Tail generation for symmetric free-product states: a general deduction
The tail of a sequence of operator algebras consists of the observables that remain after every finite initial block is discarded. For a state unchanged by finite relabelings of the copies, this note shows that the recursively defined Dykema–Köstler–Williams tail C*-algebra generates the entire tail von Neumann algebra.
The conclusion holds in the Hilbert-space representation built from the original state. It also gives an exact bound: every tail operator is a weak-operator limit of conditional expectations of represented free-product elements, with both the inputs and outputs bounded by the norm of that target.
Open the original PDF (4 pages) · Download PDF · Original files · BibTeX · DOI: 10.5281/zenodo.23567720
A general deduction from earlier methods
The contribution is a short general deduction for the explicit Dykema–Köstler–Williams recursive C*-algebra in its original representation. It answers their generation Question 5.3.5 and Question 2 in the 2015 and 2016 Oberwolfach reports, in the arbitrary-unital-algebra setting below.
Dykema, Köstler and Williams (DKW) already supplied finite-word coefficient stabilization, its bounded-vector extension and the canonical shift expectation. Liu's bilateral future-expectation argument supplies a closely related past/future mechanism. The note explains the bridge to arbitrary unital algebras and the original representation. It does not attribute this full statement verbatim to Liu.
The exact norm bound comes from Kaplansky density. The 2×2 matrix device used to retain the same bound for nonselfadjoint targets already appears in Kaplansky's proof, credited there to P. R. Halmos. DKW Theorem 7.3 already proves generation for quantum symmetric states, including nonfaithful states.
The note claims no new coefficient principle, new Kaplansky mechanism, historical firstness, exhaustive priority clearance or present-openness guarantee. The dated question statements identify the problem addressed.
The theorem in the original representation
Let A be any unital C*-algebra, let 𝔄=*j∈ℕA be its countable universal unital free product, and let ψ be any state invariant under every finite permutation of the copies. Here ℕ={1,2,…}. In the original GNS representation (π,H,ξ), write πj for the represented j-th copy and set
M=π(𝔄)″,
CN=C*(πj(A):j>N),
T=⋂N≥1CN″,
Q=C*(T∪π(𝔄)).
All commutants and operator closures are taken in B(H), on this same Hilbert space. The common unit is included. Let F:Q→T be DKW's canonical contractive expectation, given for each fixed x∈Q by the weak-operator limit of its positive shifts.
Define D to be the least unital C*-subalgebra of T closed under all expressions
F(πi₁(a₁)d₁πi₂(a₂)⋯dm−1πiₘ(aₘ)),
for every m≥1, all a₁,…,aₘ∈A, all i₁,…,iₘ∈ℕ and all inserted d₁,…,dm−1∈D. Indices may repeat. The theorem is
D″=T.
More precisely, for every t∈T there is a net of represented inputs xλ∈π(𝔄) such that
‖xλ‖≤‖t‖, ‖F(xλ)‖≤‖t‖,
F(xλ)→t in the weak-operator topology (WOT).
This means ⟨F(xλ)η,ζ⟩→⟨tη,ζ⟩ for every pair η,ζ∈H. For t=0 the constant zero net suffices. Nets retain the full statement without a separability assumption.
No faithful state or representation, separability, trace, factoriality, quantum symmetry or normal extension of F to M is assumed. The separate inclusion question D⊆π(𝔄) is not answered. Generation as a von Neumann algebra does not assert that this inclusion holds or that D=T as C*-algebras.
Why bounded future inputs suffice
Let BN be the C*-algebra of the first N copies, HN the closure of BNξ, and PN the orthogonal projection onto HN. Every finite word uses finitely many copies, so these subspaces increase densely to H and PN→I strongly. This needs no countable dense subset of A or H.
For x∈CN, finite-permutation symmetry and DKW's fixed-input shift limit give the exact compression identity
PNF(x)PN=PNxPN.
A finite future support can be moved by a finite permutation while every first-N copy stays fixed. The construction permits overlapping shifted supports and transports every repeated index consistently. Norm approximation and contractivity extend the identity from finite words to all of CN.
For a tail target t and finitely many WOT test pairs, first choose N so that replacing the vectors by their PN projections makes a small error. Since t∈CN″, norm-ball Kaplansky density gives x∈CN⊆π(𝔄) of norm at most ‖t‖ approximating the compressed coefficients. The identity transfers those coefficients to F(x). The uniform bounds on x and F(x) control the remaining vector errors, giving the required net with the exact target norm.
Setting every inserted dj=1 in the definition of D places F of each ordinary copyword in D. Norm closure gives F(π(𝔄))⊆D⊆T, and the bounded WOT density then gives D″=T.
The proof uses the shift limit on each fixed input. It never passes a varying WOT-convergent input net through F. Its compression identity is not a global equality F(x)=x, and no commutation of PN with the target is needed.
Normality and the supporting bilateral route
A normal extension of F cannot be supplied as a shortcut in general. In the compact-plus-scalar example recorded by DKW and credited there to Liu, F(λI+K)=λI+K₀₀e₀₀. The finite-rank projections qm=∑j=0mejj increase strongly to I, while F(qm)=e₀₀≠F(I)=I.
The supporting note also derives the result through Liu's closely related bilateral method. Normal expectations on each fixed bilateral future algebra are combined with a normal quotient back to the original representation. A tail target has exact-norm lifts in successive future algebras; nested ultraweakly compact fibers provide a lift in their intersection. This proves the needed tail-image equality. The argument does not commute arbitrary images with intersections or assert unrestricted normal descent of the expectation.
Original manuscript abstract
For every unital C*-algebra and every state on its countable universal unital free product invariant under finite permutations, the recursive tail C*-algebra of Dykema–Köstler–Williams generates the tail von Neumann algebra in the original GNS representation. More precisely, the canonical expectation of the represented free-product algebra is weak-operator dense in the tail with the exact norm bound of each target. This is a short general deduction from the earlier coefficient-stabilization method and Kaplansky density, answering the explicit generation question. No faithfulness or separability is required. We explain the close relation to Liu's bilateral future-expectation argument and retain bounded priority and source-version qualifications.
Original files and citation
The two unchanged native files appear in their original Zenodo order. The archive includes the editable manuscript, detailed relation to prior work, source and code provenance, licenses and portable exact verification controls.
- Manuscript PDF — tail-generation.pdf · Zenodo original
- Source and verification archive — tail-generation-source-and-verification.zip · Zenodo original
Kriebel, Alec (22 October 2026). Tail generation for symmetric free-product states: a general deduction (Version 1.0.0) [Preprint]. Zenodo. 10.5281/zenodo.23567720. Download BibTeX · Published native record.
Both the manuscript and native record identify version 1.0.0 and the date 22 October 2026. The record's creation and modification timestamps fall on that date.
The archive's actual licenses grant CC BY 4.0 for the authored prose and MIT for the authored code. Referenced third-party works retain their own rights.
Main references
- K. J. Dykema, C. Köstler and J. D. Williams, Quantum symmetric states on free product C*-algebras, Transactions of the American Mathematical Society 369 (2017), 645–679. arXiv version, §§5.1–5.3 and Theorem 7.3; Definition 5.3.1 specifies D and Question 5.3.5 asks about generation.
- W. Liu, Extended de Finetti theorems for boolean independence and monotone independence, Transactions of the American Mathematical Society 370 (2018), 1959–2003. arXiv v1, §6.2, Lemma 6.8 and Proposition 6.10 supply the cited numbering and related bilateral mechanism.
- I. Kaplansky, A theorem on rings of operators, Pacific Journal of Mathematics 1 (1951), 227–232, Theorem 1.
- K. Dykema, joint with C. Köstler and J. D. Williams, Tail algebras of symmetric states and de Finetti theorems, Oberwolfach Report 28/2015, printed pages 1618–1619, Question 2. DOI · Report PDF.
- K. Dykema, Tail algebras, amalgamated free products and KMS states, Oberwolfach Report 40/2016, printed pages 2334–2335, Question 2. DOI · Report PDF.
Read the manuscript
Open PDF in a full window · Download PDF · Original PDF on Zenodo
Research status and contributions
AI tools were used extensively in research planning, proof development, computation, source investigation, writing and adversarial verification. Alec Kriebel is responsible for the claims and artifacts. This is an unrefereed preprint; internal AI reviews do not constitute independent human peer review or formal proof certification. The finite exact controls supplement the written proof and do not certify the theorem for arbitrary C*-algebras or establish historical priority.
Researchers interested in independent verification, substantial mathematical extensions, or developing a revised manuscript for journal submission are welcome to contribute. Authorship and acknowledgments depend on the substantive work actually contributed; endorsement alone does not confer authorship.