On AI's "semistable torsion classes and canonical decompositions"

Jiarui Fei

REVIEW 5 major objections 4 minor 1 references

On AI's "semistable torsion classes and canonical decompositions"

T0 review · 5 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This note claims that the main theorems of a 2024 paper on semistable torsion classes and canonical decompositions are direct corollaries of an earlier theorem relating tropical $F$-polynomials to hom-spaces.

desk verdict A sharp but under-supported priority note: the reduction to [Ft] is plausible yet circular in places, so the sweeping claim about [AI] doesn't land as written. read the letter →

arxiv 2412.08904 v2 pith:33NF3OMV submitted 2024-12-12 math.RT math.RA

Jiarui Fei This is my paper · ORCID

classification math.RTmath.RA MSC 16G2018E40

keywords semistabletorsionclassescanonicaldecompositiontropicalF-polynomialsgeneralpresentationsweightvectorspairsE-tamealgebrasquiverrepresentations

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This short note claims that the main theorems of a 2024 paper on semistable torsion classes and canonical decompositions are not independent results but direct corollaries of a theorem the author published two years earlier. The earlier theorem says that, for any representation and any weight vector, tropical $F$-polynomials evaluated along a ray eventually coincide with ordinary hom-space dimensions. Granting that, the 2024 paper's description of torsion classes, its canonical-decomposition theorem, and its criterion for $E$-tame equivalence all reduce to two-line arguments. The note also asserts that little mathematical content in the 2024 paper goes beyond that earlier theorem, and that some historical attributions in its introduction are inaccurate.

What carries the argument

The machine is the asymptotic identity [Ft, Theorem 3.6]: for any representation $M$ and any weight vector $\delta$, there is an $n$ such that $f_M(n\delta)=\hom(n\delta,M)$ and $\check f_M(-n\delta)=e(n\delta,M)$, and the same holds for every multiple $kn$. Here $f_M(\delta)=\max_{L\hookrightarrow M}(\dim L)\cdot\delta$ is the tropical $F$-polynomial, while $\hom(n\delta,M)$ is the dimension of the kernel of the map induced by a general presentation of weight $n\delta$. The identity converts a maximum over subrepresentations into a hom-vanishing condition, which is exactly the kind of condition that defines the semistable torsion classes. The note also relies on the unproved identification of the resulting classes with [BKT]'s weight-vector torsion pairs.

What would settle it

Take a small quiver algebra (for instance, the 3-Kronecker quiver) and a weight vector $\delta$ with one positive and one negative coordinate, compute the class $\{N \mid \hom(n\delta,N)=0 \text{ for some } n\}$ and the class $\{N \mid \delta(\dim L)\le 0 \text{ for all subrepresentations } L\}$, and check whether they coincide. Any module that lies in one class but not the other would break the central identification and with it every proof in the note.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that [AI]'s main results are consequences of [Ft, Theorem 3.6]. The bridge is the assertion, made after Lemma 2.3, that the torsion-free class $\mathcal F(\delta)=\{N \mid \hom(n\delta,N)=0 \text{ for some } n\}$ and the torsion class $\check{\mathcal T}(\delta)=\{L \mid e(n\delta,L)=0 \text{ for some } n\}$ are exactly the weight-vector torsion pairs $(\mathcal T_\delta, \mathcal F^\delta)$ and $(\mathcal T^\delta, \mathcal F_\delta)$ introduced in [BKT]. Once that identification and the asymptotic identity of Theorem 2.2 are in place, each theorem in [AI]—the containment $[\theta]_{TF}\supseteq \mathrm{cone}^\circ(\mathrm{Ind}\,\theta)$, the descriptions of $\mathcal T_\theta$ and $\mathcal F_\theta$ as intersections or unions over $\ell$, and the $E$-tame equivalence criterion—is proved in a few lines. The note states this reduction explicitly: modulo [Ft], there is not much essential mathematical content left in [AI].

Load-bearing premise

Everything rests on the unproved claim that the torsion classes defined by vanishing hom-spaces after scaling a weight are the same as the torsion classes defined by sign conditions on every subrepresentation; if that identification fails, the two-line proofs of the 2024 theorems collapse.

Editorial extensions

If this is right

The semistable torsion-class theorems of [AI] are corollaries of [Ft, Theorem 3.6], so any future proof or application can cite the one identity instead of the later paper's machinery.
If the torsion-pair identification is correct, the classes $\mathcal T_\delta$ and $\mathcal F^\delta$ are determined by the asymptotic vanishing of $\hom(n\delta,-)$ and $e(n\delta,-)$, giving a uniform recipe for computing them.
The $E$-tame equivalence criterion—$\eta$ and $\theta$ are TF equivalent exactly when their indecomposable canonical summands agree—follows from the same identity.
For hereditary algebras the equality $[\theta]_{TF}=\mathrm{cone}^\circ(\mathrm{Ind}\,\theta)$ is traced to earlier work, not to [AI].

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the reduction is sound, the theory of semistable torsion classes could be rebuilt on tropical $F$-polynomials, making canonical decompositions of weight vectors a by-product of computing $\hom(n\delta,-)$ asymptotics.
The same asymptotic identity might extend to other stability frameworks wherever a hom-vanishing criterion determines the boundary of a torsion class.
A direct check of the torsion-pair identification on small quiver examples would settle whether the bridge is genuine or needs an additional argument; the note does not supply that check.
The note's priority remarks imply that the historical record in the 2024 paper's introduction should be revised if the two-line proofs are accepted.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Dear colleague,

You should know this: Fei's note has a real point about priority, but the advertised "two-line proofs" don't stand up to scrutiny. The reduction is not self-contained, and the stress-test's circularity objection is accurate.

What's genuinely useful: the note draws a clear link between the torsion pairs in Asai-Iyama and the hom-vanishing classes defined in Fei's earlier work. That connection is worth having explicitly, even if it's only asserted. The observation that Theorem 3.2 follows quickly from the definitions is fine. And the note is honest about not checking [AI] beyond the introduction, which actually narrows its own claim.

The soft spots are exactly where the reader puts them. Theorem 3.3 is central and is simply stated as a "simple consequence" with no proof; it's then used inside Theorem 3.4. More importantly, the proof of Theorem 3.4 invokes [AI, Lemma 2.10] directly. Even if that lemma is "obvious from [Fc]" as Fei says, a reduction that imports a lemma from the paper it claims to eliminate cannot support the statement that "not much essential mathematical content is left in [AI]." Also, the identification of the (T(δ), F(δ)) pairs with the [BKT] torsion pairs is just asserted; if that bridge has any subtlety, the rest of the note depends on it. These aren't manufactured flaws—they're load-bearing gaps.

That said, I don't think this note is worthless. It's a legitimate challenge to the historical narrative, and the underlying mathematics is presumably sound. But as a piece of research it doesn't deliver on its headline. It reads like a comment that belongs in a correspondence or a footnote, not a standalone paper.

For peer review: I'd send it to a competent referee who knows the torsion theory literature. The priority question matters, and a referee could quickly determine whether the gaps are fillable and whether the claims about [AI] are accurate. I wouldn't desk-reject it, but I'd expect heavy revision or conversion to a comment format.

So: worth a referee's time, but not something you'd cite for its math.

Referee Report

5 major / 4 minor

Summary. The note claims that the main theorems of Asai-Iyama's paper [AI] are direct consequences of Fei's earlier theorem [Ft, Theorem 3.6] and that modulo that theorem 'there is not much essential mathematical content left in [AI].' It reviews definitions from [DF] and [BKT], recalls the tropical F-polynomial theorem from [Ft] and torsion-free/torsion classes from [Fc], then sketches proofs of [AI, Theorems 1.1, 1.3, 1.4, 1.5], appending remarks on priority and on the introduction of [AI].

Significance. If the reduction were fully proved, the note would establish that parts of [AI] are not independent of [Ft] and would provide a shorter route to several results. The paper also points to earlier work ([HKM], [DW1]) on the torsion theory. However, the note currently does not supply complete reductions for the key theorems; in one place it invokes a lemma from the very paper being reduced. Therefore the significance, as a proof of redundancy, is not currently realized. The strengths are the precise references and the observation that Theorem 2.2 may imply the torsion-pair identifications, but these need full derivations.

major comments (5)

[§3, Theorem 3.3] Theorem 3.3 is stated without a proof; the sentence that it is 'a simple consequence of Theorem 2.2 (see the comment after Lemma 2.3)' is not a derivation. Since Theorem 3.3 is used in the proof of Theorem 3.4, the reduction of [AI, Theorem 1.4] is incomplete. Please provide the promised two-line proof or explicitly state that the result is taken from [Ft].
[§3, Theorem 3.4] The proof of Theorem 3.4 says it is 'a straightforward consequence of Theorem 3.3 and [AI, Lemma 2.10]'. Invoking a lemma from [AI], the target paper, makes the argument circular with respect to the note's meta-claim that modulo [Ft] there is little content left in [AI]. The phrase 'another easy observation made from the torsion theory of [Fc, Section 3.2]' does not supply the proof. The lemma must be stated and proved independently, or the claim about [AI] must be weakened.
[§2, after Lemma 2.3] The identification of the pairs (T(δ), F(δ)) and (ˇT(δ), ˇF(δ)) from (2.1)-(2.2) with the BKT torsion pairs (T_δ, F^δ) and (T^δ, F_δ) is asserted without proof. This identification is load-bearing for Theorems 3.2 and 3.3. In particular, the 'for some n' quantifier in (2.1) needs to be reconciled with the n-dependent statements in Theorem 2.2. A precise proof from Theorem 2.2 is required.
[§4, Remark 4] The note states that the author does not check statements in [AI] beyond the introduction, while the abstract claims that the main results of [AI] are proved. This is a scope mismatch. The verification must either be extended to all main results or the abstract should be narrowed to the statements actually proved.
[§3, Theorem 3.5] For the hereditary case, the proof says it is 'essentially an easy consequence of results in [DW]' without further detail. Since Theorem 3.5 is among the main results listed in the abstract, this is another missing step. Provide the argument or state explicitly that the hereditary case is quoted from [DW] and not part of the reduction.

minor comments (4)

[Title and throughout] There are several typos ('SEMIST ABLE' in the title, 'fro m' on page 1, 's left' in the abstract, 'central rule' for 'central role'). Please proofread.
[§3, Theorem 3.3] The display in Theorem 3.3 is garbled (e.g., 'T h /ℓθ' and 'W h /ℓθ'); the notation should be typeset properly so the reader can see which torsion classes are being intersected and unioned.
[§3, Definition 3.1] The notation K0(proj-A)_R in Definition 3.1 is not defined; specify that it is the real Grothendieck group of projective modules.
[§4, personal remarks] The personal remarks in Section 4 about conference talks (Sanya 2019, Morningside 2021) are outside the mathematical scope of the note and could be removed or moved to a footnote; they do not affect the correctness of the claimed reductions.

Circularity Check

2 steps flagged · score 6.0 of 10

The promised reduction is not self-contained: Theorem 3.4 invokes [AI, Lemma 2.10] from the very paper being reduced, and Theorem 3.3 is asserted without proof; the claim that [AI] has no essential content beyond [Ft] is only partially established.

other [Section 3, proof of Theorem 3.4]
"This is a straightforward consequence of Theorem 3.3 and [AI, Lemma 2.10], which is another easy observation made from the torsion theory of [Fc, Section 3.2]."

The note's stated aim is to give two-line proofs for the main results of [AI] from [Ft, Theorem 2.2], leaving 'not much essential mathematical content' in [AI]. But the proof of [AI, Theorem 1.4] explicitly imports [AI, Lemma 2.10] from the target paper itself, without deriving it in this note. Unless that lemma is independently proved from [Ft] or [Fc] or is shown to be a trivial reformulation, this step reduces the claimed theorem to an input taken from the very paper whose content is being dismissed. The parenthetical assertion that the lemma is 'another easy observation' does not supply the derivation, so the meta-claim that [AI] has no essential content beyond [Ft] is not upheld for this theorem.
other [Section 3, Theorem 3.3]
"As mentioned in [AI], the following Theorem 3.3 was already proved in [Ft]. In fact, it is a simple consequence of Theorem 2.2 (see the comment after Lemma 2.3)."

This is a missing derivation rather than a fully circular step, but it is load-bearing: the note says Theorem 3.3 'plays a central rule in [AI]' and then uses it in the proof of Theorem 3.4. No two-line proof is actually exhibited here; the assertion that it is a simple consequence of Theorem 2.2 and the comment after Lemma 2.3 is not developed. Consequently the claim that all main results of [AI] follow immediately from [Ft] is not completely demonstrated, and the later proof of Theorem 3.4 depends on this unproved bridge.

full rationale

The main mathematical engine, [Ft, Theorem 2.2], is an independent published result from the author's earlier work, not derived from [AI], so the paper is not globally circular. The circularity is in the meta-reduction: the note promises that modulo [Ft] there is little essential content in [AI], but the proof of Theorem 3.4 explicitly cites [AI, Lemma 2.10] from the target paper without proving it in the note. This means one of the 'two-line proofs' is not actually a reduction to [Ft] alone; it is a reduction to an unverified lemma from the paper being reduced. In addition, Theorem 3.3, which the note calls central and uses later, is asserted without proof, and the identification of the torsion pairs from (2.1)-(2.2) with the [BKT] torsion pairs is stated without proof after Lemma 2.3. These are not cases where a fitted parameter is renamed as a prediction, nor is the conclusion forced by a self-citation chain; rather, the central claim of a content-free reduction is partly circular because it relies on an input from the target paper. Score 6 reflects that partial but central circularity: one of the claimed reductions reduces to a lemma of the very paper whose independent content is being denied.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The note relies on several theorems from prior literature, most importantly Fei's own [Ft, Theorem 3.6] and [Fc, Lemma 3.8], as well as results from [DF], [AI], and [DW]. No free parameters or new entities are introduced. The heavy reliance on the author's own prior work, together with use of a lemma from [AI] itself, is recorded in the circularity score.

assumptions (5)

domain assumption Theorem 2.2 of [Ft]: for any representation M and weight δ, there is n such that f_M(nδ) = hom(nδ, M), and similarly for the dual tropical F-polynomial.
This is the main engine of the note; the author uses it to identify torsion pairs and derive AI's results. It is cited, not proven here.
domain assumption Lemma 2.3 of [Fc]: F(δ) is a torsion-free class and T(δ) is a torsion class.
Used to set up the torsion theory that is then identified with AI's.
domain assumption Theorem 1.3 of [DF]: canonical decomposition criterion in terms of e(δ_i, δ_j)=0.
Background on canonical decompositions used in the definitions of Ind θ and cone.
domain assumption Lemma 2.10 of [AI].
Used without proof in the proof of Theorem 3.4; this is a result from the very paper the note aims to reduce, which weakens the reduction.
domain assumption Results of [DW] on canonical decompositions of quiver representations.
Used for the hereditary case in Theorem 3.5.

how reviews work

0 comments

Cite this review

Pith. "Pith review of On AI's "semistable torsion classes and canonical decompositions"." pith.science (2026). https://pith.science/paper/33NF3OMV

@misc{pith2026241208904,
  author       = {Pith},
  title        = {Pith review of: On AI's "semistable torsion classes and canonical decompositions"},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/33NF3OMV}},
  note         = {Machine review of arXiv:2412.08904}
}

abstract

In this short note, we give two-line proofs for main results in "Semistable torsion classes and canonical decompositions" by Asai-Iyama from a main result in "Tropical $F$-polynomials and general presentations", which appeared on the math arXiv 2 years earlier.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

1 extracted references · 1 linked inside Pith

[1]

Semistable torsion classes and canonica l decompositions in Grothendieck groups

[AI] S. Asai and O. Iyama. “Semistable torsion classes and canonica l decompositions in Grothendieck groups”. In: Proc. Lond. Math. Soc. (3) 129.5 (2024). arXiv: 2112.14908. [BKT] Pierre Baumann, Joel Kamnitzer, and Peter Tingley. “Aﬃne Mir kovi´ c-Vilonen polytopes”. In: Publ. Math. Inst. Hautes ´Etudes Sci. 120 (2014), pp. 113–205. arXiv: 1110.3661. [DF...

arXiv 2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.

[1] [1]

Semistable torsion classes and canonica l decompositions in Grothendieck groups

[AI] S. Asai and O. Iyama. “Semistable torsion classes and canonica l decompositions in Grothendieck groups”. In: Proc. Lond. Math. Soc. (3) 129.5 (2024). arXiv: 2112.14908. [BKT] Pierre Baumann, Joel Kamnitzer, and Peter Tingley. “Aﬃne Mir kovi´ c-Vilonen polytopes”. In: Publ. Math. Inst. Hautes ´Etudes Sci. 120 (2014), pp. 113–205. arXiv: 1110.3661. [DF...

arXiv 2024