Sem II Week 1 — Witten Setup: Large $N$, Single-Trace Operators, and the Emergence of Type III

Semester II opens with Witten’s “Gravity and the Crossed Product” (arXiv:2112.12828). Its order of limits—and its power counting—matter. In the strict large-$N$ theory the non-central single-trace algebra ${\mathcal{A}}_{R,0}$ is type III${}_1$. If one also keeps the rescaled energy fluctuation $U=(H_R-⟨H_R⟩)/N$, then $U$ is central at $N=\infty$, so the extended algebra is not a factor. Beyond that limit, Witten replaces the generator $U$ by $U+\hat{h}/({\beta }_{H}N)$, whose commutator with fixed observables is already visible at order $1/N$. Separately, the usual adjoint large-$N$ coefficient expansion and holographic coupling use $1/N^2\sim G_N$; that fact does not turn the displayed $\hat{h}/({\beta }_{H}N)$ into an $O(1/N^2)$ term. The resulting perturbative algebra is the modular crossed product, hence a type II${}_{\infty }$ factor. The Rindler model used later checks the operator-algebraic continuous core; it does not derive the gravitational large-$N$ identification.

Route through this master dossier. Classroom core (two-hour seminar): §§1 and 4–5, including the four-regime table, plus the KMS part of the BTZ check in §6. Full derivation or self-study: §§2–3 and the complete §6 calculation, followed by Problems 1–6. Research extension or office hour: §7, the starred problems, and a line-by-line comparison with Witten §§2–3. This routing omits nothing from the dossier; it identifies what belongs on the board when time is finite.

0. Reading

Primary:

• Witten, “Gravity and the crossed product,” JHEP 10 (2022) 008, arXiv:2112.12828, §§1–2 (the setup and large-$N$ algebra).
• Liu, “Lectures on entanglement, von Neumann algebras, and emergence of spacetime,” arXiv:2510.07017, §§IV.1–IV.4 for type III and modular background, and especially §§VI.1–VI.3 and VI.6 for the large-$N$ AdS/CFT formulation and conserved-charge corrections.

Secondary:

• Aharony, Gubser, Maldacena, Ooguri, Oz, “Large $N$ field theories, string theory and gravity,” Phys. Rep. 323 (2000) 183 — the standard holographic background.
• Witten, “Notes on some entanglement properties of QFT,” arXiv:1803.04993, §§1–3 — modular theory in QFT (Sem I primary).

Optional research reading:

• Leutheusser & Liu, “Causal Connectability Between Quantum Systems and the Black Hole Interior in Large-N Gauge Theories,” arXiv:2110.05497 — emergence of type III at large $N$, with explicit examples.
• Faulkner, Li, Wang, “A modular toolkit for bulk reconstruction,” JHEP 04 (2019) 119, arXiv:1806.10560 — modular flow in holographic settings.

1. Why this paper

Witten 2022 is the structural anchor of the entire algebraic-gravity program. It does two things that the rest of Semester II builds on:

1. Uses the type-III${}_1$ algebra isolated by Leutheusser and Liu in the strict large-$N$ limit. The relevant algebra is generated by bounded functions of the non-central, subtracted single-trace operators. Witten reviews two reasons for its type: the bulk effective-field-theory identification and the Leutheusser–Liu analysis. Large-$N$ Gaussianity alone does not prove type III${}_1$.

2. Shows how the first noncentral correction produces the crossed product. At strict $N=\infty$, the rescaled energy fluctuation is a central collective coordinate. In Witten’s explicit order-by-order presentation it is corrected to $U+\hat{h}/({\beta }_{H}N)$, so the commutator generating boundary time translations is $O(1/N)$. This is the algebraic bookkeeping relevant to the type change. The separate $1/N^2$ expansion of gauge-invariant coefficients, and the familiar holographic scaling $G_N\sim 1/N^2$, organize other parts of the semiclassical expansion; they do not change the order of the displayed generator correction. Once the noncentral action is retained, the generator and ${\mathcal{A}}_{R,0}$ combine into ${\mathcal{A}}_{R,0}{⋊}_{\sigma }\mathbb{R}$, a type II${}_{\infty }$ factor. Sections 3.4–3.5 then construct its trace, densities, and entropy.

The combination is the punchline, at the paper’s stated perturbative scope: the entropy defined on the emergent type II${}_{\infty }$ algebra agrees with black-hole generalized entropy up to a state-independent additive constant.

Why study this paper. It gives a particularly clear bridge between Sem I’s operator-algebraic machinery and semiclassical black-hole entropy. The valuable point is not a slogan that “gravity gauges modular flow,” but the mechanism: a central energy variable at leading order acquires the modular generator $\hat{h}/({\beta }_{H}N)$ beyond the limit, and the resulting algebra is recognized as a crossed product. The $1/N$ in this generator and the $G_N\sim 1/N^2$ scaling of gravitational corrections must both remain visible.

Course strategy. We follow the paper’s structural arc and mark the status of each step. The continuous-core theorem is imported operator algebra; the identification of the large-$N$ algebra and collective energy is holographic input; the free-field Rindler model is a check of the first item only. This separation will be used throughout Semester II.

2. Recap: what Block D provides

Block D (Sem I Weeks 13–15) gave us the algebraic toolkit. Recall the structure:

Modular crossed product (Week 13). Given a vN algebra $\mathcal{M}$ with faithful normal state $\omega$ and modular flow ${\sigma }_t^{\omega }={\Delta }_{\omega }^{-it}\cdot {\Delta }_{\omega }^{it}$ (our upper-strip $\beta =+1$ convention), the modular crossed product is

$$\hat{\mathcal{M}}:=\mathcal{M}{⋊}_{{\sigma }^{\omega }}\mathbb{R}$$

acting on $\mathcal{H}\otimes L^2(\mathbb{R})$, generated by $\pi (a)={\int }^{\oplus }{\sigma }_{-s}^{\omega }(a)ds$ and $\lambda (t)=1\otimes U_t$ with $\lambda (t)\pi (a)\lambda (t{)}^{\ast }=\pi ({\sigma }_t^{\omega }(a))$.

Takesaki’s structure theorem (Week 13). If $\mathcal{M}$ is type III${}_1$, then $\hat{\mathcal{M}}$ is type II${}_{\infty }$, with a faithful normal semifinite trace $\hat{\tau }$, unique up to rescaling, satisfying $\hat{\tau }\circ {\theta }_r=e^{-r}\hat{\tau }$ under the dual action.

Dressed entropy (Week 14). When ${\hat{\rho }}_d\log {\hat{\rho }}_d$ is $\hat{\tau }$-integrable (or a common regularization is stated), $S_{\mathrm{vN}}(\hat{\rho })=-\hat{\tau }({\hat{\rho }}_d\log {\hat{\rho }}_d)$ is defined; rescaling the trace shifts a finite value by an additive constant. Faithfulness alone does not exclude divergent entropies. For two faithful normalized densities in the same semifinite core, relative to the same trace, and whenever the displayed terms are well defined, the exact trace identity is

$$S_{\mathrm{vN}}(D_{\omega })-S_{\mathrm{vN}}(D_{\varphi })=-S_{\mathrm{core}}(D_{\omega }\Vert D_{\varphi })+\hat{\tau }\left[(D_{\omega }-D_{\varphi })K_{\varphi }\right],K_{\varphi }=-\log D_{\varphi }.$$

Replacing the core relative entropy by Araki relative entropy on the original type-III algebra requires a specified lift and a proof that it preserves relative entropy; Witten and CPW supply such statements only in their stated constructions.

Free-field Rindler workhorse (Week 14 §5; Week 15 §6). For the free 2D massless scalar on $\mathcal{A}(W_R)$, Bisognano–Wichmann gives the explicit modular flow. In the regular crossed-product representation $Q=q$ is the multiplication coordinate, $P=-i{\partial }_q$, and $p$ is the Fourier spectral variable of $P$. Only in the separately checked inner-action Fourier model does the auxiliary trace take the form $e^{-p}dp$. Witten’s multiplication variable satisfies $X_{\mathrm{W}}=-p$ in this dictionary, so $e^{X_{\mathrm{W}}}d{X}_{\mathrm{W}}$ matches $e^{-p}dp$ up to orientation. If one defines $E=p/(2\pi )$, the same comparison weight is proportional to $e^{-2\pi E}dE$; it is never a weight $e^{-2\pi q}dq$ on the regular coordinate.

The missing ingredient (this block). To apply this machinery to gravity, we must identify the starting factor, its modular flow, and the collective variable retained beyond leading order. Witten’s construction uses ${\mathcal{A}}_{R,0}$, the TFD modular flow generated on the GNS Hilbert space by $\hat{H}=H_R-H_L$, and the rescaled one-sided energy fluctuation that is central only at strict $N=\infty$.

3. Large-$N$ field theory

A holographic CFT (e.g., $\mathcal{N}=4$ super-Yang–Mills with gauge group $\mathrm{U}(N)$ or $\mathrm{SU}(N)$) has a large-$N$ limit in which the gauge group becomes infinite-dimensional. This limit is governed by the ‘t Hooft expansion.

3.1 ‘T Hooft expansion

Consider a $\mathrm{U}(N)$ gauge theory with coupling $g$. The ‘t Hooft coupling is $\lambda :=g^{2}N$, held fixed as $N\to \infty$. In this limit:

• Planar diagrams dominate. Feynman diagrams that can be drawn on a sphere (genus 0) are leading order; higher-genus diagrams are suppressed by $1/N^2$.
• The expansion is in $1/N$, with $\lambda$ as a separate (independent) parameter that controls the strength of the planar interactions.

For a connected ribbon graph of genus $g$ with $b$ single-trace insertions, the standard unnormalized counting is $N^{2-2g-b}$. The normalization of the operators can move powers of $N$ between the definition and the correlator, so we will state explicitly which convention is being used below.

3.2 Single-trace operators

Definition 3.1. A subtracted single-trace operator in the normalization used by Witten is

$$\mathcal{W}(x):=\mathrm{Tr}({\Phi }_{i_1}(x)\cdots {\Phi }_{i_k}(x))-⟨\mathrm{Tr}({\Phi }_{i_1}(x)\cdots {\Phi }_{i_k}(x))⟩.$$

With the action normalized as $N\mathrm{Tr}L$, its connected two-point function is $O(1)$, while connected $k$-point functions scale as $N^{2-k}$. If instead one divides the trace by $N$, every formula below acquires the corresponding extra power of $N^{-k}$.

Multi-trace operators are products of single-traces: ${\mathcal{O}}_1(x){\mathcal{O}}_2(x)$, etc.

3.3 Large-$N$ factorization

Large-$N$ factorization. [Stated only — refs: ‘t Hooft 1974; Witten 2022 §2.] For the subtracted, unnormalized operators ${\mathcal{W}}_i$ just defined, connected correlators scale as

$$⟨{\mathcal{O}}_1\cdots {\mathcal{O}}_k{⟩}_{\mathrm{conn}}=O(N^{2-k}).$$

Hence as $N\to \infty$, $k$-point functions factorize:

$$⟨{\mathcal{O}}_1{\mathcal{O}}_2{\mathcal{O}}_3{\mathcal{O}}_4⟩\to ⟨{\mathcal{O}}_1{\mathcal{O}}_2⟩⟨{\mathcal{O}}_3{\mathcal{O}}_4⟩+⟨{\mathcal{O}}_1{\mathcal{O}}_3⟩⟨{\mathcal{O}}_2{\mathcal{O}}_4⟩+⟨{\mathcal{O}}_1{\mathcal{O}}_4⟩⟨{\mathcal{O}}_2{\mathcal{O}}_3⟩+O(1/N^2).$$

The four-point function factorizes into sums of products of two-point functions — exactly Wick’s theorem for a Gaussian random variable. The algebra of single-trace operators at $N=\infty$ is a generalized free field: a (non-Lagrangian) field whose $n$-point functions are determined by Wick contraction with a given two-point function.

This is the key structural input. The algebra is not the original gauge theory’s algebra; it is the leading large-$N$ limit, which has Gaussian (free-field-like) correlation structure but inherits its two-point function from the underlying CFT.

3.4 Generalized free fields

Definition 3.3. A generalized free field (GFF) on a spacetime $M$ is a quantum field whose correlation functions are Gaussian:

$$⟨\mathcal{O}(x_1)\cdots \mathcal{O}(x_{2k})⟩=\sum _{\text{pairings}}\prod _{\text{pairs }(i,j)}W(x_i,x_j),$$

with $W(x,y)$ a positive Wightman two-point function. Odd correlators vanish.

The two-point function $W(x,y)$ is the only input. A free scalar field is a GFF with the standard Klein–Gordon $W$; the single-trace operator at large $N$ is a GFF with the CFT’s two-point function as input.

Key difference from a free Lagrangian field. The GFF generally does not satisfy any local equation of motion. There is no Lagrangian; only a two-point function. This is exactly the structural feature of large-$N$ single-trace operators.

Physical picture: what Gaussianity does and does not buy us. Factorization isolates a generalized-free sector whose correlators are determined by a two-point function. In AdS/CFT this sector is naturally interpreted as perturbative bulk fields on a fixed background, while $1/N$ corrections encode interactions and gravitational backreaction. However, Gaussianity by itself neither fixes the bulk spacetime nor determines the von Neumann type. Those conclusions require the spectral and holographic input supplied by the state and the region.

4. The boundary single-trace algebra

4.1 Setup

Consider a holographic CFT${}_d$ dual to a $(d+1)$-dimensional bulk gravity theory. Choose a boundary subregion (e.g., one side of an eternal AdS-Schwarzschild black hole) and consider the algebra of single-trace operators smeared in that region.

At large $N$, this algebra is a Weyl-algebra-like object generated by Gaussian-correlated operators. The boundary subregion $\mathcal{O}$ has an associated local algebra ${\mathcal{A}}_R(\mathcal{O})$ in the GFF representation.

4.2 Type III${}_1$ at large $N$

Result used in this block (strict large-$N$ non-central algebra). [Stated only; refs: Witten 2022 §2; Leutheusser–Liu 2021.] In the high-temperature TFD setting analyzed by Leutheusser and Liu, the von Neumann algebras ${\mathcal{A}}_{R,0}$ and ${\mathcal{A}}_{L,0}$ generated by bounded functions of the non-central, subtracted single-trace operators are type III${}_1$ factors.

Two qualifications prevent us from turning this into an overly broad theorem.

1. The statement concerns an emergent large-$N$ representation. It is not a claim that every algebra generated by Gaussian variables is type III${}_1$.

2. The central energy variable is excluded from ${\mathcal{A}}_{R,0}$. When it is adjoined at strict $N=\infty$, the resulting algebra has a nontrivial center and is therefore not a factor.

The bulk argument identifies ${\mathcal{A}}_{R,0}$ with the algebra of low-energy effective fields in a black-hole exterior at fixed collective charges. The type-III${}_1$ conclusion is then inherited from the local-QFT algebra. This is a holographic identification, not a deduction from nuclearity and the split property of the boundary CFT.

4.3 The central energy coordinate at strict $N=\infty$

The one-sided Hamiltonian fluctuation $H_R^{\prime }=H_R-⟨H_R⟩$ is of order $N$ and has no limit as an operator on the large-$N$ TFD Hilbert space. The rescaled variable

$$U:=\frac{H_R^{\prime }}{N}$$

does have a limit, but its commutator with every fixed single-trace operator is $O(1/N)$. Thus $U$ is central at strict $N=\infty$. Adjoining it gives, schematically,

$${\mathcal{A}}_R^{(0)}\simeq {\mathcal{A}}_{R,0}\bar{\otimes }{L}^{\infty }({\mathbb{R}}_U),$$

which is not a factor. This is the point that makes a perturbatively small correction capable of changing the algebraic type.

4.4 What the first noncentral correction does

Witten retains the leading commutator by which the one-sided energy generates time translations. More explicitly, in his asymmetric presentation the bulk dual of the rescaled boundary energy is

$$\frac{H_R^{\prime }}{N}=U+\frac{\hat{h}}{{\beta }_{H}N},$$

order by order in $1/N$. The $O(1/N)$ term is what makes the generator noncentral. At the same time, the coefficient ring for gauge-invariant operator products is expanded in $1/N^2$, and the usual holographic coupling scales as $G_N\sim 1/N^2$. There is no contradiction: these are different bookkeeping statements, and the definition of $U=H_R^{\prime }/N$ has already extracted one power of $N$. In particular, the algebraic mechanism in this equation should not be relabelled an $O(1/N^2)$ commutator effect.

Once this action is retained, the extended algebra becomes

$${\mathcal{A}}_R\simeq {\mathcal{A}}_{R,0}{⋊}_{{\sigma }^{\mathrm{TFD}}}\mathbb{R},$$

a type II${}_{\infty }$ factor. The logical sequence is therefore

$$\underset{{\text{III}}_1\text{ factor}}{\underbrace{{\mathcal{A}}_{R,0}}}⟶\underset{\text{central }U;\text{not a factor}}{\underbrace{{\mathcal{A}}_R^{(0)}}}\stackrel{\text{perturbative }1/N\text{ dressing}}{\to }\underset{{\text{II}}_{\infty }\text{ factor}}{\underbrace{{\mathcal{A}}_R}}.$$

At a fixed finite integer $N$, the complete nonperturbative boundary theory is expected to be type I. This last statement concerns the full quantum system; it should not be confused with the type of a sharply localized continuum-QFT subregion, which remains type III under the usual AQFT hypotheses.

Physical picture. The type-II algebra is neither the strict classical limit nor the exact finite-$N$ theory. It is the algebra seen at a particular perturbative resolution: enough of the energy fluctuation is retained to remove the large-$N$ center, but not enough nonperturbative information is kept to resolve individual black-hole microstates. This is why it supports traces and entropies but no pure normal states.

5. The modular flow at large $N$

The crucial structural input is more precise than the slogan “modular flow equals ADM time.” On the TFD GNS Hilbert space, the difference $\hat{H}=H_R-H_L$ has a large-$N$ limit even though $H_R$ and $H_L$ separately do not. Witten obtains

$${\Delta }_{\mathrm{TFD}}=e^{-{\beta }_H\hat{H}}.$$

Consequently the modular automorphism of the right algebra is the right-boundary time translation, rescaled by ${\beta }_H$.

5.1 The ADM Hamiltonian

In the bulk gravity theory, the ADM Hamiltonian $H_{\mathrm{ADM}}$ is the energy of a configuration as measured at the asymptotic boundary. For an asymptotically AdS spacetime with a boundary time coordinate $t$, $H_{\mathrm{ADM}}$ generates time translation on the boundary.

At finite $N$, $H_R$ is a well-defined operator in the complete right CFT. In the strict large-$N$ TFD representation, however, its fluctuations diverge. Time translations survive as automorphisms of ${\mathcal{A}}_{R,0}$, but their one-sided generator is not an element of that algebra.

5.2 Modular flow = ADM time

TFD modular-flow identification (Witten 2022 §2). [Stated only.] For the right non-central algebra ${\mathcal{A}}_{R,0}$, the modular operator is $\Delta =e^{-{\beta }_H\hat{H}}$ with $\hat{H}=H_R-H_L$. Hence, in the course convention,

$${\sigma }_t^{\mathrm{TFD}}(a)={\Delta }^{-it}a{\Delta }^{it}=e^{i{\beta }_{H}t\hat{H}}a{e}^{-i{\beta }_{H}t\hat{H}}={\alpha }_{{\beta }_{H}t}^R(a),a\in {\mathcal{A}}_{R,0}.$$

The last equality is an equality of automorphisms; it does not assert that $H_R$ exists as a one-sided operator in the limiting algebra.

This is the holographic generalization of Bisognano–Wichmann (Sem I Wk 10): in the free-field Rindler case, the modular flow on the wedge algebra is the Lorentz boost; in the holographic case, the modular flow on the boundary subregion is the ADM time translation. The identification is at large $N$ and uses the holographic dictionary.

5.3 The structural picture

We now have all the algebraic pieces:

Object	Free-field Rindler (Sem I)	Holographic large-$N$ (Witten 2022)
Starting factor	$\mathcal{A}(W_R)$, type III${}_1$	${\mathcal{A}}_{R,0}$, type III${}_1$ at strict large $N$
Leading collective coordinate	an external clock variable	$U=H_R^{\prime }/N$, central at strict $N=\infty$
State	Minkowski vacuum (TFD of boost)	TFD vacuum
Modular flow	boost ${\sigma }_t^{W_R}$	ADM time ${\alpha }_{{\beta }_{H}t}^{H_R}$
Modular implementer on the standard Hilbert space	$2\pi K$, with $K$ the signed global boost generator; its right action is the wedge boost	${\beta }_H(H_R-H_L)$; its right action is ADM time
KMS temperature	Unruh $T_U=1/(2\pi )$	Hawking $T_H=1/{\beta }_H$
Crossed product algebra	$\hat{\mathcal{A}}(W_R)=\mathcal{A}(W_R){⋊}_{\mathrm{boost}}\mathbb{R}$	${\mathcal{A}}_R={\mathcal{A}}_{R,0}{⋊}_{{\sigma }^{\mathrm{TFD}}}\mathbb{R}$ after the noncentral $\hat{h}/({\beta }_{H}N)$ dressing is retained
Type of dressed algebra	type II${}_{\infty }$	type II${}_{\infty }$
Dressed entropy	trace entropy with a geometric interpretation only after extra input	generalized entropy $A/(4G_N)+S_{\mathrm{out}}$ in Witten’s stated semiclassical regime, up to the trace-normalization constant

The free-field Rindler model supplies a concrete realization of the continuous core and its dual action. It does not reproduce the holographic derivation of the collective energy variable, Witten’s large-$N$ bookkeeping, or the area interpretation of the entropy. For an inner unitary with the same clock lift, traciality instead gives an exact entropy cancellation.

6. Worked check: the BTZ thermal correlator

This example checks the KMS normalization. It is not a derivation of the algebraic type.

6.1 BTZ black hole

The 3D Bañados–Teitelboim–Zanelli black hole is the simplest non-trivial AdS${}_3$ black hole, with metric

$$d{s}^2=-\frac{r^2-r_h^2}{{\ell }^2}d{t}^2+\frac{{\ell }^2}{r^2-r_h^2}d{r}^2+r^{2}d{\phi }^2,$$

where $\ell$ is the AdS radius and $r_h$ is the horizon radius. The Hawking temperature is $T_H=r_h/(2\pi {\ell }^2)$.

The two-sided BTZ is the maximally extended version, with two asymptotic boundaries (left and right). The boundary CFT${}_2$ lives on a cylinder ${\mathbb{R}}_t\times S_{\phi }^1$ on each boundary.

6.2 Boundary CFT and single-trace operators

For a primary with weights $(h,\bar{h})$ on the thermal line, the Lorentzian two-point function is, up to normalization and the usual $iϵ$ prescription,

$$⟨\mathcal{O}(t,x)\mathcal{O}(0,0){⟩}_{\beta }\propto {\left[\frac{\pi /\beta }{\sinh \left(\frac{\pi }{\beta }(x-t+iϵ)\right)}\right]}^{2h}{\left[\frac{\pi /\beta }{\sinh \left(\frac{\pi }{\beta }(x+t-iϵ)\right)}\right]}^{2\bar{h}}.$$

For a spatial circle, the BTZ answer includes the corresponding image sum. Analytically continuing $t\mapsto t+i\beta$ and exchanging the operator order verifies the KMS boundary condition. This is the only conclusion needed here.

The boundary single-trace algebra is generated by smearings ${\mathcal{O}}_{\Delta }(f)=\int d^{2}xf(x){\mathcal{O}}_{\Delta }(x)$, with the GFF structure determined by the above two-point function.

6.3 Modular structure

The boundary subregion is the right asymptotic region $\{t\in \mathbb{R},\phi \in S^1\}$, with $r\to \infty$. The single-trace algebra is the GFF algebra at $N=\infty$.

The TFD vector at $\beta ={\beta }_H$ is cyclic and separating for the one-sided algebra in the standard large-$N$ representation. The modular flow on that algebra is ${\sigma }_t^{\mathrm{TFD}}={\alpha }_{{\beta }_{H}t}^R$. The thermal two-point function checks the KMS normalization, while cyclicity, separatingness, and the von Neumann type require separate algebraic input.

The two-point function on the right side, restricted to one boundary, looks thermal at ${\beta }_H$. This is the holographic analog of “the Minkowski vacuum looks thermal at the Unruh temperature to a Rindler observer.”

6.4 Type III${}_1$ for the BTZ boundary subregion

[Stated only — refs: Witten 2022 §2; Leutheusser–Liu.] In the high-temperature large-$N$ representation, the corresponding non-central one-sided algebra is type III${}_1$. The thermal correlator establishes KMS behavior; it does not by itself compute Connes’ $S$-invariant. The type statement requires the full algebraic analysis or the bulk-effective-field-theory identification.

7. Discussion: two large-$N$ bookkeepings

A few interpretive remarks before Week 2’s construction:

7.1 The two limits

• Strict $N=\infty$, non-central sector: ${\mathcal{A}}_{R,0}$ is type III${}_1$ and its modular flow is outer.
• Strict $N=\infty$, energy adjoined: $U$ is central, so the extended algebra is not a factor.
• Perturbative algebra: $U$ is dressed by $\hat{h}/({\beta }_{H}N)$, and this $O(1/N)$ term removes the center and produces the type-II${}_{\infty }$ crossed product. The separate $1/N^2$ coefficient expansion and $G_N\sim 1/N^2$ scaling remain part of the physical large-$N$ dictionary, not a replacement for the order shown in the generator.
• Exact finite integer $N$: the complete boundary quantum system is expected to be type I; this lies beyond the perturbative algebra.

Thus the type II${}_{\infty }$ algebra belongs to the semiclassical expansion, not to strict $N=\infty$ and not to the exact finite-$N$ theory.

7.2 Why this is the right framework

The operator-algebra theorem about the continuous core is structural. Its gravitational realization is conditional on Witten’s large-$N$ setting, on the chosen collective charges, and on the semiclassical identification of the trace density. It should not be advertised as a theorem for every holographic CFT or every boundary subregion.

7.3 What’s missing (next-week preview)

The Witten 2022 setup gives us:

1. ✓ The algebra: boundary single-trace algebra at large $N$.
2. ✓ The flow: ADM time translation = modular flow.
3. ✗ The crossed product itself: not yet constructed.
4. ✗ The dressed entropy formula: not yet derived.
5. ✗ Identification with generalized entropy: Witten’s key claim, pending.

Weeks 2–3 build the missing pieces. Week 4 tests the continuous-core formulas in the free-field analog and marks the gravitational identifications as external input.

8. What to take away

• Witten 2022 architecture, with the order of limits visible: ${\mathcal{A}}_{R,0}$ is III${}_1$; adjoining $U$ at strict $N=\infty$ gives a nonfactor with center; retaining $U+\hat{h}/({\beta }_{H}N)$ order by order gives the II${}_{\infty }$ crossed product.
• Large-$N$ factorization turns single-trace operators into a Gaussian random field (GFF) at $N=\infty$. The boundary algebra is the GFF algebra.
• Gaussianity is not a type-classification theorem. The III${}_1$ statement is imported from the Leutheusser–Liu analysis and the bulk local-QFT identification.
• Three algebraic regimes must not be conflated: strict large $N$, the perturbative $1/N$ construction with physical corrections organized in $G_N\sim 1/N^2$, and exact finite $N$.
• Modular flow = ADM time at large $N$ — the holographic generalization of Bisognano–Wichmann.
• The free-field Rindler model tests the continuous core, but not the gravitational origin of the clock or the generalized-entropy identification.

9. Looking ahead

Week 2 turns the setup into Witten’s central construction: form ${\mathcal{A}}_R={\mathcal{A}}_{R,0}{⋊}_{{\sigma }^{\mathrm{TFD}}}\mathbb{R}$, use Takesaki’s structure theorem to identify its type, and study the trace. Week 3 reads Witten’s entropy construction in §§3.4–3.5. Week 4 then checks the operator-algebraic formulas in a free-field Rindler model without claiming to reproduce the gravitational $1/N$ mechanism.

10. Problem set

Core problems.

1. Wick contractions in a GFF. For a generalized free field with two-point function $W(x,y)$, prove that

$$⟨\mathcal{O}(x_1)\mathcal{O}(x_2)\mathcal{O}(x_3)\mathcal{O}(x_4)⟩=W(x_1,x_2)W(x_3,x_4)+W(x_1,x_3)W(x_2,x_4)+W(x_1,x_4)W(x_2,x_3),$$

i.e., the four-point function factorizes into the three Wick contractions. Compare with the free scalar field’s four-point function (Sem I Wk 8).

2. The ‘t Hooft expansion for $\mathrm{Tr}({\Phi }^2)\mathrm{Tr}({\Phi }^2)$. In a $\mathrm{U}(N)$ matrix theory with $\Phi$ a Hermitian random matrix and Gaussian measure $e^{-N\mathrm{Tr}({\Phi }^2)/2}$, compute the leading $1/N$ behavior of $⟨\mathrm{Tr}({\Phi }^2)\mathrm{Tr}({\Phi }^2){⟩}_{\mathrm{connected}}$. (Hint: the connected part is $O(N^0)$, while the disconnected is $O(N^2)$, illustrating large-$N$ factorization.)

3. The central energy variable. Starting from $[H_R,\mathcal{W}]=-i{\partial }_t\mathcal{W}$, show that $U=H_R^{\prime }/N$ commutes with every fixed single-trace operator in the $N\to \infty$ limit. Explain why adjoining $U$ at leading order produces a nontrivial center rather than a type II factor.

4. Thermal two-point function and KMS. Derive the formula in §6.2 from the exponential map from the plane to the thermal cylinder. Verify its KMS continuation at $t\mapsto t+i\beta$, keeping the $iϵ$ prescription and operator order explicit.

5. Read Witten 2022 §§1–3.2. Make a three-row table for ${\mathcal{A}}_{R,0}$, the strict-$N=\infty$ algebra with central $U$, and the perturbatively corrected crossed product. For each row record the generators, center, and type. This table is the main correctness check for the week.

Starred problems.

6*. Read Liu’s §§VI.1–VI.3 and VI.6. Liu (arXiv:2510.07017) gives a pedagogical algebraic formulation of large-$N$ AdS/CFT and then discusses conserved-charge corrections. Identify the input for the type-III statement in the TFD sector, and compare Liu’s bookkeeping of $1/N$ corrections with Witten’s $U+\hat{h}/({\beta }_{H}N)$ construction. Consult §§IV.1–IV.4 only for the operator-algebra background.

7*. KMS at ${\beta }_H$ for the TFD. Show that the ordered boundary two-point functions obey the KMS boundary relation after analytic continuation by $i{\beta }_H$. Explain why simple imaginary-time periodicity is sufficient for bosonic Euclidean correlators but the Lorentzian KMS statement must retain the operator ordering.

8*. Finite-$N$ corrections. Sketch the structure of $1/N$ corrections to the GFF correlators. Show that connected $k$-point functions for $k>2$ scale as $1/N^{k-2}$, hence the algebra becomes non-Gaussian at finite $N$.

9*. The Leutheusser–Liu approach. Read Leutheusser & Liu, arXiv:2110.05497, §§1–3. Identify their construction of the type-III boundary algebra and how it differs from Witten 2022’s. In particular: where does the “type III emerges at large $N$” structurally come from in their framework?

Project problems.

10. Self-contained writeup. Write a 5-page exposition of the large-$N$ type-III${}_1$ result for a non-specialist physics graduate student. Identify which steps are CFT input, which are large-$N$ structural arguments, and which are operator-algebraic. Cite Witten 2022 and Liu lectures.

---

Notes prepared for the algebraic-QFT course, Semester II Block 1. Last revised 2026-09-29.