O(N) BCFT: new data from conformal partial wave expansions

arXiv:2606.14733 · hep-th, cond-mat.str-el · Submitted 2026-06-02 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: High Energy Physics: Theory. The day in high-energy theory: gravity, strings, fields and the questions left open.

Kai: We're looking at the paper: "O(N) BCFT: new data from conformal partial wave expansions".

Mira: Here's the gist of the paper. The O(N) BCFT is analyzed in the large N expansion in a generic bulk dimension 2 < d + 1 < 4 to derive complete bulk/boundary conformal block decompositions of two simplest bulk-bulk 2-pt functions, and to extract as much Boundary Conformal Field Theory data as possible The O(N) BCFT is analyzed in the large N expansion in a generic bulk dimension 2 < d + 1 < 4 to derive complete bulk/boundary conformal block decompositions of two simplest bulk-bulk 2-pt functions, and to extract as much Boundary Conformal Field Theory data as possible.

Kai: Boundary Data Extraction The study focuses on extracting BCFT data from the expansion coefficients of bulk/boundary conformal block decompositions of the two simplest bulk-bulk 2-pt functions, specifically by analyzing the boundary spectral functions The authors prove conjectures about boundary spectral functions in Dujava et al.

1: and reproduce the bulk expansion of ⟨ϕϕ⟩ in Giombi et al.

2: .

Paper summary: Kai: So we're looking at this paper today titled "O(N) BCFT: new data from conformal partial wave expansions". Essentially, they are taking the O(N) Gross-Neveu model with a boundary and analyzing it in the large N expansion in a generic bulk dimension where two less than d plus one less than four.

Mira: Right, so the main thesis is that they use this large N expansion to get complete bulk and boundary conformal block decompositions for the two simplest bulk two point functions, which are phi phi and sigma sigma.

Lev: So what they claim is that by expanding these functions in terms of bulk and boundary conformal partial waves, they can extract as much Boundary Conformal Field Theory data as possible from the coefficients of that expansion.

Kai: It sounds like they are using these expansions to basically map out the BCFT data, which is really about understanding physics at the edge of a system.

Mira: Exactly. They focus on extracting BCFT data by analyzing the boundary spectral functions that come out of those bulk partial wave decompositions and they prove some conjectures about those spectral functions in Dujava et al. one <ref:2606.14733#pg1>.

Lev: And they also reproduce the bulk expansion of phi phi, which is something Giombi et al. two did, but they go further by deriving a new prediction for an unknown OPE coefficient called C sigma sigma three, which matches a result from Derkachov et al <ref:2606.14733#pg1>. three <ref:2606.14733#pg2>.

Kai: So we're talking about extracting specific numbers and coefficients that are usually hard to find directly in the boundary theory by looking at these bulk expansions.

Mira: And on the bulk side, they compute spectral functions associated with phi phi and sigma sigma from their coefficients of those bulk conformal block decompositions, and they verify it by plotting the sequence of partial bulk block sums.

Lev: They found a unique infinite tower of bulk primaries contributing to both the phi times phi and sigma times sigma OPE in this analysis, which is quite a detailed finding for the structure of the theory.

Kai: That infinite tower sounds like a lot to deal with when you're trying to pin down the physics of how these operators interact.

Mira: For the phi phi correlator, they extracted an OPE coefficient called C phi phi sigma and successfully matched it against a known result, which is pretty strong verification of their method here.

Lev: The analysis also leads to predictions for several unknown OPE coefficients, like C sigma sigma two sixty-seven, which they predict to be equal to the square root of two times five point three seven.

Kai: And they also predict the anomalous dimension of sigma squared, which they state is predicted by the calculation in equation five point three eight <ref:2606.14733#pg2>.

Paper summary: Mira: The convergence of this boundary expansion is shown to be fast across different dimensions, uniformly on any interval rho between epsilon and one where epsilon is greater than zero and doesn't include zero.

Lev: They suggest looking into running conformal perturbation theory in reverse, and extending these computations to the next large N order as future directions.

Kai: So they're saying that this method works pretty well regardless of the dimension you pick, which is important because it gives us a consistent way to extract boundary information.

Mira: And in appendix A, they derive a general relation between the coefficient of a bulk and boundary conformal block and actual BCFT data for any two point function O1 O2.

Lev: That relation looks like this: A equals two minus delta one minus delta two plus delta three C O1 O2 O3 C3 C3 triple sigma, which is a bit complex but it gives us a way to connect the bulk expansion coefficients to the BCFT data.

Kai: It seems like they're providing a systematic tool for linking different descriptions of the same physics, moving from bulk expansions to boundary data.

Mira: They also bootstrapped the boundary spectral function for delta sigma delta sigma using bootstrap arguments, which leads to a specific expression for Spec partial delta sigma delta sigma that depends on the displacement operator.

Lev: That final spectral function is determined to be Spec partial delta sigma delta sigma equals one over pi of tau b minus one times tan pi over two times tau b times tau b, where tau b is the scaling dimension.

Kai: So they pinned down that boundary spectral function by linking its zeroes to the poles of the phi two phi two function, which fixes what we're trying to bootstrap.

Mira: And this is connected via a Ward identity to the one point function sigma, which gives them a concrete result for C minus one half sigma sigma three, predicted as negative tau zero times five point three seven times N FP sigma.

Lev: The norm of the displacement operator two point function CDb Db is found to be equal to eight minus d times d minus three times d minus one squared pi over three minus gamma of d plus half over gamma of two d minus two cos pi over two times pi power d/two gamma of d plus one over gamma of forty nine <ref:2606.14733#pg1>.

Kai: That formula for the norm seems really specific, and they say it precisely matches results from reference twenty-four and also reference two <ref:2606.14733#pg2>.

Mira: They also derived a prediction for the boundary limit of the connected sigma propagator, which is dominated by the boundary operator of lowest scaling dimension in the BOE of delta sigma.

Paper summary: Lev: That prediction for it is C minus one half sigma sigma times N over two times d plus one equals B minus d Db two C Db Db, which gives us a way to estimate that boundary limit.

Kai: So they're connecting the bulk structure, through these spectral functions, to specific physical observables at the boundary.

Mira: The bulk OPE structure for phi times phi and sigma times sigma is inferred directly from those bulk conformal block expansions they computed earlier.

Lev: The leading operator contributing to the phi times phi OPE is sigma, and its squared coefficient C minus one half phi phi sigma squared equals Ope N of Op of S zero one, which verifies that hypothesis from point five point four in the paper.

Kai: It seems like they're confirming a structure they suspected by matching it against the bulk expansion coefficients.

Mira: The subleading bulk conformal block expansion for the modified connected correlator has coefficients C and D that are complicated functions expressed in terms of derivatives of four F three hypergeometric functions whose arguments depend only on the space dimension d and scaling dimensions delta k equal to two k where k is greater than or equal to two.

Lev: The convergence analysis shows that subsequent coefficients change sign due to oscillatory behavior, which means you can't just use standard numerical bootstrap algorithms because they need positive expansion coefficients.

Kai: So the oscillatory behavior in the bulk expansion prevents us from using some straightforward numerical techniques to check for positivity.

Mira: They also predict that both the phi times phi OPE and the sigma times sigma OPE contain an infinite tower of operators of a specific schematic form, which is something important because it tells us about the full spectrum.

Lev: For the operator k equals one sigma, it contributes only to the subleading term if d isn't equal to two, because its coefficient in that leading term vanishes due to some identity.

Kai: So for that specific operator, its contribution is suppressed at leading order unless you're in a certain dimension.

Mira: The contribution of the one sigma operator via the bulk channel sigma times sigma OPE with the coefficient of its block gives us a prediction involving two minus delta zero minus delta zero plus delta zero times two minus one half plus delta zero times three over two and kappa over sixteen, which simplifies to negative N FP sigma d minus one d minus two divided by four.

Lev: For the operator k equals two sigma squared, it participates in both the leading and subleading equations because it appears in both equation five point three zero and five point three two.

Kai: So for that term, it's active at both orders of approximation we're looking at here.

Paper summary: Mira: At leading order, using the standard strategy of matching the bulk OPE with the computed coefficient of the block and large N factorization gives us a prediction involving two minus delta zero times sigma minus delta zero times sigma plus delta zero times sigma squared over one and then then something that looks like tau b squared at zero.

Lev: For k equals three operators, like sigma three or sigma boxsigma, they enter both the leading part of the correlator and the subleading one, with undifferentiated blocks involving CFT data that needs to be evaluated at a subleading order.

Kai: So for these higher order terms, we don't have all the necessary information yet because we need to evaluate some things at a lower level of approximation.

Mira: They do give us a prediction for the unknown OPE coefficient C minus one half sigma sigma three, which they state is equal to q three factorial N times five point four times sigma squared over d minus one d minus two over sixteen times negative four q six pi d minus two r of d minus three cos pi over two times pi power d/two gamma of five minus d over forty nine <ref:2606.14733#pg1>.

Lev: The mixing coefficient b minus one half one is expressed in terms of the displacement operator norm, giving a complicated expression involving tau zero times three over two and A one sigma bracket tau zero times one bracket sigma one zero bracket tau zero ones bracket sigma one zero divided by four N FP sigma C C0 sigma six seven minus tau omega bracketssigma ones zero divided by four N FP.

Kai: It seems like the complexity of these higher order terms is really increasing, leading to these very complicated expressions for the coefficients.

Mira: The overall picture here is that they're using this method to systematically map out how bulk expansions connect to boundary data through those relations and bootstrap methods.

Lev: The paper's limitation, as stated in the text, is that the convergence analysis shows oscillatory behavior in subsequent coefficients, which prevents the ordinary transition from being investigated by numerical bootstrap algorithms because they require positivity of expansion coefficients.

Kai: So basically, the method is very powerful for deriving these specific predictions and relating bulk structure to boundary physics under certain conditions but it has limitations when trying to use general numerical methods that demand positive numbers.

Mira: They suggest running conformal perturbation theory in reverse and extending the computations to the next large N order as future work.

Lev: So basically, they're showing a path forward by suggesting how to build on this analysis for even more complete results without relying on those positivity constraints.

Conclusion: Kai: So we're wrapping up this look at the paper titled "O(N) BCFT: new data from conformal partial wave expansions."

Mira: It’s about taking this large N expansion and using it to extract concrete boundary data from bulk calculations of two simple point functions, phi phi and sigma sigma.

Lev: Essentially, they're connecting the structure we see in the middle of a system to what happens right at the edge.

Kai: What that means is they’re using these complex mathematical expansions—the conformal block decompositions—to get actual numbers for boundary physics that we usually can’t measure directly.

Mira: Right, and they did this by analyzing those spectral functions coming out of the bulk analysis and proving some things about them in Dujava et al.

Lev: The main thing I see is how they connect these bulk expansions to real boundary observables like the displacement operator norm, which they compare against known results.

Kai: They also predicted some new OPE coefficients, like C sigma sigma three, which are usually just theoretical guesses until you have experimental data to check them against.

Mira: So for someone listening who doesn't know CFT, it means they’ve built a bridge from the deep inside of the system to the surface physics using these large N tools.

Lev: The paper flags that while the convergence is fast, those subsequent coefficients in their expansion oscillate, which means we can't just use simple numerical bootstrap methods that demand positive numbers.

Kai: So what they’re suggesting next is running this kind of perturbation theory backwards and pushing it to the next large N order to get more reliable results.

Mira: It’s a lot of intricate work, but if this works consistently across different dimensions, it opens up a new way to probe boundary effects in these strongly coupled systems.

Jozef Csipesa, Petr Vaškob

Institute of Theoretical Physics, Charles University · Institute of Particle and Nuclear Physics, Charles University · Institute of Physics of the Czech Academy of Sciences

hep-th, cond-mat.str-el

Submitted: 2026-06-02

Updated: 2026-11-16

Comments: version accepted for publication in JHEP

Journal ref: JHEP 11 (2026) 086

DOI: 10.1007/JHEP11(2026)086

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

Importance score: 81/100

The gist: The O(N) BCFT is analyzed in the large N expansion in a generic bulk dimension 2 < d + 1 < 4 to derive complete bulk/boundary conformal block decompositions of two simplest bulk-bulk 2-pt functions…

Key concepts

O(N) BCFT
This is a framework used to study Quantum Field Theories in the large N limit, specifically focusing on the properties of operators in theories with an O(N) symmetry. It provides a systematic way to decompose complex correlation functions into simpler building blocks (conformal blocks).
Conformal Block Decomposition
This is a mathematical tool that breaks down complicated correlation functions into a sum of simpler, fundamental components called conformal blocks. These blocks are determined by the underlying conformal symmetry of the theory and are crucial for understanding how fields interact in different regimes.
Boundary Spectral Function
This function describes the behavior of correlation functions at the boundary of spacetime. In this paper, it is derived using bootstrap arguments to predict its exact form, which helps extract specific data points from the BCFT analysis.

Terminology

Summary

The O(N) BCFT is analyzed in the large N expansion in a generic bulk dimension 2 < d + 1 < 4 to derive complete bulk/boundary conformal block decompositions of two simplest bulk-bulk 2-pt functions, and to extract as much Boundary Conformal Field Theory data as possible

Summary of Results

The O(N) BCFT is analyzed in the large N expansion in a generic bulk dimension 2 < d + 1 < 4 to derive complete bulk/boundary conformal block decompositions of two simplest bulk-bulk 2-pt functions, and to extract as much Boundary Conformal Field Theory data as possible.

Boundary Data Extraction

The study focuses on extracting BCFT data from the expansion coefficients of bulk/boundary conformal block decompositions of the two simplest bulk-bulk 2-pt functions, specifically by analyzing the boundary spectral functions The authors prove conjectures about boundary spectral functions in Dujava et al. [1] and reproduce the bulk expansion of ⟨ϕϕ⟩ in Giombi et al. p. 0

Where the paper says it

We prove conjectures about boundary spectral functions in Dujava et al. [1] and reproduce the bulk expansion of ⟨ϕϕ⟩ in Giombi et al.

[2]. They also derive a new prediction for the unknown OPE coefficient Cσσσ3, which is a byproduct of the analysis and matches an existing result by Derkachov et al. p. 0
Where the paper says it

[1] and reproduce the bulk expansion of ⟨ϕϕ⟩ in Giombi et al. The bulk expansion of ⟨σσ⟩ is new and allows to extract the leading large N expression for an unknown OPE coefficient Cσσσ3 (the computation outputs as a byproduct also a mixing coefficient, which matches an existing result by Derkachov et al.

[3].

Bulk Data Analysis

The paper computes bulk spectral functions associated with ⟨ϕϕ⟩ and ⟨σσ⟩ from their coefficients of bulk conformal block decompositions, verifying their correctness by explicitly plotting the sequence of partial bulk block sums. The analysis identifies a unique infinite tower of bulk primaries contributing to both the ϕ × ϕ and σ × σ OPE in (5. p. 4

Where the paper says it

However, let as point out that their correctness was verified by explicitly plotting the sequence of partial bulk block sums and checking that they converge to the exact correlators, as displayed in Figure 4 and Figure 5. 4), a unique infinite tower of bulk primaries contributing both to the ϕ × ϕ and σ × σ OPE defined in (4.

4). For the ⟨ϕϕ⟩ correlator, the OPE coefficient Cϕϕσ is extracted and successfully matched against a known result. p. 4
Where the paper says it

– Bulk-bulk 2−pt function ⟨ϕϕ⟩: • The OPE coefficient Cϕϕσ was extracted in (5. 11) and successfully matched against a known result.

Extraction of OPE Coefficients

The analysis leads to predictions for several unknown OPE coefficients, including Cσσσ3 (5.45). For the leading operator k=2, the prediction for the leading OPE coefficient is Cσσσ2[67] = √2 (5. p. 5

Where the paper says it

• As a byproduct of the analysis leading to its derivation, also a mixing coefficient, defined in (5. 25), linking the two primaries σ 3 and [σσ]1,0 (a double-twist operator defined in (5.

37). The anomalous dimension of σ squared is predicted by (5. p. 4
Where the paper says it

37) was found to coincide with existing calculations. • Anomalous dimension of σ 2 is determined in (5.

38).

Convergence and Outlook

The convergence of the boundary expansion is shown to be fast across different dimensions, uniform on any interval ρ ∈ (ε, 1] with ε > 0 that doesn’t include the bulk OPE point ρ = 0. The authors suggest future directions include running conformal perturbation theory in reverse and extending computations to the next large N order.

A Relation between bulk/boundary block coefficients and BCFT data

In this appendix, the relation between the coefficient of a bulk/boundary conformal block and BCFT data for the general case of a 2-pt function ⟨O1O2⟩ is derived, yielding A = 2−∆1−∆2+∆3 CO1O2O3 CO3O3 ⟨O3⟩ (A.7).

Boundary Expansion Bootstrap

The boundary spectral function for ⟨δσδσ⟩ is derived by bootstrap arguments, leading to Spec∂⟨δσδσ⟩ = 1/π (∆b − 1) tan π/2 ∆b. This result is pinned down by determining a single residue, which corresponds to the boundary operator of scaling dimension ∆b = 3—the displacement operator—whose BOE coefficient is linked via the Ward identity (C. p. 1

Where the paper says it

1 ⟨ϕϕ⟩ bulk expansion 25 5. 2 ⟨σσ⟩ bulk expansion 30 6 Outlook 34 A Relation between bulk/boundary block coefficients and BCFT data 35 B Bootstrap of the boundary spectral function for ⟨δσδσ⟩ 37 C Ward identity for BOE coefficient of displacement operator 39 D Summing bulk conformal block expansion for ϕ 40 – 1 –

6) to the 1-pt function ⟨σ⟩. p. 1
Where the paper says it

1 ⟨ϕϕ⟩ bulk expansion 25 5. 2 ⟨σσ⟩ bulk expansion 30 6 Outlook 34 A Relation between bulk/boundary block coefficients and BCFT data 35 B Bootstrap of the boundary spectral function for ⟨δσδσ⟩ 37 C Ward identity for BOE coefficient of displacement operator 39 D Summing bulk conformal block expansion for ϕ 40 – 1 –

Bulk Conformal Block Decomposition

The bulk conformal block expansion of the modified connected correlator is shown to be consistent with its spectral representation, and the coefficients C⟨ϕϕ⟩ k are explicitly known coefficients in (4.24). The convergence analysis for the bulk expansion shows oscillatory behavior in subsequent coefficients, which prevents the ordinary transition from being investigated by numerical bootstrap algorithms requiring positivity of expansion coefficients. p. 3

Where the paper says it

Note that all computations were done for the ordinary transition. The expansion coefficients calculated in this text are also available in the supplementary Wolfram Mathematica notebook 1 including some that did not fit into this text.

C Ward identity for BOE coefficient of displacement operator

The norm of the displacement operator 2-pt function CDbDb is found to be ∆σ⟨σ⟩√NσvolSd/2 = 8−d(d−3)(d−1)2π/3 −dΓ(d+1/2)/Γ(2d−2)cos πd/2 Γ(d+1)/Γ(49). This result precisely matches [24] and [2, (C. p. 0

Where the paper says it

[1] and reproduce the bulk expansion of ⟨ϕϕ⟩ in Giombi et al. The bulk expansion of ⟨σσ⟩ is new and allows to extract the leading large N expression for an unknown OPE coefficient Cσσσ3 (the computation outputs as a byproduct also a mixing coefficient, which matches an existing result by Derkachov et al.

9)].

Bootstrap of the boundary spectral function for ⟨δσδσ⟩

The set of zeroes of Spec∂⟨δσδσ⟩ is given by the set of poles of Spec∂⟨ϕ2ϕ2⟩, which fixes the set of zeroes we are trying to bootstrap. The final spectral function is determined to be Spec∂⟨δσδσ⟩ = 1/π (∆b − 1) tan π/2 ∆b. p. 1

Where the paper says it

1 ⟨ϕϕ⟩ bulk expansion 25 5. 2 ⟨σσ⟩ bulk expansion 30 6 Outlook 34 A Relation between bulk/boundary block coefficients and BCFT data 35 B Bootstrap of the boundary spectral function for ⟨δσδσ⟩ 37 C Ward identity for BOE coefficient of displacement operator 39 D Summing bulk conformal block expansion for ϕ 40 – 1 –

The boundary limit of the connected σ-propagator ⟨δσ(x)δσ(y)⟩ is dominated by the boundary operator of lowest scaling dimension in the BOE of δσ – the displacement operator. The resulting prediction for it is C(−1/2)σσσ[67] = −2∆(0)σqN FP σ⟨σ>(1/2)(d−1)(d−2) 16 = 4(d−2)Γ(d−1/2)/√πΓ(5-d/49). p. 7

Where the paper says it

1) as can be seen by the equation of motion for σ which just sets σ = √ λ N ϕ 2 . Here, the critical theory can be obtained by performing the limits m2 → 0, λ → ∞, and ξ → 1 4 d−1 d , resulting in the simplified action SHS[ϕ i , σ] = Z M d d+1x 1 2 ∂ϕi 2 + d − 1 8d Rϕiϕi + 1 2 √ N σϕiϕi .

Bulk OPE

The structure of the ϕ × ϕ and σ × σ OPEs is inferred from the bulk conformal block expansions (4.28). The leading operator contributing to the ϕ × ϕ OPE is σ, and its squared coefficient C(−1/2)ϕϕσ[67] squared = OpeN[Op[S,0,1]] verifies the hypothesis (5. p. 4

Where the paper says it

15) we define an infinite tower of operators, σbk, k ∈ Z≥0, appearing in the BOE of σ and in (3. 16) their squared BOE coefficients bσbσk 2 are provided for any d.

4).

Subleading N0 order

The subleading bulk conformal block expansion for the modified connected correlator is given by (4.7). The coefficients C, D are rather complicated functions expressed in terms of (derivatives of) 4F3[. p. 4

Where the paper says it

Correctness of the BOE coefficients is verified by explicit boundary conformal block expansions, see Figure 2 and Figure 3. In Section 4 bulk spectral functions associated with ⟨ϕϕ⟩ and ⟨σσ⟩ were computed and from those coefficients of bulk conformal block decompositions derived.

] hypergeometric functions with arguments dependent only on the space dimension d and scaling dimensions ∆k = 2k, k ≥ 2 of bulk primaries contributing to the σ × σ OPE. p. 4
Where the paper says it

However, let as point out that their correctness was verified by explicitly plotting the sequence of partial bulk block sums and checking that they converge to the exact correlators, as displayed in Figure 4 and Figure 5. 4), a unique infinite tower of bulk primaries contributing both to the ϕ × ϕ and σ × σ OPE defined in (4.

The convergence analysis shows that subsequent coefficients change sign as can be seen from the oscillatory behavior.

The subleading bulk conformal block expansion for the modified connected correlator is given by (4.27). The main outcome of the above analysis is that, like the ϕ × ϕ OPE, also the σ × σ OPE contains an infinite tower of operators of the schematic form (5. p. 4

Where the paper says it

However, let as point out that their correctness was verified by explicitly plotting the sequence of partial bulk block sums and checking that they converge to the exact correlators, as displayed in Figure 4 and Figure 5. 4), a unique infinite tower of bulk primaries contributing both to the ϕ × ϕ and σ × σ OPE defined in (4.

4).

k = 1: σ

The operator k=1: σ contributes just to the subleading term (5.32) (if d ≠ 2) as its coefficient in the leading term (5. p. 0

Where the paper says it

[1] and reproduce the bulk expansion of ⟨ϕϕ⟩ in Giombi et al. The bulk expansion of ⟨σσ⟩ is new and allows to extract the leading large N expression for an unknown OPE coefficient Cσσσ3 (the computation outputs as a byproduct also a mixing coefficient, which matches an existing result by Derkachov et al.

30) vanishes due to (5. p. 9
Where the paper says it

For later convenience, it is useful to have a basic picture of these functions across different dimensions. Their OPE bulk limit grows like a power ⟨δσδσ⟩ ∼ ρ→0 1 ρ 2 , while their boundary limit vanishes as ⟨δσδσ⟩ ∼ ρ→1 (1 − ρ) d+1.

31). Comparing its contribution via the bulk channel σ × σ OPE with the coefficient of its block yields 2−∆(0)σ−∆(0)σ+∆(0)σ(−1/2) FP⟨σσ⟩(0) FP⟨σ>(1/2)(d−1)(d−2)/16 = −N FP σ(d-1)(d-2)/4 (5. p. 8
Where the paper says it

10) Then the bulk OPE limit ρ → 0 yields the 2-pt function of the canonically normalized free masless scalar on R d+1 1 x⊥ d−1 2 1 y⊥ d−1 2 ⟨ϕ iϕ j ⟩ = ⟨ϕ iϕ j ⟩FHP −−−→ ρ→0 ⟨ϕ iϕ j ⟩FP = δ ij N FP ϕ h (x − y) 2 i d−1 2 , (2. 11) and we introduced the normalization N FP σ in the OPE limit since it will play a role when we will be computing CFT data from bulk conformal block expansions.

35).

k = 2: σ 2

The σ squared operator participates both in the leading (5.30) and subleading (5. p. 1

Where the paper says it

2 Leading ⟨σσ⟩ bulk conformal block expansion 22 4. 3 Subleading ⟨σσ⟩ bulk conformal block expansion 23 4.

32) equations, which we discuss in order. p. 4
Where the paper says it

Note that for OPE coefficients we offer slightly more than just a test as we fixed also their signs (in most of the literature OPE coefficients were computed from bulk 4−pt functions sensitive just to their squares, contrary to BCFT capable of fixing also signs). All checks regard the leading non-trivial large N order (emphasized as a superscript enclosed in parenthesis for each quantity of interest) and references used for crosschecking are included at relevant places where the links bellow point to.

At leading order, using the by now familiar strategy of matching the bulk OPE with the computed coefficient of the block, together with large N factorization (as all quantities are computed at leading order) and the conversion dictionaries (5. p. 4
Where the paper says it

Note that for OPE coefficients we offer slightly more than just a test as we fixed also their signs (in most of the literature OPE coefficients were computed from bulk 4−pt functions sensitive just to their squares, contrary to BCFT capable of fixing also signs). All checks regard the leading non-trivial large N order (emphasized as a superscript enclosed in parenthesis for each quantity of interest) and references used for crosschecking are included at relevant places where the links bellow point to.

8)/(5.2), gives 2−∆(0)σ−∆(0)σ+∆(0)σ squared [z]1 σσσ2 FP⟨σ 2σ 2>(0) FP⟨σ 2>(1/4)= (4.26).

k = 3: σ cubed, σ□σ

The operators k=3: σ cubed, σ□σ enter the leading part of the correlator (5.31), as well as the subleading one (5. p. 7

Where the paper says it

Our interest will be in the ordinary transition which can intuitively be understood—if one considers the model as a lattice of spins—as a transition connecting a phase where both bulk and boundary are ordered with a phase where they are both disordered. A different perspective on the boundary critical theory is provided by AdS/CFT techniques, from which it is well known that one can map the half-space R d+1 ≥0 onto a hyperboloid (the Euclidean Anti-de Sitter space) Hd+1 by a Weyl transformation.

32) – both the terms with and without derivatives of conformal blocks. p. 5
Where the paper says it

Bulk/boundary objects. • Bulk/boundary conformal blocks, partial waves, spectral functions are denoted by the superscript B/∂: G B/∂ , CPWB/∂ , SpecB/∂ .

The subleading equation containing undifferentiated blocks involves CFT data that needs to be evaluated at subleading order and is currently unknown. p. 1
Where the paper says it

3 Subleading ⟨σσ⟩ bulk conformal block expansion 23 4. 4 Bulk OPE 25 5 Extraction of BCFT data 25 5.

The resulting system of two equations for two unknowns yields a prediction for the yet unknown OPE coefficient C(−1/2)σσσ3[67] = q 3!N FP σ 4⟨σ>(d−1)(d−2)/16 = −4q 6π(d−2)r(d−3)cos πd/2Γ(5-d/49).

The mixing coefficient b(−1/2)1 is expressed as b(−1/2)1 = 1⟨σ>(3/2)[A(1)[σσ]1,0 q⟨[σσ]1,0[σσ]1,0>4N FP σ C(0)σσ[67]−⟨[σσ]1,0>(1) = −2(d−9)(d−2)(d+1)d/2 − 32d+63.

k = 4: σ 4, □σ 3

The next two members of this family, the k=4 operators, impose an infinite number of constraints on BCFT data.

In this appendix we derive the relation between the coefficient of a bulk/boundary conformal block and BCFT data for the general case of a 2-pt function ⟨O1O2⟩, yielding A = 2−∆1−∆2+∆3 CO1O2O3 CO3O3 ⟨O3⟩ (A.7).

Boundary expansion

The boundary limit of the connected σ-propagator ⟨δσ(x)δσ(y)⟩FHP is dominated by the boundary operator of lowest scaling dimension in the BOE of δσ – the displacement operator. The resulting prediction for it is Nσ2/2(d+1) = BσDb 2CDbDb. p. 7

Where the paper says it

1) as can be seen by the equation of motion for σ which just sets σ = √ λ N ϕ 2 . Here, the critical theory can be obtained by performing the limits m2 → 0, λ → ∞, and ξ → 1 4 d−1 d , resulting in the simplified action SHS[ϕ i , σ] = Z M d d+1x 1 2 ∂ϕi 2 + d − 1 8d Rϕiϕi + 1 2 √ N σϕiϕi .

Bulk expansion

The goal is to verify the bulk expansion of the modified correlator ρ−1ρ∆ϕ⟨ϕϕ⟩ given in (5.5) with coefficients (4. p. 1

Where the paper says it

1 ⟨ϕϕ⟩ bulk expansion 25 5. 2 ⟨σσ⟩ bulk expansion 30 6 Outlook 34 A Relation between bulk/boundary block coefficients and BCFT data 35 B Bootstrap of the boundary spectral function for ⟨δσδσ⟩ 37 C Ward identity for BOE coefficient of displacement operator 39 D Summing bulk conformal block expansion for ϕ 40 – 1 –

24). The proof shows that the double sum on the right hand side is equal to the coefficient in front of the power of ρ on the left hand side (D. p. 7
Where the paper says it

Starting with the special transition in the UV, one can add a relevant boundary interaction term cϕˆ2 (with positive c) which drives the RG flow to the ordinary transition while adding the same operator with a negative c leads to an RG flow to the extraordinary transition. On the other hand, if one starts in the ordinary transition, a deformation by the operator h · ϕˆ (where h is a vector of constants) leads again to the extraordinary transition.

2).

The task is to output a formula for the coefficient A in terms of BCFT data, yielding A = 2−∆1−∆2+∆3 CO1O2O3 CO3O3 ⟨O3⟩ (A.7).

The boundary limit of the conformal block behaves like (1 − ρ)∆b, which in coordinates means RHS ∼ B(1 − ρ)∆b ∼ y⊥→0 B 4x⊥y⊥ x 2⊥ + (x−y) 2!∆b.

The key feature of these plots is that subsequent coefficients of the expansion change sign as can be seen from the oscillatory behavior. This prevents the ordinary transition from being investigated by (rigorous, SDPB like) numerical bootstrap algorithms as they require positivity of expansion coefficients. p. 3

Where the paper says it

Note that all computations were done for the ordinary transition. The expansion coefficients calculated in this text are also available in the supplementary Wolfram Mathematica notebook 1 including some that did not fit into this text.

The boundary spectral representation is Spec∂⟨δσδσ⟩ = -1/(8π)2 1/Spec∂⟨ϕ2ϕ2⟩ (B.1). The inverse relation implies that the set of zeroes of Spec∂⟨δσδσ⟩ is given by the set of poles of Spec∂⟨ϕ2ϕ2⟩, which is the boundary spectrum of the bubble function ⟨ϕ2ϕ2⟩. p. 8

Where the paper says it

4) around the saddle point and only keeping the leading terms in large N gives the connected part of effective propagator for the σ field ⟨δσδσ⟩ = −4B −1 (x, y) (2. 7) where B(x, y) = ⟨ϕ 2 (x)ϕ 2 (y)⟩ N = 2 N ⟨ϕ iϕ j ⟩⟨ϕiϕj ⟩ is the bubble function and by B−1 we mean its inverse with respect to convolutions on Hd+1.

The final spectral function is determined to be Spec∂⟨δσδσ⟩ = 1/π (∆b − 1) tan π/2 ∆b (B.8). This finishes the proof of (B.2) we set out to achieve in this appendix.

D Summing bulk conformal block expansion for ϕ

The double sum on the right hand side is equal to the coefficient in front of the power of ρ on the left hand side (D.2). This finishes the proof of (5.5).

The prediction for it is C(−1/2)σσσ[67] = −2∆(0)σqN FP σ⟨σ>(1/2)(d−1)(d−2)/16 = 4(d−2)Γ(d−1/2)/√πΓ(5-d/49).

The prediction for the leading OPE coefficient is Cσσσ2[67] = √2 (5.37).

The prediction for it is C(−1/2)σσσ[67] = −2∆(0)σqN FP σ⟨σ>(1/2)(d−1)(d−2)/16 = 4(d−2)Γ(d-1/2)/√πΓ(5-d/49).

The boundary spectral representation is Spec∂⟨δσδσ⟩ = -1/(8π)2 1/Spec∂⟨ϕ2ϕ2⟩ (B.1). The inverse relation implies that the set of zeroes of Spec∂⟨δσδσ⟩ is given by the set of poles of Spec∂⟨ϕ2ϕ2⟩, which fixes the set of zeroes we are trying to bootstrap. p. 1

Where the paper says it

1 ⟨ϕϕ⟩ bulk expansion 25 5. 2 ⟨σσ⟩ bulk expansion 30 6 Outlook 34 A Relation between bulk/boundary block coefficients and BCFT data 35 B Bootstrap of the boundary spectral function for ⟨δσδσ⟩ 37 C Ward identity for BOE coefficient of displacement operator 39 D Summing bulk conformal block expansion for ϕ 40 – 1 –

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. Improved ability to predict unknown OPE coefficients using constraints from known data. The system can solve a system of two equations for two unknowns, such as determining C sigma sigma3[67] by matching its prediction against known results and solving for the mixing coefficient b1 using equation (5.46).

  2. Enhanced capability in analyzing convergence properties of bulk conformal block expansions. The AI can identify oscillatory behavior in coefficients, which signals that numerical bootstrap algorithms cannot be used, and determine when convergence is uniform on specific intervals of the cross ratio rho.

  3. Improved capacity to derive anomalous dimensions for composite operators from bulk spectral functions. The system can compute the anomalous dimension of sigma squared by taking the ratio of coefficients multiplying differentiated blocks in equation (5.31)/(5.33), matching known results and verifying consistency with those computed in other literature like [67].

  4. New ability to extract boundary spectral functions for correlators using bootstrap arguments. The system can fix the set of poles and zeroes of the boundary spectral function by applying bulk shadow symmetry and assuming simple poles, leading to a closed-form expression like SpecB ⟨δσδσ⟩ = A X k∈Z 1 ∆b − (2k + 1).

  5. Determination of scaling dimensions for an infinite tower of operators. The system can identify the family of bulk primaries contributing to the OPE expansion by matching their subleading anomalous dimensions, such as identifying sigma k=1, sigma squared o k=2, etc., by comparing them against known CFT data in [67].

Abstract

The (N) is analyzed in the large N expansion in a generic bulk dimension 2<d+1<4. Focus is on boundary conditions corresponding to the ordinary transition, however techniques used can be generalized to special or extraordinary transitions. We study the system of bulk 2-pt functions ϕϕ, ϕ squared ϕ squared and σσ, where σ is the Hubbard--Stratonovich field. They are expanded in bulk/boundary conformal partial waves. The coefficient functions of this expansion -- the spectral functions -- encode data and determine bulk/boundary conformal block expansions. We prove conjectures about boundary spectral functions in Dujava et al. [arXiv:2503.16345] and reproduce the bulk expansion of ϕϕ in Giombi et al. [arXiv:2007.04955]. The bulk expansion of σσ is new and allows to extract the leading large N expression for an unknown OPE coefficient C σσσ cubed (the computation outputs as a byproduct also a mixing coefficient, which matches an existing result by Derkachov et al. [arXiv:hep-th/9705020], providing further support for the main result). Besides these derived data infinitely many constraints are produced, though due to operator mixing growing in complexity as degeneracy increases, they cannot be disentangled without further input.

Sources

Related papers