Sem II Week 13 — Ahmad–Jefferson III: Open Questions and Positioning the Field

AAJ show that a crossed-product entropy can be followed under a unitary deformation of the full algebra-state system, and they apply the construction to a traversable wormhole through second order in the deformation and through $O(1/N^2)$. The next questions are not all questions AAJ themselves pose. This week therefore uses two labels: “AAJ outlook” for directions stated in §5, and “course-generated direction” for connections we propose after comparing the paper with modular reconstruction and entanglement embezzlement. The distinction makes the literature map useful rather than decorative.

Route through this master dossier. Classroom core (two-hour seminar): §§1–3, ending with the claim-status and literature-positioning ledger. Full derivation or self-study: §4’s catalyst calculation and the core exercises. Research extension or office hour: §5, the starred/project problems, and the proposal-building exercise that separates AAJ’s outlook from course-generated directions. The aim is to teach how a research frontier is delimited, not merely how it is advertised.

0. Reading and corrected references

Primary.

• Ahmad and Jefferson, arXiv:2501.01487v2, §5, “Discussion: the fate of the algebraic approach in quantum gravity.” The paper has five numbered sections; there is no §6.
• Re-read Eqs. (77)–(78): five additional linear terms and fifteen quadratic terms, twenty new terms through quadratic order.

Secondary.

• CPW, arXiv:2209.10454, final discussion.
• T. Faulkner, M. Li, and H. Wang, “A modular toolkit for bulk reconstruction,” arXiv:1806.10560. The title and identifier should be kept together; arXiv:2206.00027 is not this paper.

Embezzlement reading.

• W. van Dam and P. Hayden, “Universal entanglement transformations without communication,” quant-ph/0201041.
• L. van Luijk, A. Stottmeister, R. F. Werner, and H. Wilming, “Embezzlement of entanglement, quantum fields, and the classification of von Neumann algebras,” arXiv:2401.07299.

1. What AAJ established

AAJ study type-III${}_1$ algebra-state systems related by a unitary on the full algebra and its commutant. They transport the modular data, reconstruct the corresponding crossed products, and compute how the spectral weighting in the type-II trace changes. In the GJW application they obtain an entropy expansion containing

$$5\text{additional }O(h)\text{ terms}\text{and}15O(h^2)\text{ terms},$$

through $O(1/N^2)$. The original GJW first-law term is among the linear structures but is not counted among the five additional terms.

This achievement has a clear boundary:

• the deformation of the full system is unitary;
• the regional change is controlled perturbatively;
• the detailed application is a bilocal double-trace deformation;
• evaluating every term still requires a concrete boundary Hamiltonian/correlator input;
• the interpretation of the total type-II entropy as separate area and matter contributions is regime- and normalization-sensitive.

The reusable tool is not “a Connes-cocycle expansion.” It is the combination of unitary modular covariance, a changed trace weight/Jacobian, and a BCH expansion of the modular charge.

2. Outlook questions stated or directly motivated by AAJ §5

2.1 Beyond global unitary transformations

AAJ explicitly point to quantum channels as a natural extension. A global unitary that mixes an algebra with its commutant need not descend to unitary dynamics on either subsystem. In a type-I regulator the reduced map is a completely positive trace-preserving channel; the algebraic question is how to formulate the corresponding perturbative crossed product and entropy weighting without relying on a tensor-product partial trace.

Concrete deliverable. Start with a finite-dimensional Stinespring dilation, derive the subsystem channel, and identify which steps in AAJ’s covariance argument fail when the effective map has more than one Kraus operator. Then state what operator-algebraic replacement would be needed.

2.2 Regions that move, fluctuate, or change topology

AAJ emphasize a basic tension: AQFT normally assigns an algebra to a fixed region, while in gravity the region and its QES can move under a perturbation. Controlled unitary deformation handles a small change, but a topology-changing or nongeometric regime is not covered.

The difficult question is therefore not simply whether ${\mathcal{A}}_R^{\prime }={\mathcal{A}}_L$ continues to hold. Tomita commutant relations are representation-theoretic and can survive many physically different geometries. One needs a rule that assigns algebras to quantum regions and compares those assignments across states.

Concrete deliverable. Define a family $h\mapsto ({\mathcal{A}}_h,{\omega }_h)$ and specify embeddings or channels between different $h$. Without such comparison maps, the phrase “the algebra changed type/topology” has no precise content.

2.3 Heavy operators and loss of semiclassical geometry

AAJ note that sufficiently heavy boundary excitations may leave the regime in which a smooth bulk region is available. The crossed product still has an abstract algebraic meaning, but the geometric interpretation of its modular charge may fail.

This gives a useful research boundary: determine which statements require only a standard pair and which require a semiclassical bulk. Semifiniteness of the continuous core belongs to the first column; “this term is a QES area shift” belongs to the second.

2.4 Evaporation as mixing of algebra and commutant

AAJ use the traversable wormhole as a controlled analog of information transfer between a black hole and radiation. The global process may be unitary while the effective one-sided description is nonunitary and nonlocal. Extending this picture to a genuinely evaporating one-sided black hole requires a state- and time-dependent algebraic assignment, not merely reusing the eternal-TFD formulas.

3. Course-generated directions

The directions below are motivated by AAJ but are not presented as claims from their §5.

3.1 Higher-order organization

AAJ give an all-orders formal expression for the perturbed weight and work out the entropy through $h^2$. A natural project is to develop a verified symbolic organization of $h^3$ terms. The goal is not to guess a number from “cocycle × logarithm × clock sectors,” but to expand AAJ’s own weight/Jacobian formula, impose BCH order, and test the result against finite matrices.

Status: course-generated computational project.

3.2 Other interactions

Conserved-current or stress-tensor bilocals bring Ward identities and contact terms absent from a generic scalar double trace. One may ask which of AAJ’s displayed structures simplify, which new renormalization data enter, and whether the unitary deformation still has a controlled regional interpretation.

Status: course-generated extension; the number of terms cannot be predicted without carrying out the expansion and fixing grouping conventions.

3.3 Bulk matching term by term

Terms containing ${\beta }_1$ or ${\beta }_2$ invite an area interpretation, while commutator terms can contain matter and quantum-gravity information. A rigorous term-by-term map to a Shapiro time advance or a changed QES requires a bulk calculation in the same state, normalization, and perturbative scheme.

Status: partially motivated by AAJ’s interpretation, but the proposed detailed matching is open.

3.4 Embezzlement and the flow of weights

The strongest proven bridge between embezzlement and von Neumann-algebra type comes from van Luijk–Stottmeister–Werner–Wilming, not from AAJ. They define an operational worst-case error. A state is embezzling when the infimum error vanishes: arbitrary targets can be produced with arbitrarily small disturbance/error. In this sense type-III${}_1$ factors are universal embezzlers, and every normal state is embezzling.

“Error infimum zero” must not be rewritten as “one finite protocol has literally zero error.” It means that for every tolerance $ϵ>0$ there is an admissible protocol with error below $ϵ$.

The paper relates embezzling states to invariant probability measures on the flow of weights. That suggests a genuine question for the course: how does passing from a type-III${}_1$ algebra to its type-II${}_{\infty }$ crossed product change the relevant operational invariants? This is much sharper than saying that the same Connes cocycle “implements both” AAJ perturbation and embezzlement. AAJ do not compute a Connes-cocycle protocol, and the embezzlement theorem is organized by the flow of weights.

Status: course-generated bridge to the group’s research program.

4. Finite-dimensional embezzlement: what can be checked

4.1 The van Dam–Hayden family

The standard catalyst has Schmidt coefficients

$$|{\mu }_n⟩=\frac{1}{\sqrt{H_n}}\sum _{j=1}^n\frac{1}{\sqrt{j}}|j{⟩}_A|j{⟩}_B,H_n=\sum _{j=1}^n\frac{1}{j}.$$

Its Schmidt probabilities $p_j=1/(j{H}_n)$ are approximately scale invariant over a long range. Van Dam and Hayden prove that for any fixed finite target state, local unitaries can produce the target while returning the catalyst with fidelity tending to one as $n\to \infty$; the error decreases only logarithmically in the catalyst size.

The useful hand calculation is not an alleged exact fidelity formula. It is the edge-mass estimate. Rescaling the index by a fixed target Schmidt rank $m$ changes $1/j$ approximately by the compensating factor $m$ away from the first and last few indices. The unmatched edge weight is bounded by a fixed harmonic sum divided by $H_n$:

$$\text{edge weight}\lesssim \frac{H_m}{H_n}\underset{n\to \infty }{\to }0.$$

This explains why a harmonic spectrum works. The precise fidelity/error bound should be cited to van Dam–Hayden rather than reconstructed from an informal pairing of Schmidt coefficients.

4.2 Why a flat catalyst fails

Let

$$|{\nu }_n⟩=\frac{1}{\sqrt{n}}\sum _{j=1}^n|jj⟩.$$

Its Schmidt rank is $n$. Adding a product ancilla does not change that rank, while the desired state $|{\nu }_n⟩\otimes |{\Phi }^{+}⟩$ has Schmidt rank $2n$ with uniform coefficients. Local unitaries preserve the Schmidt spectrum. Even after embedding in a larger space, the largest possible squared overlap of a rank-$n$ state with that rank-$2n$ uniform target is at most

$${\left[\sum _{j=1}^n\left(\frac{1}{\sqrt{n}}\frac{1}{\sqrt{2n}}\right)\right]}^2={\left(\frac{1}{\sqrt{2}}\right)}^2=\frac{1}{2}.$$

after optimal alignment of the nonzero Schmidt directions. Increasing $n$ does not improve the bound. Large rank alone is therefore insufficient; the slowly varying, scale-free tail is essential.

4.3 What type III${}_1$ changes

The operator-algebraic theorem should be stated operationally:

For a type-III${}_1$ factor, every normal state is an embezzling state in the sense that the optimal worst-case error is zero.

The theorem is not proved by calling the modular spectrum a literal infinite Schmidt spectrum. That analogy is suggestive but can mislead because a type-III factor admits no underlying tensor-product density matrix of the kind used above. The proof uses the flow of weights and von Neumann-algebraic invariants.

5. A positioning matrix for final projects

Direction	Exact starting point	Missing ingredient	Defensible deliverable
AAJ $h^3$	v2 all-orders weight; BCH	controlled symbolic bookkeeping	verified finite-matrix expansion
Quantum-channel extension	AAJ §5 suggestion	crossed-product analog of reduced channel	finite Stinespring model + formal gap
Moving QES/algebra	unitary covariance	comparison maps between region algebras	precise family of inclusions/channels
Bulk term matching	Eqs. (77)–(78)	same-scheme gravity calculation	one term matched with assumptions
Holographic Bell	AQFT existence theorems	explicit boundary observables and theorem hypotheses	reproducible lower bound, no bridge iff
Crossed-product embezzlement	arXiv:2401.07299 invariants	type-II operational analysis	theorem-led literature memo or toy model

This table forces a project to name both its controlled input and its missing step. That is what “positioning” should do.

6. Student presentations

Each student gives a ten-minute talk with four slides:

1. Question. One sentence, narrow enough to answer.
2. Controlled input. The theorem, equation, or finite model actually available.
3. Proposed calculation. A deliverable that can be checked.
4. Scope ledger. Exact/proved, numerical, heuristic, and open claims in separate rows.

Recommended topic forms:

• finite-regulator BCH/entropy perturbation following AAJ;
• explicit Bell lower bound for a specified observable family;
• operational embezzlement under passage to a crossed product;
• critical, equation-by-equation exposition of CPW, AAJ, or CLPW.

“Free-field cocycle perturbation reproducing AAJ” is no longer offered without qualification. A good version is “finite-regulator comparison between an Araki cocycle and AAJ’s unitary BCH deformation.”

7. What to take away

• AAJ v2 ends with §5, not §6.
• AAJ’s method is unitary crossed-product perturbation, not a Connes-cocycle expansion.
• Their new-term count is five linear plus fifteen quadratic.
• AAJ’s stated outlook includes quantum channels and the problem of algebras attached to moving, fluctuating, or topology-changing regions.
• Higher-order diagrams, other bilocals, detailed bulk matching, and the embezzlement connection are useful course-generated directions unless a more specific source is supplied.
• Type-III${}_1$ universal embezzlement is an arbitrarily-accurate operational theorem. It does not mean a single finite protocol has exactly zero error.
• The correct embezzlement source is van Luijk–Stottmeister–Werner–Wilming, arXiv:2401.07299; the finite catalyst is due to van Dam–Hayden, quant-ph/0201041.

8. Problem set

Core problems

1. Source/outlook split. Make two lists from this week: statements directly present in AAJ §5 and course-generated research proposals. Verify each source statement by paragraph or footnote.

2. Count audit. Explain in one paragraph why “twenty quadratic corrections” is false. Reproduce the $2+2+2$ linear and $2+6+7$ quadratic groupings from Eqs. (77)–(78).

3. Flat catalyst. Prove the $1/2$ squared-overlap bound in §4.2 from Schmidt coefficients. State which assumptions about local operations are used.

4. Harmonic edge estimate. For fixed $m$, show $H_m/H_n\to 0$. Explain why this is only intuition for the van Dam–Hayden construction and cite the source for the actual protocol error bound.

Starred problems

5. Channel extension.* Work out a two-qubit global unitary whose reduced dynamics has two Kraus operators. Identify the precise step in AAJ’s conjugation argument that cannot be copied for the reduced channel.

6. Embezzlement definition.* Read the definitions in arXiv:2401.07299. Distinguish “error zero,” “infimum error zero,” and “exact finite operation.” Which one appears in the universal-embezzler theorem?

7. Nonperturbative diagnostic.* Propose comparison maps for a family of region algebras ${\mathcal{A}}_h$. Explain why the bare relation $J{\mathcal{A}}_{h}J={\mathcal{A}}_h^{\prime }$ cannot by itself decide whether a wormhole has changed topology.

Project problem

8. Final-topic presentation. Deliver the four-slide talk of §6 and revise the written project scope in response to questions. Every claimed source result must carry an equation, theorem, or section pointer.

Wiki connections. Bell inequalities in QFT (research area) · Bell–CHSH in holographic settings (open question)

---

Notes prepared for the algebraic-QFT course, Semester II Block 4. Last revised 2026-08-24.