Sem II Week 2 — Witten’s Construction: Energy as a Collective Coordinate

Witten’s mechanism is more precise than the slogan “gauge time.” At strict large $N$, the noncentral single-trace algebra is type III${}_1$, while a rescaled black-hole energy survives as a central collective coordinate. Perturbative $1/N$ corrections couple that coordinate to the modular generator, remove the centre, and produce the crossed-product factor of type II${}_{\infty }$. This week reconstructs that mechanism before translating it into the standard crossed-product representation. The distinction matters physically: an ADM Hamiltonian is a boundary charge, not a constraint that annihilates every physical state.

0. Reading

Primary:

• Witten, “Gravity and the crossed product,” arXiv:2112.12828, §§3.1–3.5.
• Sem I Week 13 and Appendix D for the regular representation and variable dictionary.

Secondary:

• Takesaki, Theory of Operator Algebras II, ch. X.
• Chandrasekaran, Penington, Witten, arXiv:2209.10454, §2, for the microcanonical formulation used in the next block.

0.1 How to use this master dossier

• Classroom core: §§1.1–1.4, 2.1–2.3, and 3.1–3.2, followed by Problems 1–5. This route establishes the physical mechanism, the regular representation, and the valid trace formula in a two-hour seminar.
• Full derivation / self-study: §§3.3–4 and §6, with Problems 6–9. This route checks the analytic trace domain, state independence, every convention sign, and the inner-action control model.
• Research extension: §5, Problems 10–11, and the primary-source comparison. This route separates Witten’s perturbative canonical construction from CPW’s strict microcanonical limit and from the finite-$N$ CFT.

The minimum learning outcome is not the slogan “gravity makes type III into type II.” A student should be able to say which algebra is type III, which energy variable is central at leading order, which retained term removes the centre, and where the gravitational interpretation enters.

1. What actually changes beyond strict large $N$

1.1 The noncentral algebra and the TFD generator

Let ${\mathcal{A}}_{R,0}$ denote the strict-large-$N$ von Neumann algebra generated by the noncentral right single-trace fluctuations in the TFD representation. In the Leutheusser–Liu setting it is a type III${}_1$ factor. Its commutant is the corresponding left algebra ${\mathcal{A}}_{L,0}$.

The individual subtracted Hamiltonians $H_R^{\prime }$ and $H_L^{\prime }$ have fluctuations of order $N$ and do not have limits as operators on the large-$N$ TFD Hilbert space. Their difference does:

$$\hat{H}:=H_R^{\prime }-H_L^{\prime }.$$

The TFD modular operator satisfies

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

Thus the automorphism of ${\mathcal{A}}_{R,0}$ is well-defined even though $H_R^{\prime }$ itself is not an element of that algebra.

1.2 The energy collective coordinate at $N=\infty$

Introduce the two rescaled boundary-energy fluctuations

$$U_R:=\frac{H_R^{\prime }}{N},U_L:=\frac{H_L^{\prime }}{N}.$$

Their difference is

$$U_R-U_L=\frac{\hat{H}}{N},$$

which vanishes at leading order because $\hat{H}=O(1)$ on the limiting TFD Hilbert space. Thus both approach one common leading collective coordinate, which we denote by $U$. Moreover,

$$[U_R,a]=\frac{1}{N}[H_R,a]⟶0$$

for fixed right single-trace $a$, while $U_L$ commutes with right operators by left–right commutativity. Hence the common limit $U$ is central at strict large $N$. Adjoining bounded functions of $U$ gives an algebra with a nontrivial centre, schematically

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

not a factor.

1.3 The perturbative coupling that removes the centre

At the next order the difference between the two energies can no longer be erased. Choose the asymmetric convention in which the auxiliary multiplication variable represents $U_L$ and write it simply as $U$. Since

$$\hat{h}:={\beta }_H\hat{H}={\beta }_H(H_R^{\prime }-H_L^{\prime })=-\log \Delta ,$$

one has the exact bookkeeping identity

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

Here the equality is exact once $U$ denotes $H_L^{\prime }/N$; what is perturbative is retaining its $O(1/N)$ noncommuting term in the large-$N$ algebra. The $U$ term commutes with ${\mathcal{A}}_{R,0}$, but $\hat{h}$ does not: $e^{it\hat{h}}={\Delta }^{-it}$ implements the modular automorphisms. After keeping the scaled combination that retains this action, the enlarged algebra is the crossed product

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

The qualitative transition is therefore

$$\text{III1 factor plus a central energy variable}⟶\text{II∞ crossed-product factor}.$$

Witten’s result is perturbative in $1/N$. At fixed integer $N$, the full CFT algebra is expected to be type I; the type II${}_{\infty }$ factor describes the perturbative large-$N$ regime, not the exact finite-$N$ theory.

The three levels should remain separate:

Regime	Energy sector	Algebraic conclusion
leading strict $N=\infty$	one central variable $U$	${\mathcal{A}}_{R,0}\bar{\otimes }{L}^{\infty }({\mathbb{R}}_U)$, not a factor
perturbative large $N$ with the modular term retained	$U+\hat{h}/({\beta }_{H}N)$ in the rescaled generator	type II${}_{\infty }$ continuous core
fixed finite $N$	full boundary Hamiltonian exists in the ordinary CFT representation	expected type I full algebra

The middle row is not obtained by taking the first row and merely renaming $U$ a clock. The noncommuting modular term is the mechanism.

1.4 Boundary time is not a vanishing constraint

In canonical gravity on a spacetime with an asymptotic boundary, the Hamiltonian is a sum of bulk constraints and a boundary ADM charge. Diffeomorphisms that vanish at the boundary are gauge; an asymptotic time translation acts nontrivially and is generated by the physical ADM charge. It is therefore incorrect to impose $H_{\mathrm{ADM}}|\Psi ⟩=0$ on every black-hole state.

The relational language remains useful. The auxiliary collective coordinate records the asymptotic energy and its conjugate timeshift. But “gauging time” should be read as shorthand for incorporating the gravitational constraint structure together with the boundary charge—not as declaring the ADM energy to be zero.

2. The crossed product in course notation

2.1 Variable dictionary

We keep the notation of Appendix D:

Symbol	Role
$t$	dimensionless modular parameter
$u={\beta }_{H}t$	physical boundary time
$q$	scalar coordinate in the regular representation $L^2({\mathbb{R}}_q)$
$Q$	multiplication operator, $(Q\xi )(q)=q\xi (q)$
$P=-i{\partial }_q$	conjugate momentum; $\lambda (t)=e^{-itP}$
$p$	Fourier spectral variable of $P$
$r$	dual-action parameter

Witten instead uses a multiplication variable $X$ in a Fourier-equivalent realization. With the Fourier convention adopted here, the clean sign map is

$$p=-X_{\mathrm{Witten}}.$$

Consequently, Witten’s trace weight $e^{X}dX$ becomes the course weight $e^{-p}dp$. His conjugate operator also changes role:

$$P_{\mathrm{Witten}}=-i{\partial }_X=i{\partial }_p=Q$$

in the Fourier-transformed course representation. Thus $X_{\mathrm{Witten}}$ is a multiplication coordinate in Witten’s realization, but it is not our regular coordinate $q$; it is the negative of our Fourier variable $p$.

2.2 Regular covariant representation

On ${\mathcal{H}}_{\mathrm{TFD}}\otimes L^2({\mathbb{R}}_q)$ define

$$(\pi (a)\xi )(q)={\sigma }_{-q}^{\mathrm{TFD}}(a)\xi (q),(\lambda (t)\xi )(q)=\xi (q-t).$$

The right algebra is

$${\hat{\mathcal{A}}}_R:={\mathcal{A}}_{R,0}{⋊}_{{\sigma }^{\mathrm{TFD}}}\mathbb{R}=\{\pi ({\mathcal{A}}_{R,0}),\lambda (t):t\in \mathbb{R}{\}}^{\prime \prime }.$$

Direct computation gives

$$\lambda (t)\pi (a)\lambda (t{)}^{\ast }=\pi ({\sigma }_t^{\mathrm{TFD}}(a)).$$

Since $u={\beta }_{H}t$, the unitary $\lambda (u/{\beta }_H)$ implements physical right-boundary time translation on $\pi ({\mathcal{A}}_{R,0})$. It is an abstract implementer in the crossed product. We should not identify it with $e^{iu{H}_R}$ as an operator on the strict-large-$N$ TFD Hilbert space, because $H_R$ itself has no such limit.

2.3 Dual action

The dual action is

$${\theta }_r(\pi (a))=\pi (a),{\theta }_r(\lambda (t))=e^{itr}\lambda (t).$$

It is implemented in the ambient regular representation by $e^{irQ}$ and acts in the Fourier picture by

$${\theta }_r(f(P))=f(P-r).$$

The fixed-point algebra is $\pi ({\mathcal{A}}_{R,0})$.

3. Type, trace, and the correct explicit formula

3.1 Type promotion

Because ${\mathcal{A}}_{R,0}$ is a type III${}_1$ factor and ${\sigma }^{\mathrm{TFD}}$ is its modular flow,

$$\boxed{{\hat{\mathcal{A}}}_R={\mathcal{A}}_{R,0}{⋊}_{{\sigma }^{\mathrm{TFD}}}\mathbb{R}\text{ is a type II∞ factor}.}$$

It carries a faithful normal semifinite trace $\hat{\tau }$, unique up to scale, with

$$\hat{\tau }\circ {\theta }_r=e^{-r}\hat{\tau }.$$

3.2 Witten’s $X$-representation

For a classical-quantum vector

$$|\hat{\Psi }⟩=|\Psi ⟩\otimes g^{1/2}(X),$$

Witten realizes the crossed product as the algebra generated by $a\in {\mathcal{A}}_{R,0}$ and

$${\Delta }_{\Psi }^{-it}{e}^{itX}.$$

For operators in the trace domain, his §3.4 formula is

$${\mathrm{Tr}}_{\mathrm{W}}(\hat{a})={\int }_{\mathbb{R}}{e}^X⟨\Psi |\hat{a}(X)|\Psi ⟩dX.$$

The expectation is generally a nontrivial function of $X$; that dependence is required for convergence. With $p=-X$, this becomes

$$\hat{\tau }(\hat{a})={\int }_{\mathbb{R}}{e}^{-p}⟨\Psi |\hat{a}(-p)|\Psi ⟩dp,$$

up to the overall normalization of the trace.

This is a legitimate formula in Witten’s specified realization. It is not the distributional expression

$$\int e^{-{\beta }_{H}q}⟨\Psi \otimes {\delta }_q|\hat{a}|\Psi \otimes {\delta }_q⟩dq,$$

which mixes the regular coordinate $q$ with the Fourier/energy variable and does not define the type-III-core trace.

3.3 The trace ideal

The trace is not finite on the identity or on a nonzero element of the original type III algebra. A useful dense family is obtained from smooth crossed-product Fourier integrals

$$\hat{a}={\int }_{\mathbb{R}}a(t)\lambda (t)dt,$$

with sufficient decay and analytic continuation to the KMS strip. Witten derives, on an appropriate analytic subalgebra, a contour-shift expression in terms of the value of $a(t)$ at imaginary modular time. The analytic hypothesis is essential; the formula is not a delta-function rule for individual monomials.

Here is the derivation in the form useful for the blackboard. In Witten’s variables write

$$\hat{a}={\int }_{\mathbb{R}}a(v)e^{iv(\hat{h}+X)}dv,$$

where $a(v)$ is operator-valued, analytic in $0\le \mathrm{Im}v\le 1$, and decreases fast enough uniformly in that strip. Since $\hat{h}|\Psi ⟩=0$,

$${\mathrm{Tr}}_{\mathrm{W}}(\hat{a})={\int }_{{\mathbb{R}}^2}{e}^{X+ivX}⟨\Psi |a(v)|\Psi ⟩dXdv.$$

Move the $v$ contour from $\mathbb{R}$ to $\mathbb{R}+i$. No pole may be crossed, and the vertical pieces at infinity must vanish; those are precisely the places where analyticity and decay are used. Put $v=w+i$. Then

$$e^{X+i(w+i)X}=e^{iwX},$$

so Fourier inversion gives

$$\begin{array}{rl}{\mathrm{Tr}}_{\mathrm{W}}(\hat{a})& ={\int }_{{\mathbb{R}}^2}{e}^{iwX}⟨\Psi |a(w+i)|\Psi ⟩dXdw\\ & =2\pi ⟨\Psi |a(i)|\Psi ⟩.\end{array}$$

This computation establishes the value on the analytic core. To prove cyclicity on products $\hat{a}\hat{b}$, one additionally uses the KMS boundary relation to move a modularly evolved coefficient through the expectation value before shifting the contour. “Analyticity,” “decay,” and “KMS” therefore do three different jobs; none can be replaced by a formal delta vector.

3.4 Why the result is II${}_{\infty }$, not II${}_1$

In Witten’s representation,

$${\mathrm{Tr}}_{\mathrm{W}}(1)={\int }_{\mathbb{R}}{e}^{X}dX=\infty .$$

The trace is nevertheless semifinite: the analytic trace ideal contains nonzero positive elements of finite trace, and every positive element can be approximated from below in the sense required by semifiniteness. The identity having infinite trace is exactly what distinguishes the II${}_{\infty }$ result from a finite II${}_1$ algebra. Factoriality is a separate conclusion supplied by the type III${}_1$ continuous-core theorem.

3.5 Entropy normalization

If $\hat{\tau }\to c\hat{\tau }$, then the density of a fixed normalized state changes by $D\to D/c$, so

$$S\to S+\log c.$$

This is the additive ambiguity Witten compares with the normalization ambiguity of black-hole entropy.

4. State independence of the core

Let $\Psi$ and $\Omega$ be faithful normal states of ${\mathcal{A}}_{R,0}$. Their modular actions are related by the Connes cocycle. The corresponding crossed products are canonically isomorphic: schematically, the cocycle corrects the group generator so that

$${\Delta }_{\Omega }^{-it}{e}^{itX}⟼{\Delta }_{\Psi }^{-it}{e}^{itX}.$$

This is the precise sense in which the continuous core depends on the algebra rather than on a chosen background state. Two cautions accompany the statement:

1. “Canonically isomorphic” does not mean that the two concrete subalgebras are literally equal before the cocycle identification.
2. A simple product class $\Psi \otimes g^{1/2}(X)$ need not remain a simple product after the identification.

These cautions become important in AAJ, where the state and the modular structure are perturbed simultaneously.

5. What the gravitational interpretation adds

The abstract continuous core already explains why the algebra is type II${}_{\infty }$ and why its trace is unique up to scale. Gravity supplies three further statements:

1. the auxiliary spectral variable is the black-hole energy collective coordinate;
2. its conjugate variable is a relative boundary timeshift;
3. the additive normalization of trace entropy is naturally measured relative to a reference black-hole entropy. The explicit equality with $A/(4G_N)+S_{\mathrm{out}}$ requires the additional semiclassical argument developed by CPW.

The timeshift is physical in the asymptotic theory: eliminating it requires a diffeomorphism that acts nontrivially at the boundary. Thus it should not be described as a pure gauge artifact.

Physical picture. At strict large $N$, one can choose a black-hole mass and study type-III quantum fluctuations on that fixed background. The mass then behaves as a classical label. The crossed product promotes that label and its conjugate timeshift into a quantum collective coordinate coupled to modular flow. The centre disappears because energy no longer merely labels superselection sectors: it participates in the algebra.

6. Exact finite-dimensional audit

Take $\mathcal{M}=M_n(\mathbb{C})$ with faithful $\rho$. Its modular action is inner, so

$$\mathcal{M}{⋊}_{{\sigma }^{\rho }}\mathbb{R}\cong M_n(\mathbb{C})\bar{\otimes }{L}^{\infty }({\mathbb{R}}_p).$$

On finite-rank/compact-support positive elements,

$$\hat{\tau }(a\otimes f(P))=\mathrm{Tr}(a)\int f(p)e^{-p}dp.$$

The dual action sends $f(P)$ to $f(P-r)$, hence

$$\hat{\tau }({\theta }_r(a\otimes f(P)))=e^{-r}\hat{\tau }(a\otimes f(P)).$$

One can also audit the density sign. Let ${\rho }_0$ be a density matrix and $\mu (p)dp$ a probability law. Define

$$\hat{\omega }(a\otimes f(P))=\mathrm{Tr}({\rho }_{0}a)\int \mu (p)f(p)dp.$$

The equation $\hat{\omega }(x)=\hat{\tau }(Dx)$ forces

$$D(p)={\rho }_0\mu (p)e^p,$$

because the factor $e^p$ must cancel the trace weight $e^{-p}$. Provided the logarithmic moments exist,

$$\begin{array}{rl}{S}_{\hat{\tau }}(D)& =-\int e^{-p}\mathrm{Tr}\left[{\rho }_0\mu (p)e^p\log ({\rho }_0\mu (p)e^p)\right]dp\\ & =S({\rho }_0)+H(\mu )-{\mathbb{E}}_{\mu }[p].\end{array}$$

This is the cleanest local check of the signs in the trace and density. It is not a substitute for Witten’s type-III construction: because the original action is inner, the model has centre $L^{\infty }({\mathbb{R}}_p)$ and is not a II${}_{\infty }$ factor. The type-III theorem is doing real work in Witten’s setting.

7. What to take away

• Witten’s type II${}_{\infty }$ algebra appears after the energy collective coordinate couples to the modular generator at perturbative order in $1/N$.
• At strict large $N$, the rescaled energy variable is central; the intermediate extended algebra is not a factor.
• The well-defined TFD generator is $\hat{H}=H_R-H_L$. The individual $H_R,H_L$ need not exist on the large-$N$ TFD Hilbert space.
• An ADM Hamiltonian is a physical boundary charge. “Gauge time” is only shorthand for the full constraint-plus-boundary-charge construction.
• In course notation, $q$ is the regular coordinate, $(Q\xi )(q)=q\xi (q)$, $P=-i{\partial }_q$, $p$ is the $P$-spectrum, $u={\beta }_{H}t$, and $r$ labels the dual action.
• Witten’s $X$ is Fourier-equivalent to $-p$ in our convention, while $P_{\mathrm{Witten}}$ corresponds to $Q$; $e^{X}dX$ therefore agrees with $e^{-p}dp$.
• The trace formula is valid on its trace ideal. There is no general ${\delta }_q$ diagonal formula for the core.
• The continuous core is state-independent up to the Connes-cocycle isomorphism, while simple product states are not preserved automatically by that isomorphism.

8. Looking ahead

Week 3 uses Witten’s density-matrix formula to identify the type-II entropy with black-hole entropy up to a constant. Week 4 then performs two exact checks: the inner-action Fourier calculation and the inner-unitary coherent-state cancellation. These checks are deliberately less ambitious than a free-field derivation of $1/(4G_N)$, which the algebra alone cannot provide.

9. Problem set

Core problems.

1. Centrality at strict large $N$. Starting from $U=H_R^{\prime }/N$, use the Heisenberg equation to show that $[U,a]\to 0$ for a fixed single-trace operator. Explain why adjoining $U$ alone produces a centre rather than a factor.

2. Covariance relation. Verify directly that $\lambda (t)\pi (a)\lambda (t{)}^{\ast }=\pi ({\sigma }_t(a))$ in the regular representation.

3. Seven-symbol dictionary. Distinguish $t,u,q,Q,P,p,r$. Write the actions of $Q$, $P$, $\lambda (t)$, and ${\theta }_r$ on a Schwartz vector. Fourier transform and verify ${\theta }_r(f(P))=f(P-r)$.

4. Trace scaling. In the finite-dimensional Fourier model, take positive finite-rank $a$ and compactly supported $f$. Verify $\hat{\tau }\circ {\theta }_r=e^{-r}\hat{\tau }$.

5. Trace rescaling. Show that $\hat{\tau }\to c\hat{\tau }$ sends $D\to D/c$ and $S\to S+\log c$.

Starred problems.

6*. Witten notation map. Read Witten §§3.2–3.4. Starting from his generators $a$ and ${\Delta }^{-it}{e}^{itX}$, derive the operator map $X_{\mathrm{Witten}}=-P$, hence $X_{\mathrm{Witten}}=-p$ on the $P$-spectrum, and $P_{\mathrm{Witten}}=Q$ for the Fourier convention of this course. Track the trace weight and the sign of the dual scaling. Explain why writing both $X_{\mathrm{Witten}}=Q$ and $X_{\mathrm{Witten}}=-p$ would mix two representations.

7*. Analytic trace core. In Witten’s convention take $\hat{a}=\int a(t)e^{it(\hat{h}+X)}dt$, with $a(t)$ rapidly decreasing and analytic in the required KMS strip. Reproduce the contour-shift argument leading to ${\mathrm{Tr}}_{\mathrm{W}}(\hat{a})=2\pi ⟨\Psi |a(i)|\Psi ⟩$. Mark where analyticity, decay, and the KMS boundary relation are each used; then translate the result to the course sign convention.

8*. State independence. Use the Connes cocycle to explain how the group generator changes when the reference state changes. Why does the isomorphism not preserve the naive product form $\Psi \otimes g^{1/2}(X)$?

9*. Boundary charge versus constraint. In canonical gravity with an asymptotic boundary, write schematically $H[\xi ]={\int }_{\Sigma }\xi \cdot \mathcal{C}+H_{\partial \Sigma }[\xi ]$. Explain which transformations are gauge and why the ADM boundary term need not annihilate physical states.

Project problems.

10. Witten mechanism memo. Write a four-page derivation organized as: strict-$N$ type III${}_1$ algebra; central energy coordinate; perturbative coupling to $-\log \Delta$; type II${}_{\infty }$ core; trace and entropy. Every arrow must state its approximation regime.

11. CPW bridge. Read CPW §2 and compare its microcanonical energy variable $h_R$ with Witten’s canonical-ensemble variable $U$. Explain why the algebra is already a II${}_{\infty }$ factor in the CPW microcanonical scaling.

10. Instructor checkpoints (internal)

These are grading checkpoints, not substitutes for the derivations.

1. Centrality: with ${\partial }_{u}a=i[H_R,a]$, one gets $[U_R,a]=-(i/N){\partial }_{u}a=O(N^{-1})$. The limiting bounded functions of $U$ commute with ${\mathcal{A}}_{R,0}$, so the centre contains $L^{\infty }({\mathbb{R}}_U)$.
2. Covariance: the fibre is evaluated at $q-t$, giving ${\sigma }_{t-q}(a)={\sigma }_{-q}({\sigma }_t(a))$. Any answer with $q+t$ has used the opposite translation convention.
3. Dictionary: $Q$ multiplies by $q$, $P=-i{\partial }_q$, $\lambda (t)=e^{-itP}$, $p$ is the $P$-spectrum, and $e^{irQ}$ implements $P\mapsto P-r$. The physical variable is $u={\beta }_{H}t$; $r$ is not time.
4. Trace scaling: the substitution $p^{\prime }=p-r$ yields the factor $e^{-r}$. The test element must remain in the stated trace ideal.
5. Trace rescaling: $D^{\prime }=D/c$ and the normalized-state entropy shift is $+\log c$.
6. Witten map: Fourier transformation gives $\lambda (t)\mapsto e^{-itp}$, so comparison with $e^{it{X}_{\mathrm{W}}}$ gives $X_{\mathrm{W}}=-P$ as an operator and $X_{\mathrm{W}}=-p$ spectrally; then $-i{\partial }_{X_{\mathrm{W}}}=i{\partial }_p=Q$ in the chosen convention.
7. Analytic trace: the contour is shifted to $v=w+i$, after which $e^{X+ivX}=e^{iwX}$ and the $X$ integral gives $2\pi \delta (w)$. Analyticity permits the shift, decay removes the vertical contours, and KMS proves cyclicity on products.
8. State independence: the appropriate Connes cocycle modifies the crossed-product implementing unitaries and gives an isomorphism of cores. Because that cocycle acts jointly on the system and group sector, a product wavefunction need not map to another product.
9. Boundary charge: transformations whose descriptors vanish at the boundary are generated by constraints and are gauge; a nonvanishing asymptotic time translation carries the ADM surface charge and acts physically.
10. Memo rubric: all three regimes in §1.3 must appear, and every use of “type II${}_{\infty }$” must be tied to the retained modular action rather than to the leading central algebra.
11. CPW bridge: CPW keep $h_R=H_R-E_0$ with $O(1)$ fluctuations in the strict microcanonical $N\to \infty$ limit. It already acts nontrivially on ${\mathcal{A}}_{R,0}$; no leading central $U$ algebra is the endpoint of that scaling.

---

Notes prepared for the algebraic-QFT course, Semester II Block 1. Last revised 2026-08-24.