A new Cartier duality for gerbes of vector bundles shows that solid quasi-coherent sheaves on the Hodge-Tate stack of a smooth rigid variety are exactly the weight-1 sheaves on the Simpson gerbe.

1. self definitional [Definition 3.4.2; Theorem 3.4.5 (statement and proof)]

   "Tensoring the map η_HT : X → B^2T†_X(1) with the analytic cotangent bundle T^{*,an}_X(−1) produces a map T^{*,an}_X(−1) → B^2T†_X(1) ×_X T^{*,an}_X(−1), we compose with the natural pairing B^2T†_X(1) ×_X T^{*,an}_X(−1) → B^2G†_{a,X} → B^2G_{m,X} ... Definition 3.4.2. The Simpson gerbe S_X of X is the pullback square ... Theorem 3.4.5. There is a natural pairing of Z-modules X_HT^{ext} ⊗_Z S_X → BG_m ... [X_HT^{ext}]_* ≅ [S_X]_! ... The first statement about Cartier duality is a formal consequence of Theorem 3.2.35 and Theorem 3.3.3."

   S_X is constructed from the same HT class η_HT and the same suspended exponential pairing that the duality (3.10) then pairs with X_HT^{ext}. In the kernel-category formalism (Prop. 2.4.3) Cartier duality is the identity on underlying objects and merely swaps the *- and !-structures. Theorem 3.3.3 lifts the rank-one exponential duality by defining N_ψ as the pullback of that pairing, and Def. 3.4.2 is exactly that N_ψ after pushout to G_m. Thus (3.10) holds for the paper's own S_X by construction once the base vector-bundle duality (Thm. 3.2.35) is assumed; it does not independently identify a pre-existing Simpson gerbe. The bridge to Bhatt–Zhang is only asserted in Rem. 3.4.3 ('a bookkeeping of the construction shows'), so the advertised application is not established by this derivation.

2. self citation load bearing [§1.1, footnote 8, Def. 3.4.1; reference [ALBRCS]]

   "In joint work in progress with Anschütz, Le Bras and Scholze on the analytic prismatization [ALBRCS], we introduce the analytic Hodge-Tate stack X_HT of a smooth rigid space X ... Footnote 8: The formalism of [ABB+25] only uses induced analytic ring structures so a priori it can be directly applied only to partially proper rigid spaces or to smooth dagger spaces. However, since any smooth rigid space in the sense of Huber is (locally) an open subspace as analytic stacks of a smooth dagger space, the following discussion on the Hodge-Tate stack also applies to them."

   The theorem's input D(X_HT) is the analytic Hodge-Tate stack of [ALBRCS], an in-preparation work by the same authors; the paper does not independently construct this object in the required generality. The ad-hoc Def. 3.4.1 uses η_HT, and footnote 8 asserts the extension from partially proper/dagger spaces to general smooth rigid spaces by locality, but that extension is part of the unpublished [ALBRCS] theory. The central premise of the application is therefore supported by a self-citation that is not machine-checked or externally reproduced and whose assumptions include the target stack. This makes the headline generality load-bearing on the authors' own unpublished claims.

3. other [Remark 3.4.3; abstract]

   "A bookkeeping of the construction shows that this gerbe is precisely that of Theorem 3.4.2 after taking the pushout along the natural morphism G^{an}_m → G_m from the analytic to the algebraic multiplicative group."

   This sentence is the entire bridge from the abstract's 'Bhatt–Zhang’s Simpson gerbe' to the object S_X for which Theorem 3.4.5 is proved. It is a sketch, treats only partially proper spaces, and changes the band from analytic G^{an}_m to algebraic G_m without proving that the weight-1 categories are canonically identified under the band change; the bibliography contains no Bhatt–Zhang reference. Consequently the headline equivalence is proved for the paper's renamed/constructed gerbe, while the independently constructed Bhatt–Zhang gerbe is not shown to satisfy (3.10)–(3.11). This is a missing external comparison rather than an internal contradiction, but it is load-bearing for the abstract's claim.

The internal derivation chain in Sections 2–3.3 is largely self-contained: Theorem 2.5.13, the rank-one exponential computations (Theorems 3.2.6, 3.2.11, 3.2.20, 3.2.26, 3.2.31), and the descent/devisage to universal vector bundles (Theorem 3.2.35) are parameter-free and do not presuppose the HT/Simpson duality. Theorem 3.3.3 then formally lifts vector-bundle Cartier duality to gerbes. The circularity risk is concentrated in Section 3.4: the object called 'Simpson gerbe' in the main theorem is constructed in Definition 3.4.2 from the same η_HT and the same exp-pairing that the duality exchanges, so equation (3.10) is a formal consequence for that ad-hoc object rather than an independent prediction about Bhatt–Zhang's gerbe. The only bridge to the abstract's Bhatt–Zhang claim is a sketched comparison in Remark 3.4.3 involving a band change and a locality assertion in footnote 8, and the Hodge–Tate stack in full generality is sourced to the in-preparation [ALBRCS]. Thus the headline application is partly built into definitions and partly supported by unpublished self-citations. Score 4: the rank-one exponential and descent content is genuine and non-circular, so this is not a fully circular paper, but the advertised external identification is not established by the derivation.