Truncated 5-Point Bootstrap

• Truncated 5-Point Bootstrap reduces infinite-dimensional crossing equations to a finite linear system by truncating the exchanged operator spectrum and enforcing functional sum rules.
• It employs both analytic functional truncation and derivative-based methods to approximate OPE coefficients, achieving low relative errors even at low truncation orders.
• The technique leverages Polyakov blocks and orthogonality properties to extract CFT data from five-point correlators, bridging exact solutions with computational approximations.

Searching arXiv for the cited and closely related papers on five-point and truncated bootstrap methods. Truncated 5-Point Bootstrap denotes a class of approximate conformal-bootstrap schemes in which a five-point crossing problem is reduced to a finite-dimensional system by retaining only finitely many exchanged operators and finitely many constraints. In the recent five-point Polyakov-bootstrap formulation for one-dimensional conformal field theories, the term refers specifically to taking a known or assumed exchanged spectrum in a chosen OPE channel, truncating the set of operator pairs $(\Delta_1,\Delta_2)$, and solving a finite set of functional sum rules for approximate five-point OPE coefficients (Antunes et al., 7 Aug 2025). More broadly, related work in higher dimensions implements the same truncation principle by evaluating five-point crossing equations with a finite OPE cutoff and minimizing residual crossing violations near a symmetric kinematic configuration (Poland et al., 2023). The approach is motivated by the fact that a single higher-point correlator constrains the OPE data of infinitely many four-point correlators, so a controlled finite truncation can provide nontrivial access to CFT data that are difficult to extract from scalar four-point functions alone (Antunes et al., 7 Aug 2025).

1. Higher-point motivation and the role of truncation

In conformal field theory, a four-point function determines spectra and OPE coefficients in a given channel, but higher-point correlators encode substantially more data. For five-point functions, every sequential fusion pattern defines an effective four-point problem involving composite operators, so one correlator constrains infinitely many four-point correlators at once (Antunes et al., 7 Aug 2025). In one-dimensional CFT, the paper "Five points for the Polyakov Bootstrap" develops this idea for identical scalars and emphasizes that a single five-point function of a light scalar operator $\phi$ already encodes the OPE data of infinitely many four-point correlators involving heavier operators (Antunes et al., 7 Aug 2025).

The practical obstacle is that exact five-point crossing equations are infinite-dimensional. A truncated 5-point bootstrap replaces the full problem with two finite choices: a finite exchanged spectrum and a finite family of equations. In the one-dimensional Polyakov-bootstrap setting, the exchanged dimensions are taken as known input and only finitely many operator pairs $(\Delta_1,\Delta_2)$ in the $12$-$3$-$45$ channel are retained, after which one solves finitely many functional sum rules for the OPE coefficients $a_{\Delta_1,\Delta_2}$ (Antunes et al., 7 Aug 2025). In the three-dimensional scalar five-point bootstrap, truncation is instead imposed through a dimension cutoff $\Delta \leqslant \Delta_{\rm cutoff}$ and a finite set of derivative constraints around a symmetric point, leading to an overdetermined approximate crossing system (Poland et al., 2023).

This suggests a useful distinction between two operational realizations of the same idea. One is functional truncation, where one chooses analytic functionals or derivative functionals and solves a finite linear system. The other is residual minimization, where one chooses a finite operator basis and minimizes a cost function built from crossing violations. Both are truncated five-point bootstrap methods in the broad sense, but they differ sharply in how crossing information is compressed (Antunes et al., 7 Aug 2025, Poland et al., 2023).

2. One-dimensional five-point kinematics and channel decomposition

For identical scalars in one dimension, the five-point correlator is written as

$$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$

with two independent cross-ratios

$$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$

In the $\phi$0-$\phi$1-$\phi$2 OPE channel, the correlator admits the block expansion

$\phi$3

where the exchanged data are naturally indexed by operator pairs $\phi$4 rather than a single intermediate dimension (Antunes et al., 7 Aug 2025).

Crossing symmetry is generated by cyclic permutations and takes the explicit form

$\phi$5

This already displays the main structural difference from four-point bootstrap: the unknown OPE data form a two-index array $\phi$6, and the crossing transformation acts nontrivially on two variables (Antunes et al., 7 Aug 2025).

The Mellin representation clarifies why truncation can be organized efficiently. In the $\phi$7-$\phi$8-$\phi$9 channel, the OPE limits $(\Delta_1,\Delta_2)$0 and $(\Delta_1,\Delta_2)$1 are controlled by poles in $(\Delta_1,\Delta_2)$2 and $(\Delta_1,\Delta_2)$3. The Mellin amplitude must satisfy crossing symmetry, Regge boundedness, and OPE compatibility, and these requirements provide the analytic backbone for the Polyakov-block construction (Antunes et al., 7 Aug 2025). A truncated scheme then amounts to retaining only finitely many physical pole families or finitely many operator pairs in the corresponding position-space decomposition.

3. Polyakov blocks, spurious families, and functional sum rules

The one-dimensional Polyakov-bootstrap formulation generalizes the four-point Polyakov block to five points by constructing a crossing-symmetric combination of Witten diagrams. The five-point Polyakov block $(\Delta_1,\Delta_2)$4 includes a double-exchange diagram exchanging $(\Delta_1,\Delta_2)$5 in the $(\Delta_1,\Delta_2)$6 channel and $(\Delta_1,\Delta_2)$7 in the $(\Delta_1,\Delta_2)$8 channel, a set of single-exchange diagrams, and a unique crossing-symmetric contact diagram consistent with the Regge bound (Antunes et al., 7 Aug 2025). In Mellin space, the Polyakov block contains 30 terms, one for each pair $(\Delta_1,\Delta_2)$9 with all indices distinct, together with the term obtained by $12$0 (Antunes et al., 7 Aug 2025).

Its importance for truncation lies in its decomposition in the physical $12$1-$12$2-$12$3 channel. In position space,

$12$4

equals the desired physical blocks plus a tower of spurious blocks. These spurious contributions fall into four families, associated with coefficients denoted

$12$5

Demanding cancellation of the spurious blocks yields the corresponding functional sum rules (Antunes et al., 7 Aug 2025).

These sum rules are

$12$6

$12$7

$12$8

and

$12$9

The first two are genuinely multipoint constraints because they simultaneously relate OPE data across an infinite set of effective four-point functions. The last two behave more like analytic functionals acting on correlators $3$0 at fixed $3$1 (Antunes et al., 7 Aug 2025).

The families are organized by double-twist and triple-twist dimensions,

$3$2

and by the mixed family

$3$3

The orthogonality properties of the $3$4 functionals are central to the truncated scheme. The single-variable functionals are dual to specific twist families, while the two-variable functionals approximately satisfy Kronecker-delta behavior on matched double- or triple-twist labels and vanish on mismatched families, up to special lowest-twist shifts (Antunes et al., 7 Aug 2025). This makes the truncated linear systems unusually sparse in functional space even when they are dense in operator space.

The Mellin Polyakov block contains two free regular terms multiplied by $3$5 and $3$6. These are fixed by eliminating two chosen functionals, for example

$3$7

This choice functions as a gauge fixing of the Polyakov-block representation and is the five-point analogue of choosing a contact term to remove a particular spurious derivative block in the four-point Polyakov bootstrap (Antunes et al., 7 Aug 2025).

4. The truncated 5-point bootstrap as a finite linear system

In the one-dimensional Polyakov-bootstrap implementation, “truncated 5-point bootstrap” has a precise operational meaning. One starts from a specific known five-point correlator, assumes the exchanged spectrum is given, retains only finitely many pairs $3$8 in the $3$9-$45$0-$45$1 channel, and enforces only finitely many functional equations. The result is a finite linear system for finitely many OPE coefficients $45$2 (Antunes et al., 7 Aug 2025).

The test case is the “all-to-all” correlator, obtained from a generalized free field $45$3 and the external operator $45$4. After stripping the kinematic prefactor, the reduced correlator is

$45$5

Its $45$6-$45$7-$45$8 OPE is controlled by a single tower

$45$9

and the five-point OPE coefficients form a matrix $a_{\Delta_1,\Delta_2}$0 (Antunes et al., 7 Aug 2025).

For $a_{\Delta_1,\Delta_2}$1, the first exact entries are

$a_{\Delta_1,\Delta_2}$2

The truncation chosen in the Polyakov-functional test retains the seven operator pairs

$a_{\Delta_1,\Delta_2}$3

equivalently

$a_{\Delta_1,\Delta_2}$4

Setting $a_{\Delta_1,\Delta_2}$5, one solves for six unknowns using six functional equations: three $a_{\Delta_1,\Delta_2}$6 rules, one $a_{\Delta_1,\Delta_2}$7 rule, and two $a_{\Delta_1,\Delta_2}$8 rules (Antunes et al., 7 Aug 2025).

The resulting numerical solution is

$a_{\Delta_1,\Delta_2}$9

Compared with the exact values

$\Delta \leqslant \Delta_{\rm cutoff}$0

the agreement is good, with relative errors typically at the few-percent level despite the very low truncation order (Antunes et al., 7 Aug 2025). This is the cleanest explicit realization of truncated 5-point bootstrap presently available in the cited literature.

5. Comparison with derivative-based five-point truncations

A central claim of the Polyakov-bootstrap work is that the new functionals outperform standard derivative functionals in low-order truncations. The derivative approach starts from the five-point crossing equation

$\Delta \leqslant \Delta_{\rm cutoff}$1

and defines functionals

$\Delta \leqslant \Delta_{\rm cutoff}$2

at the crossing-symmetric point

$\Delta \leqslant \Delta_{\rm cutoff}$3

Choosing finitely many derivative pairs $\Delta \leqslant \Delta_{\rm cutoff}$4 again yields a finite linear system for the truncated OPE data (Antunes et al., 7 Aug 2025).

The comparison was performed by using the same truncated operator set and selecting six random derivative functionals with $\Delta \leqslant \Delta_{\rm cutoff}$5, repeated 10 times. The quality metric was the relative error

$\Delta \leqslant \Delta_{\rm cutoff}$6

For the lightest coefficients, such as $\Delta \leqslant \Delta_{\rm cutoff}$7, derivatives and Polyakov functionals perform comparably on average. For heavier coefficients, such as $\Delta \leqslant \Delta_{\rm cutoff}$8, $\Delta \leqslant \Delta_{\rm cutoff}$9, and $$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$0, the Polyakov-functional solution consistently gives smaller errors, and even the best derivative system performs worse (Antunes et al., 7 Aug 2025).

The stated explanation is structural rather than purely empirical. Polyakov functionals encode Regge behavior and OPE structure globally, are built so that spurious double- and triple-twist contributions enter with controlled coefficients, and enjoy orthogonality properties that make them effectively dual to particular twist sectors. Derivative functionals, by contrast, are local probes of the crossing equation in cross-ratio space and only recover comparable global information when many derivatives or multiple points are used (Antunes et al., 7 Aug 2025). A plausible implication is that truncation error at five points is governed less by local Taylor accuracy than by the extent to which the chosen functionals resolve the correct twist-family organization.

A separate derivative-based five-point truncation was developed in general dimension for the critical 3d Ising model. There, five-point blocks for arbitrary exchanged spins were constructed in generalized radial coordinates by solving two quadratic Casimir differential equations, and crossing was approximately enforced by taking derivatives up to third order near a symmetric point (Poland et al., 2023). The unknowns included $$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$1 and several OPE coefficients involving two spinning operators and $$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$2, with best-fit values extracted by minimizing a random-weighted cost function over an overdetermined derivative system (Poland et al., 2023). That work demonstrates that derivative truncations can be numerically useful, but it also documents sizable instability for tensor structures whose contributions are comparable to the neglected tail (Poland et al., 2023). The later Polyakov-functional results in one dimension sharpen the methodological comparison by providing a setting where both truncation schemes can be tested against exact data (Antunes et al., 7 Aug 2025).

6. Alternative formulations: residual minimization and stochastic search

Although the one-dimensional Polyakov-bootstrap study formulates truncation as a finite linear problem, other truncated-bootstrap literature uses a different architecture based on residual minimization. "A Monte Carlo approach to the conformal bootstrap" considers a truncated four-point system, evaluates crossing on a finite set of points near the crossing-symmetric region, and defines an action

$$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$3

with $$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$4 and $$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$5 (Laio et al., 2022). At fixed dimensions, the OPE coefficients are integrated out by weighted least squares, and the remaining optimization over scaling dimensions is performed by a Metropolis search (Laio et al., 2022).

That paper does not solve a five-point problem, but it explicitly presents its method as conceptually general and sketches how the same logic could be carried over to truncated five-point crossing: choose a finite operator basis, sample a compact cross-ratio region, define residuals for a finite set of independent crossing equations, and minimize a weighted action over dimensions and OPE data (Laio et al., 2022). This makes it relevant to the encyclopedia topic because it clarifies that “truncated bootstrap” need not mean a purely linearized functional method. In practice, truncated 5-point bootstrap can refer either to a finite system of sum rules or to a finite-dimensional variational problem.

The three-dimensional five-point bootstrap of $$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$6 is closer to the residual-minimization paradigm than to the Polyakov-functional paradigm. It truncates the $$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$7 OPE to

$$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$8

with $$\langle \phi(x_1)\dots \phi(x_5)\rangle = \mathcal{L}(x_i)\,\mathcal{G}(\chi_1,\chi_2),$$9, computes five-point blocks in generalized radial coordinates $$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$0, expands crossing near the symmetric point

$$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$1

and minimizes a cost function

$$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$2

built from derivative constraints $$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$3 and random weights $$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$4 (Poland et al., 2023). The extracted quantities include

$$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$5

as well as $$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$6 (Poland et al., 2023). This demonstrates that truncated five-point bootstrap is not confined to one-dimensional scalar problems; it is a broader numerical strategy for extracting CFT data from finite higher-point crossing information.

7. Scope, limitations, and conceptual significance

The main conceptual significance of truncated 5-point bootstrap is that it operationalizes the use of higher-point crossing as a practical data-extraction tool. In one dimension, the Polyakov-functional construction shows that the method can be organized around analytically controlled spurious-family cancellation, with functionals that directly probe double- and triple-twist sectors across infinitely many effective four-point correlators (Antunes et al., 7 Aug 2025). In three dimensions, the derivative-based five-point bootstrap shows that the same truncation philosophy can estimate OPE coefficients involving two spinning operators and a scalar, such as $$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$7, $$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$8, and $$\chi_1= \frac{x_{12} x_{34}}{x_{13}x_{24}}, \qquad \chi_2= \frac{x_{23} x_{45}}{x_{24}x_{35}},\qquad x_{ij}=|x_i-x_j|\, .$$9, from scalar five-point data (Poland et al., 2023).

Several limitations are explicit in the literature. In one dimension, five-point correlators are not reflection-positive, so standard semidefinite programming is not directly available, motivating truncated schemes instead (Antunes et al., 7 Aug 2025). In higher dimensions, the lack of a manifestly positive expansion for five-point blocks prevents an SDP-style treatment and forces reliance on approximate cost functions or local minimization strategies (Poland et al., 2023). Systematic errors from omitted operators remain difficult to quantify sharply; derivative-based methods can become unstable when neglected higher-spin or higher-dimension contributions are comparable to the terms being extracted (Poland et al., 2023). Residual-minimization methods more generally sacrifice rigor for exploratory reach, since a local minimum of the action does not prove the existence of a CFT and the absence of a minimum does not exclude one (Laio et al., 2022).

At the same time, the recent one-dimensional Polyakov-bootstrap results suggest that the quality of a truncation is strongly controlled by the choice of functionals rather than only by the truncation size. Polyakov functionals appear to provide more stable and better conditioned linear systems, especially for heavier exchanged operators (Antunes et al., 7 Aug 2025). This suggests that future progress may come less from naively enlarging derivative bases and more from building analytic functionals adapted to higher-point Regge and OPE structure.

A broader antecedent for such truncations exists in supersymmetric settings where symmetry collapses parts of the bootstrap to polynomial constraints. In defect $\phi$00 SYM, special protected sectors reduce crossing to finite algebraic equations in BPS data, illustrating a distinct but related notion of truncated bootstrap based on closure of a subsector rather than numerical OPE cutoff (Liendo et al., 2016). This suggests that higher-point truncations may become especially powerful in theories where Ward identities or cohomological reductions compress the crossing problem before numerical approximation is imposed.

Taken together, these developments indicate that truncated 5-point bootstrap is emerging as a methodological bridge between exact higher-point crossing and computationally feasible approximations. In its most concrete current form, it is a finite-system reconstruction of five-point OPE data from analytically designed functional constraints (Antunes et al., 7 Aug 2025). In a broader sense, it is any controlled approximation scheme that uses a finite operator basis and finite five-point crossing data to solve for otherwise inaccessible CFT observables (Poland et al., 2023, Laio et al., 2022).