page Pith review arXiv 2001
[41]

I. Peschel,LETTER TO THE EDITOR: Calculation of reduced density matrices from correlation functions,Journal of Physics A Mathematical General36(Apr., 2003) L205–L208 [cond-mat/0212631]

work page arXiv 2003
[42]

Entanglement entropy in free quantum field theory

H. Casini and M. Huerta,Entanglement entropy in free quantum field theory,J. Phys. A42(2009) 504007 [0905.2562]

work page Pith review arXiv 2009