{"id":"2a2743dc-bf92-42a8-9584-8f6108114e67","arxiv_id":"2505.09120","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Excellent metrics are a subclass of good metrics for which the completion map S to S(S)^op is an almost involution, up to idempotent completion.","lead":"Neeman identifies a class of 'excellent metrics' on triangulated categories under which the construction S to S(S)^op is almost an involution. This is groundwork for a sequel proving new results about uniqueness of enhancements for triangulated categories.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The main theorem is conditionally applicable: the only established excellent metrics depend on statements imported from the unpublished preprint [6], leaving the non-vacuity of the class unverified in this paper.","rationale":"The reader's weakest assumption correctly identifies the dependency on the unpublished preprint [6] as the main source of conditionality. I read the paper's central theorem as a careful conditional: if a metric satisfies Definition 2.1, then the almost-involution conclusions follow. The proof of that conditional is long and appears internally coherent; the paper is explicit about the auxiliary constructions and gives detailed proofs for the intermediate lemmas. The genuinely load-bearing point is the non-vacuity of the class of excellent metrics. Example 2.3 is the only systematic construction of excellent metrics before Section 8, and its verification of Definition 2.1(iii) invokes [6, Lemma 8.5] for the existence of strong approximating sequences. Section 8 similarly imports results from [6] and [1]. This is a legitimate external dependency, not an internal flaw: the author does not hide it, and the theorem would be unaffected if the imported statements were supplied. A conditional verdict is therefore appropriate. My stress-test does not reveal any additional internal inconsistency that would force a stronger or weaker verdict, so I recommend keeping the reader's CONDITIONAL verdict unchanged.","tokens_in":75116,"tokens_out":6436,"duration_ms":63220,"concrete_test":"Provide a self-contained proof of the specific statements imported from [6] that are used in Example 2.3 and Section 8, at minimum [6, Proposition 2.14] and [6, Lemma 8.5]. The minimal check is whether every Cauchy sequence c_* in T^c admits a subsequence that is strong T^c-approximating for its homotopy colimit, with the bound d_i in T^{≤ -m-1} used in Example 2.3; if this fails for a weakly approximable coherent T, then Definition 2.1(iii) is not established for the motivating examples.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 0.9 is a conditional statement about excellent metrics, and the paper's main supply of excellent metrics is Example 2.3, whose verification of Definition 2.1(iii) relies essentially on [6, Remark 0.24 plus Lemma 2.8, Proposition 2.14, and Lemma 8.5]. The crucial step replaces the Cauchy sequence c_* with a 'strong T^c-approximating sequence' and uses [6, Lemma 8.5(iii)] to obtain d_i in T^{≤ -m-1}; these are not proved in the present work. If [6, Lemma 8.5] is false, or if not every Cauchy sequence in T^c admits such an approximating subsequence, then the motivating metrics on T^c and (T^b)^op are not known to be excellent, and Theorem 0.9 would have few or no known instances. This is not an internal inconsistency in the proof of Theorem 0.9 itself; it is an unresolved external dependency that controls whether the central claim has any substantive applications. The paper is transparent about this, but the conditionality remains load-bearing because the 'large class' of excellent metrics is otherwise unsupported.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a new subclass of good metrics on triangulated categories, called excellent metrics, and studies the involutivity of the construction that sends a triangulated category S with a metric {M_i} to the opposite of the category S(S) with the induced metric {N_i^op}. The main theorem (Theorem 0.9) states that excellence is preserved under this passage, and that the double passage is almost an involution: there is a fully faithful triangulated functor bY from S to the second double opposite whose essential image generates the target up to direct summands, with an isometry in the idempotent-complete case. The paper also develops very good metrics, proves recognition criteria for excellence in the presence of good extensions (Proposition 6.2), and constructs examples: for weakly approximable coherent triangulated categories the metrics on T^c and (T^b_c)^op are excellent, and in Section 8 it identifies T^sb = union_n <G>^[-n,n] with an excellent metric for weakly approximable T with a compact generator.","tokens_in":75375,"tokens_out":3328,"duration_ms":34227,"significance":"If the results stand, this is a substantial structural contribution to the theory of metrics on triangulated categories and to the study of the S-construction: it gives a precise conceptual explanation of when the passage S -> S(S)^op is involutive, namely excellence, and it packages the idempotent-completion phenomenon cleanly in Theorem 0.9(iii)-(v). The promised sequel on uniqueness of enhancements indicates that these notions are likely to have genuine applications. The paper is careful to state the main theorem with explicit caveats (almost involution, direct summands, idempotent-complete isometry), and the proof is structured through many small lemmas with precise references to [2] and [6]. A significant limitation is that all the concrete examples of excellent metrics supplied in the paper rely on results from the unpublished preprint [6]; the non-vacuity of the class is therefore not established within this manuscript alone.","major_comments":[{"comment":"The paper's motivating examples of excellent metrics are not self-contained. In Example 2.3, the verification of Definition 2.1(iii) uses [6, Remark 0.24, Lemma 2.8, Proposition 2.14, Lemma 8.5], and Example 8.3 relies on [6, Proposition 2.6 and Corollary 2.2.1]. Since Theorem 0.9 is a statement about excellent metrics, the applicability of the main theorem to the motivating examples depends entirely on the correctness of the unpublished preprint [6]. This is an unresolved external dependency that is load-bearing for the claim that the class of excellent metrics is 'large'. I recommend that the author either include proofs of the needed results from [6] in an appendix, or explicitly state in the introduction and in Theorem 0.9 that the examples are conditional on [6] being correct, with a clear indication of which statements are imported. Without this, a reader cannot presently verify that the main theorem has any non-formal instances.","section":"Example 2.3 and Section 8"}],"minor_comments":[{"comment":"There is a typo in Theorem 0.9(ii): 'catgeories' should be 'categories'. Also the abstract contains a malformed symbol '\\mathfrak(\\mathcal{S})' that should be '\\mathfrak{S}(\\mathcal{S})'.","section":"Abstract and Theorem 0.9(ii)"},{"comment":"In the statement of Lemma 1.16(ii), 'repsectively' is a typo for 'respectively'. The same typo appears in the proof and in a few other places (e.g., Lemma 3.8 and Lemma 3.10).","section":"Lemma 1.16(ii)"},{"comment":"In Lemma 7.8, 'untegern' should be 'integer n'. Also in Example 7.9, 'nessecary' should be 'necessary'.","section":"Lemma 7.8"},{"comment":"In the proof of Lemma 7.6, 'nust' should be 'must'.","section":"Lemma 7.6"},{"comment":"The notation for the functor Ψ and its relation to bY is somewhat confusing, since bY is introduced earlier as a functor from (Mod-S)^op to Mod-S(S)^op, while Ψ is a functor S^op -> S(S(S)^op). A direct sentence stating that Ψ is the restriction of bY to Y(S)^op (up to the equivalence of Proposition 2.17) would help the reader.","section":"Definition 4.5 and Theorem 4.6"}],"recommendation":"major_revision","confidential_remarks":"The paper leans very heavily on the author's own prior work, especially the unpublished preprint [6]. The main theorem is conditional on the class of excellent metrics being nonempty, and the paper's own examples of such metrics all depend on [6]. I would ask the editor to have the author either make [6] publicly available in a stable form or to include the necessary statements and proofs in this paper. This is not a correctness issue internal to the arguments, but it is a matter of verifiability of the central claim's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main take: this is a solid sequel. The genuinely new object is the class of excellent metrics (Definition 2.1), sitting inside good metrics, for which the passage S -> S(S)^op is almost an involution: a fully faithful triangulated functor whose image is idempotent-complete (Theorem 0.9(iii)-(v)). That theorem is stated with precise caveats, and the proof is long, lemma-heavy, and internally consistent as far as I can tell. The paper also computes a new category T^sb, generalizing K^b(R-Proj) (Examples 0.24 and Definition 8.5). The author is transparent about the limits, e.g., Discussions 1.9 and 6.8.\n\nThe credit: this is not a hand-wavy programmatic paper. The diagrams and technical conditions are stated carefully, and the idempotent-completion formulation is honest about what can fail. The dependence on the author's earlier papers [2] and [6] is substantial but appropriate for a sequel; it is dependency, not circularity.\n\nThe soft spot: load-bearing reliance on the unpublished preprint [6]. Example 2.3, which is the main source of excellent metrics, verifies condition (iii) of Definition 2.1 through [6, Lemma 8.5] and related statements. Example 8.3 also uses [6]. If any of those statements fail, the class of excellent metrics could be nearly empty, and Theorem 0.9 would have few or no known instances. This is not an internal flaw in the proof; but for a referee, it means the paper's applicability rests on a source that is not publicly available in final form. I would want [6] posted or the needed lemmas proved in an appendix before final acceptance.\n\nMinor: the paper is very long and the notation is heavy, typical for this program. No machine-checked proofs, but I do not see that as a flaw at this level.\n\nAudience: derived-category people working on enhancements, t-structures, and the T^c/T^b_c program. If the sequel with Canonaco and Stellari indeed proves new uniqueness-of-enhancements results, this paper will be an important reference.\n\nRecommendation: send to peer review; the referee should require that the dependence on [6] be resolved.","headline":"A careful, internally consistent theory of excellent metrics and their almost-involutive completion; the main caveat is that the motivating examples depend on an unpublished preprint.","tokens_in":75895,"tokens_out":2307,"would_cite":true,"duration_ms":24950,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["18G80"],"pacs":[],"model":"deepseek-v4-flash","headline":"For triangulated categories with an excellent metric, the S-to-S(S)^{op} passage is an almost-involution: iterating it returns S up to idempotent completion, and on categories of the form S(R) it is an isometric equivalence.","keywords":["triangulated categories","excellent metrics","good metrics","t-structures","derived categories","completions","involution","uniqueness of enhancements"],"falsifier":"Exhibit a triangulated category $\\mathcal{S}$ with a good metric satisfying Definition 2.1(i) and (ii) but not (iii), and check whether the comparison functor $\\widehat{Y} : \\mathcal{S} \\to \\mathfrak{S}(\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}})^{\\mathrm{op}}$ is fully faithful and whether every object of the target is a direct summand of an image object: if both hold for such a metric, then excellence is not necessary for the almost-involution, and if either fails, condition (iii) is essential. A concrete test case is the homotopy category $K^b(R\\text{-}\\mathrm{Proj})$ with its truncation metric, for a ring $R$ for which $D(R\\text{-}\\mathrm{Mod})$ is weakly approximable but not coherent; in that setting condition (iii) is the only clause in doubt.","tokens_in":74891,"feed_emoji":"🔁","tokens_out":24017,"duration_ms":193500,"temperature":0.7,"pith_summary":"Starting from a triangulated category $\\mathcal{S}$ equipped with a 'good metric' (a nested family of full subcategories $\\{M_i\\}$ playing the role of balls of radius $2^{-i}$), the paper's predecessor built a new triangulated category $\\mathfrak{S}(\\mathcal{S})$ out of Yoneda-image colimits of Cauchy sequences. This paper singles out a subclass of good metrics, the 'excellent' ones, and proves that for them the passage $(\\mathcal{S},\\{M_i\\}) \\mapsto (\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}}, \\{N_i^{\\mathrm{op}}\\})$ is almost an involution: applying it twice returns $\\mathcal{S}$ up to direct-summand completion, and exactly when $\\mathcal{S}$ already has the form $\\mathfrak{S}(\\mathcal{R})$ the return is an isometric triangle equivalence. Excellence is a strong condition, yet it is satisfied by the natural truncation metrics coming from t-structures on weakly approximable triangulated categories, giving new examples such as $K^b(R\\text{-}\\mathrm{Proj})$ with its truncation metric. The paper announces that these involutivity results will feed into new theorems about uniqueness of enhancements in a sequel.","feed_headline":"Excellent metrics make S to S(S) almost involutive","feed_subtitle":"For excellent metrics, applying the construction twice returns the category up to idempotent completion.","key_machinery":"The load-bearing object is the Yoneda-based completion $L(\\mathcal{S})$: the full subcategory of right $\\mathcal{S}$-modules whose objects are colimits of Yoneda images of Cauchy sequences in $\\mathcal{S}$, inside which the category $\\mathfrak{S}(\\mathcal{S})$ is cut out by the formula $\\mathfrak{S}(\\mathcal{S}) = L(\\mathcal{S}) \\cap \\bigcup_n Y(M_n)^\\perp$. The central mechanism is the notion of a 'type-$n$ morphism' between Cauchy sequences, meaning that the third vertices of the associated triangles lie in $M_n$; condition (iii) of excellence postulates the existence of type-$m$ morphisms $Y(F) \\to D$ with $D$ in $\\mathfrak{S}(\\mathcal{S}) \\cap L_n^\\perp$. These type-$m$ morphisms allow the restricted Yoneda functor $\\widehat{Y} : (\\mathrm{Mod}\\text{-}\\mathcal{S})^{\\mathrm{op}} \\to \\mathrm{Mod}\\text{-}\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}}$ to induce an equivalence $L(\\mathcal{S})^{\\mathrm{op}} \\cong L(\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}})$ and to transport strong triangles, and this equivalence of completions is what yields the almost-involution.","core_discovery":"The paper's central claim is Theorem 0.9: inside the class of good metrics on triangulated categories there is a subclass of 'excellent' metrics for which the assignment $(\\mathcal{S},\\{M_i\\}) \\mapsto (\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}}, \\{N_i^{\\mathrm{op}}\\})$ is almost an involution. Concretely, there is a fully faithful triangulated functor $\\widehat{Y} : \\mathcal{S} \\to \\mathfrak{S}(\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}})^{\\mathrm{op}}$ whose image is dense up to direct summands: every object of the double category is a direct summand of some $\\widehat{Y}(X)$, and every object of the induced metric ball $\\widehat{M}_i$ is a direct summand of some $\\widehat{Y}(M_i)$. If $\\mathcal{S}$ itself is of the form $\\mathfrak{S}(\\mathcal{R})$ for some triangulated category $\\mathcal{R}$ with an excellent metric, or if $\\mathcal{S}$ is idempotent-complete and each $M_i$ is idempotent-complete, then $\\widehat{Y}$ is a triangle equivalence and is an isometry of metric categories. The paper also proves that excellence is preserved by the construction, so the process can be iterated and the idempotent completion stops after the first step.","pith_inferences":["Read as a duality statement, the construction behaves like an antitone involution on the 'completed' categories: since $L(\\mathcal{S})^{\\mathrm{op}} \\cong L(\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}})$ (Proposition 2.17), the metric completions are exactly exchanged by the construction, and $\\mathcal{S}$ itself is captured only up to idempotent completion; a testable consequence is that idempotent co","Condition (iii) of excellence reads like a metric-theoretic 'enough approximability' hypothesis; one could test whether, for a weakly approximable triangulated category $\\mathcal{T}$, excellence of the truncation metric on $\\mathcal{T}^{c}$ is equivalent to coherence of $\\mathcal{T}$, which would turn Example 0.24 into a sharper dichotomy.","Because the motivating metrics are intrinsic (they are produced by recipes from the triangulated category alone, up to equivalence), the involutivity would transfer any uniqueness-of-enhancement property in both directions between $\\mathcal{S}$ and $\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}}$; this suggests a proof strategy for the announced sequel: prove uniqueness for the fixed points of the constr","The paper proves the almost-involution for excellent metrics but leaves open whether the 'almost' can be dropped for metrics that are only 'very good' (Definition 5.9); a concrete test would be to run the double construction on a very-good-but-not-excellent metric and check whether the fully faithfulness or the direct-summand density of the comparison functor survives."],"forward_implications":["Excellent metrics are closed under the construction: if $\\{M_i\\}$ is excellent on $\\mathcal{S}$, the induced metric $\\{N_i^{\\mathrm{op}}\\}$ is excellent on $\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}}$ (Proposition 4.1), so the process can be iterated freely.","For an excellent metric, the comparison functor $\\widehat{Y} : \\mathcal{S} \\to \\mathfrak{S}(\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}})^{\\mathrm{op}}$ is fully faithful and triangulated, with every object of the target a direct summand of an image object; if $\\mathcal{S}$ is idempotent-complete with idempotent-complete $M_i$, or if $\\mathcal{S}$ is already of the form $\\mathfrak{S}(\\mathcal{R})$, thi","For weakly approximable triangulated categories $\\mathcal{T}$ with a compact generator $G$, the subcategory $\\mathcal{T}^{sb} = \\bigcup_m \\langle G \\rangle^{[-m,m]}$ carries an explicit excellent metric given by $\\mathcal{T}^{sb} \\cap \\mathcal{T}^{\\leq -\\ell}$, and it equals $\\mathfrak{S}((\\mathcal{T}^b)^{\\mathrm{op}})^{\\mathrm{op}}$ (Definition 8.5).","When $\\mathcal{T} = D(R\\text{-}\\mathrm{Mod})$, the subcategory $\\mathcal{T}^{sb}$ is $K^b(R\\text{-}\\mathrm{Proj})$; the truncation metric is always excellent on it, while its restriction to the finitely generated projective complexes is excellent exactly when $D(R\\text{-}\\mathrm{Mod})$ is coherent (Example 0.24).","The paper states that a forthcoming sequel will use these involutivity results to prove new and surprising statements about uniqueness of enhancements."],"supporting_citations":[{"why":"introduces good metrics, the construction S(S), strong triangles, and good extensions on which this paper builds.","marker":"[2]"},{"why":"supplies the weakly approximable/coherent results (including the propositions the paper cites as [6, Proposition 2.14] and [6, Lemma 8.5]) used to verify excellence for the motivating examples.","marker":"[6]"},{"why":"provides the bound on products/coproducts in compactly generated t-structures used to certify the new excellent metrics of Section 8.","marker":"[1]"},{"why":"is the classical derived-category Morita theorem whose generalization motivated the metric framework and the involutivity question.","marker":"[7]"}],"fun_headline_variants":["Excellent metrics yield near-involutive construction","Excellent metrics make double construction almost involutive","Excellent metrics give almost involutive double construction","For excellent metrics, the construction is almost its own inverse","Excellent metrics: double construction is involution up to idempotent completion"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire conclusion depends on the hardest clause in the definition of an excellent metric: at every scale m, each object of the starting category must admit a controlled map into an object of the constructed category that receives no maps from the finer objects of scale n; this existence clause is not automatic for good metrics, and the paper can verify it in examples only by importing the weakly approximable/coherent machinery of [2] and an unpublished preprint.","fun_headline_variants_meta":{"raw":{"variants":["Excellent metrics yield near-involutive construction","Excellent metrics make double construction almost involutive","Excellent metrics give almost involutive double construction","For excellent metrics, the construction is almost its own inverse","Excellent metrics: double construction is involution up to idempotent completion"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001151,"raw_usage":{"total_tokens":4834,"prompt_tokens":1068,"completion_tokens":3766,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":684,"completion_tokens_details":{"reasoning_tokens":3690}},"tokens_in":684,"tokens_out":3766,"duration_ms":24867,"temperature":1.0,"reasoning_tokens":3690,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:40:28.081286+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a triangulated category $\\mathcal{S}$ with a good metric satisfying Definition 2.1(i) and (ii) but not (iii), and check whether the comparison functor $\\widehat{Y} : \\mathcal{S} \\to \\mathfrak{S}(\\mathfrak{S}(\\mathcal{S})^{\\mathrm{op}})^{\\mathrm{op}}$ is fully faithful and whether every object of the target is a direct summand of an image object: if both hold for such a metric, then excellence is not necessary for the almost-involution, and if either fails, condition (iii) is essential. A concrete test case is the homotopy category $K^b(R\\text{-}\\mathrm{Proj})$ with its truncation metric, for a ring $R$ for which $D(R\\text{-}\\mathrm{Mod})$ is weakly approximable but not coherent; in that setting condition (iii) is the only clause in doubt.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the weakly approximable/coherent results (including the propositions the paper cites as [6, Proposition 2.14] and [6, Lemma 8.5]) used to verify excellence for the motivating examples."},{"cited_title":"F. Enriques","cited_arxiv_id":null,"evidence_quote":"is the classical derived-category Morita theorem whose generalization motivated the metric framework and the involutivity question."}],"review_version":1}