{"id":"9744b5d0-6b6a-4e4c-824d-4a2315d2fbf1","arxiv_id":"1908.04132","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Using Freyd categories and generalized morphisms, the paper gives explicit algorithms for computing natural transformations between finitely presented functors and for constructing spectral sequence differentials in abelian categories.","lead":"This paper shows how to make abstract category theory computable, with explicit algorithms for natural transformations between functors like Ext and Tor over a ring, and for the differentials on pages of spectral sequences. A generalist might read it because it outlines a path to automating homological algebra, a tool used across geometry and algebra.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Computational claim over arbitrary commutative coherent rings outruns the supplied algorithms: coherence gives existence of syzygies, not decidability of equality or lifts.","rationale":"The reader's weakest assumption identifies exactly the same load-bearing concern: the cascade requires decidable lifts and weak kernels, and these are only demonstrated for Gröbner-base rings, not for arbitrary commutative coherent rings. My stress-test confirms this is the primary gap. The paper's mathematical equivalences and the generalized-morphism spectral-sequence construction appear sound or deferred to prior work, so the central issue is scope of the computational claim rather than an internal contradiction. The appropriate remedy is a conditional acceptance requiring the authors to qualify the R-generality and to state the algorithmic preconditions explicitly. I therefore do not alter the reader's CONDITIONAL verdict.","tokens_in":30122,"tokens_out":11044,"duration_ms":122651,"concrete_test":"Formalize the minimal assumptions needed for Construction 1.54 and attempt to prove decidable lifts in Rows_R from coherence alone. As a countercheck, take R = Z[x1,...,xn]/I with I finitely generated and ideal membership undecidable (possible by encoding Diophantine equations). Show that the lift-decision problem for the cospan with C = 0 reduces to deciding whether a given element of R equals 0, i.e., ideal membership. Since this is undecidable for such R, Construction 1.54 cannot be instantiated for every commutative coherent ring. This settles that the abstract's universal claim must be narrowed to rings explicitly equipped with decidable equality, decidable lifts, and computable syzygies.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's first guiding question promises computation of natural transformations over a commutative coherent ring R (abstract and Section 1.7). Construction 1.54, however, requires as input that Cp(R) be computable, that Rows_R have decidable lifts, and that weak kernels (syzygies) be computable. Coherence alone guarantees only the classical existence of finite syzygy generating sets; it does not guarantee algorithms for them. Decidable lifts are strictly stronger than existence and are not implied by coherence. The paper's own Example 1.4 shows that equality in a finitely presented single-object category can be undecidable, and analogous finitely presented commutative rings (Noetherian, hence coherent) can have undecidable ideal membership, making Cp(R) non-computable. The Gröbner-base examples (1.35, 1.44, 1.45) cover quotients of polynomial rings and localizations, not all coherent rings. Thus the central algorithmic claim is conditional on base decidability assumptions that are not stated in the abstract or in Section 1.7. The mathematical equivalence fpp(R-fpmod,Ab) ≅ Ap(Ap(C(R)^+)^op) is not in question; what is in question is whether the promised computation can be carried out for every commutative coherent R.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents an introduction to constructive category theory organized around two computational guiding questions. The first is how to compute sets of natural transformations between finitely presented functors such as Ext and Tor over a commutative coherent ring R. The proposed answer is a cascade of category constructors: starting from the single-object category Cp(R), passing to the additive closure Cp(R)⊕, then to Freyd categories, to obtain an equivalence fpp(R-fpmod, Ab) ≅ Ap(Ap(Cp(R)⊕)op). The authors develop computable versions of categories, Ab-categories, additive closure, homomorphism structures, and Freyd categories, and show how kernels, cokernels, lifts along monomorphisms, and homomorphism sets can be computed under additional assumptions such as decidable lifts and computable weak kernels. The second question asks how to construct spectral-sequence differentials for a filtered cochain complex using only operations provided by the axioms of an abelian category. The answer uses a calculus of generalized morphisms (spans up to stable equivalence), including pseudo-inverses, pullback/pushout computation rules, and a generalized homomorphism theorem, yielding explicit formulas for the differentials d^{p,q}_r = emb·B^{p+q}·proj. The paper includes worked examples over Z and Q[x,y], and points to an implementation in the GAP package CAP.","tokens_in":30259,"tokens_out":7405,"duration_ms":78368,"significance":"If the claims are brought into line with their hypotheses, this paper is a useful contribution that connects computational algebra with categorical homological algebra. The central categorical equivalence is classical (Freyd, Auslander) and is not at issue, and the generalized-morphism calculus is proven in the text (Theorems 2.10, 2.15, 2.17), giving explicit, implementable formulas for connecting maps and spectral-sequence differentials. The paper also gives concrete examples (Examples 1.55–1.57) and references a working software implementation (CAP project), which strengthens its practical relevance. The main weakness is that the first guiding question is stated for all commutative coherent rings, but the provided algorithms require strictly stronger computational hypotheses; this is a fixable scope problem rather than an error in the underlying mathematics. The spectral-sequence section provides a constructive description of differentials but does not address convergence, which is a limitation that should be stated explicitly.","major_comments":[{"comment":"The abstract and §1.7 promise computation of natural transformations between finitely presented functors over an arbitrary commutative coherent ring R. The algorithm in Construction 1.54 requires more than coherence: by Remark 1.15 and Example 1.24, Cp(R) must be computable (decidable equality); by Definition 1.33, Rows_R must have decidable lifts; by Example 1.44 and Definition 1.43, weak kernels (syzygies) must be computable. Coherence guarantees only the existence of finite syzygy generating sets, not algorithms for them, and decidable lifts are strictly stronger than existence. The paper's own Example 1.4 shows that finitely presented monoids can have undecidable equality, and finitely presented commutative rings with undecidable word problem are coherent, so Cp(R) is not computable for such R. The Gröbner-basis examples (1.34, 1.35, 1.44, 1.45) cover quotients of polynomial rings over fields with decidable equality, localizations thereof, and path-algebra quotients, not all coherent rings. I therefore recommend either adding explicit constructive hypotheses (e.g., decidable equality, computable weak kernels, and decidable lifts in Rows_R) to the abstract and §1.7, or restricting the computational claim to the classes covered by the examples. The categorical equivalence itself is unaffected.","section":"§1.7, Construction 1.54; also Abstract and §1.5"},{"comment":"The correctness of the central Freyd-category algorithms is not proved in the manuscript; §1.6 states: \"For details about the correctness of these constructions, we refer the reader to [Pos17a].\" For a paper whose stated goal is to present methods of constructive category theory, this leaves the reader unable to verify the algorithms from the paper alone. The constructions appear correct, but I ask the author to state the correctness results as numbered propositions with proof sketches, or at least to give precise pointers to the corresponding statements in [Pos17a] so that each algorithm's correctness can be checked without reconstructing the proofs.","section":"§1.6, Constructions 1.40, 1.50, 1.53; Remark 1.52"},{"comment":"The paper constructs pages E_r and differentials d^{p,q}_r = emb·B^{p+q}·proj for a filtered cochain complex and asserts that the cohomologies of the r-th honest complex determine the objects of the (r+1)-th page. The verification is compressed into a variable substitution and does not show in detail that the constructed d^{p,q}_r satisfy d_r^2 = 0 and that the isomorphism E_{r+1}^{p,q} ≅ ker(d_r^{p,q})/im(d_r^{p-r,q+r-1}) is the canonical one. Since the second guiding question is specifically about constructing spectral-sequence differentials, I would like to see a proof outline explaining how the standard E_r-page of the filtered complex is recovered, along with a statement of any boundedness conditions needed for convergence, or an explicit remark that convergence is not addressed.","section":"§2.7, final paragraph and Definition 2.21"}],"minor_comments":[{"comment":"There is a typo in the abstract: \"by a nswering two guiding computational questions\" should read \"by answering two guiding computational questions.\"","section":"Abstract"},{"comment":"\"Futhermore\" should be \"Furthermore.\"","section":"Remark 1.15"},{"comment":"The diagrams labeled Figure 1 and Figure 2 do not appear as numbered floats in the text; they should either be converted into proper figures or referred to as inline diagrams, depending on the journal's style.","section":"Figures 1 and 2"},{"comment":"The construction of the Cp(R)-homomorphism structure uses commutativity of R via H(a,b) = a·b; since composition is defined as precomposition and Example 1.9 notes Cp(R) equals R^op, the role of commutativity should be stated explicitly.","section":"Example 1.24"},{"comment":"The displayed formula for d^{p,q}_r in the final paragraph contains a formatting artifact (\"M p`q`1\") and should be typeset as M^{p+q+1}.","section":"§2.7"}],"recommendation":"major_revision","confidential_remarks":"The manuscript leans heavily on the author's previous work [Pos17a, Pos17b] for correctness proofs of the Freyd-category constructions and for parts of the generalized-morphism theory. The editor should verify that those deferred proofs are in a sufficiently accessible and stable venue, because as it stands the present paper is not fully self-contained. The main reason for the major-revision recommendation is the scope mismatch in the first guiding question: the abstract promises computation over all commutative coherent rings, while the algorithms require decidability and computability assumptions that fail for some coherent rings; this is straightforwardly fixable by revising the stated hypotheses."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper is a well-organized introduction to constructive category theory built around two computational questions: computing natural transformations between finitely presented functors, and computing spectral sequence differentials explicitly. The genuinely new material is in Section 2.7: a clean formula for the spectral sequence differentials d^{p,q}_r = emb ∘ B^{p+q} ∘ proj, plus a construction that extracts an honest cochain complex (and its cohomology) from a generalized cochain complex. The worked examples over Q[x,y] — Hom(Tor_1(M,–), Ext^1(M,–)) — are informative and show the Freyd-category machinery in action.\n\nMost of the Freyd category content is a summary of Posur's prior work [Pos17a], and the generalized morphism calculus follows Barakat [Bar09]. That's fine: the paper is upfront about this and the synthesis is valuable. It emphasizes computational issues that are often glossed over, especially the need to treat equality of morphisms as an algorithm rather than a given, and Example 1.4 (undecidable word problem) makes that point nicely. The correctness statements are mostly deferred, but the central equivalences are standard.\n\nThe main soft spot is the generality claimed in the abstract and Section 1.7. The paper promises computation over \"a commutative coherent ring R,\" but Construction 1.54 and the surrounding algorithms require more: Cp(R) must be computable, Rows_R must have decidable lifts, and weak kernels (syzygies) must be computable. Coherent means only that syzygy generating sets exist; it does not mean they can be found algorithmically, nor that equality in R is decidable. The Gröbner-base examples cover quotients of polynomial rings and localizations, not all coherent rings. So the computational claim outruns the supplied algorithms. The fix is easy: state the extra hypotheses in the abstract and in Section 1.7, and note explicitly where coherence alone is insufficient.\n\nTwo smaller issues: the spectral sequence part is presented as an introduction, and convergence is not addressed, which is fine if labeled as such; and no code is shipped, so reproducibility rests on the CAP/GAP packages and the cited papers, which is acceptable for a theory paper but worth stating more explicitly.\n\nOverall, the paper is a useful map for anyone who wants to implement constructive homological algebra, and the spectral sequence formulas deserve wider circulation. It should go to referees; I would send it with a request to fix the overclaim before acceptance.\n\nBest,\n[You]","headline":"A clear, useful synthesis of Posur's and Barakat's computational machinery, with real new spectral-sequence formulas, but the abstract overpromises computation over arbitrary coherent rings.","tokens_in":30885,"tokens_out":2955,"would_cite":true,"duration_ms":31119,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["18E10","18E05","18A25","18E25"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper shows that homomorphism sets between finitely presented functors over a commutative coherent ring can be computed by building an equivalent category through a cascade of constructors, and that every spectral sequence…","keywords":["constructive category theory","finitely presented functors","Freyd categories","homomorphism structures","generalized morphisms","spectral sequences","diagram chases","coherent rings"],"falsifier":"On a double complex $C^{\\bullet,\\bullet}$ filtered by rows, the spectral sequence's first differential must equal $d_h + (-1)^p d_v$; compute $d^{p,q}_1 = \\operatorname{emb}\\cdot B^{p+q}\\cdot \\operatorname{proj}$ for a small example, say the double complex associated to a square of abelian groups, and compare the result with the classical formula. Any mismatch would falsify Section 2.7's construction.","tokens_in":29803,"feed_emoji":"🧮","tokens_out":13470,"duration_ms":124585,"temperature":0.7,"pith_summary":"The paper's project is to show that two high-level outputs of homological algebra can be computed by explicit algorithms that stay within the operations a category provides. The first output is the set of natural transformations between two finitely presented functors, e.g., $\\operatorname{Ext}^1(M,-)$ and $\\operatorname{Tor}_1(M,-)$, over a commutative coherent ring $R$; the paper builds a category equivalent to the category of such functors by a cascade of category constructors, $\\mathrm{fpp}(R\\text{-}\\mathrm{fpmod},\\mathrm{Ab}) \\simeq \\mathcal{A}(\\mathcal{A}(\\mathcal{C}(R)^+)^{op})$, and transfers computability up the cascade. The second output is every differential $d^{p,q}_r$ of the spectral sequence of a filtered cochain complex; the paper gives the closed formula $d^{p,q}_r = \\operatorname{emb} \\cdot B^{p+q} \\cdot \\operatorname{proj}$, computed from kernels, cokernels, pullbacks, and pushouts alone. The point of the project is that existence theorems of homological algebra, like the snake lemma or the pages of a spectral sequence, become concrete recipes that an implementation can execute.","feed_headline":"Hom-sets of Ext and Tor become computable via category constructors","feed_subtitle":"Finitely presented functors reduce to a double Freyd category; spectral sequence differentials get closed formulas.","key_machinery":"The Freyd category constructor $\\mathcal{A}(\\mathcal{A})$ is the central object: it takes an additive category $\\mathcal{A}$ and returns the category of finitely presented functors on $\\mathcal{A}$, with objects $A \\leftarrow RA$ as formal cokernels and morphisms as equivalence classes of lifts; iterating it builds the cascade equivalent to $\\mathrm{fpp}(R\\text{-}\\mathrm{fpmod},\\mathrm{Ab})$. Its companion is the transfer of homomorphism structures (Section 1.6.5): a homomorphism structure—a way of representing each hom set as an object of another category—on $\\mathcal{A}$ induces one on $\\mathcal{A}(\\mathcal{A})$ via a subquotient diagram, so computability of hom sets climbs the cascade. For spectral sequences, the load-bearing mechanism is the category $\\mathcal{G}(\\mathcal{A})$ of generalized morphisms, whose pseudo-inverse operation and the pullback/pushout computation rules rewrite diagram chases algebraically; the generalized homomorphism theorem then decomposes every generalized morphism into honest maps, yielding the differentials.","core_discovery":"The paper's central discovery is that the intangible objects of homological algebra—finitely presented functors and spectral sequence differentials—have finite, syntactic representatives that support algorithms. On the functor side, the paper identifies finitely presented contravariant functors on an additive category $\\mathcal{A}$ with its Freyd category $\\mathcal{A}(\\mathcal{A})$: objects are morphisms $A \\leftarrow RA$ thought of as formal cokernels, and morphisms are equivalence classes of commutative squares. Applying this twice, with an opposite in between, yields $\\mathcal{A}(\\mathcal{A}(\\mathcal{C}(R)^+)^{op}) \\simeq \\mathrm{fpp}(R\\text{-}\\mathrm{fpmod},\\mathrm{Ab})$ for a commutative coherent ring $R$, so computing homomorphism sets in the double Freyd category computes natural transformations such as $\\operatorname{Hom}(\\operatorname{Ext}^1(M,-),\\operatorname{Tor}_1(M,-))$. On the spectral sequence side, the paper's category $\\mathcal{G}(\\mathcal{A})$ of generalized morphisms—spans $A \\leftarrow C \\rightarrow B$ modulo stable equivalence, with pseudo-inverses and pullback/pushout computation rules—lets diagram chases be written as algebraic compositions. The paper shows that the $r$-th page objects are the canonical subquotients of these generalized morphisms, and the differentials are the honest maps obtained by restricting and projecting: $d^{p,q}_r = \\operatorname{emb}\\cdot B^{p+q}\\cdot \\operatorname{proj}$. The upshot is that both computations require no ambient module category: kernels, cokernels, pullbacks, and pushouts suffice.","pith_inferences":["Because the spectral-sequence formula uses only abelian-category operations, it should port essentially unchanged to abelian categories where element chases are awkward, such as sheaf categories or categories of filtered modules; testing $d_1$ on a two-term double complex filtered by rows would be a concrete check.","The paper's decidability requirement suggests a sharp boundary: the cascade computes natural transformations exactly for commutative coherent rings whose row modules admit a syzygy algorithm, and the undecidable word problem of Example 1.4 marks where no implementation can go.","The generalized-morphism calculus appears suited to deriving other connecting homomorphisms—boundary maps in long exact sequences of Ext or Tor—as explicit compositions of pseudo-inverses, following the snake lemma template."],"forward_implications":["Computing $\\operatorname{Hom}(\\operatorname{Ext}^i(M,-), \\operatorname{Tor}_j(M,-))$ reduces to matrix and kernel/cokernel computations over $R$, and becomes programmable whenever $R$ admits Gröbner bases and syzygy algorithms.","For graded rings and path-algebra quotients, the same cascade yields computational models of finitely presented graded modules and their functors.","Every page of the spectral sequence of a filtered cochain complex, including its differentials, is constructible from kernels, cokernels, pullbacks, and pushouts, with no embedding into a module category.","The identities $gker(B^{p,q}_r)=dom(B^{p,q}_{r+1})$ and $gimp(B^{p,q}_r)=def(B^{p-r,q+r-1}_{r+1})$ determine the page-to-page isomorphisms of the spectral sequence."],"supporting_citations":[{"why":"Supplies the Freyd category construction and the theorem that $\\mathcal{A}(\\mathcal{A})$ is abelian exactly when $\\mathcal{A}$ has weak kernels.","marker":"[Fre66]"},{"why":"Identifies Ext and Tor as finitely presented functors, the target objects of the computation.","marker":"[Aus66]"},{"why":"Proves correctness of the kernel, cokernel, and homomorphism-structure constructions inside Freyd categories.","marker":"[Pos17a]"},{"why":"Introduces generalized morphisms and their use for effective homological algebra and spectral sequences.","marker":"[Bar09]"},{"why":"Provides the Gröbner basis algorithms that make $\\mathrm{Rows}_R$ compute lifts and weak kernels.","marker":"[GP02]"},{"why":"Presents the pullback/pushout construction of the snake lemma that the generalized-morphism rules mirror.","marker":"[ML98]"},{"why":"Supplies the comparison theorem for presentations and the abelian category axioms.","marker":"[Wei94]"},{"why":"Contains the detailed development of generalized morphisms and the generalized homomorphism theorem.","marker":"[Pos17b]"}],"fun_headline_variants":["Ext and Tor hom-sets: now computable","Spectral sequence differentials get constructive formulas","Diagram chases without ambient categories","Finite syntax for homological algebra","Category constructors make functors calculable"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the base category $\\mathrm{Rows}_R$ has decidable lifts and computable weak kernels (syzygies); Gröbner bases supply this in the paper's examples, but for a commutative coherent ring with undecidable word problem the computation is impossible, so the method's scope is exactly the class of rings where such algorithms exist.","fun_headline_variants_meta":{"raw":{"variants":["Ext and Tor hom-sets: now computable","Spectral sequence differentials get constructive formulas","Diagram chases without ambient categories","Finite syntax for homological algebra","Category constructors make functors calculable"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000351,"raw_usage":{"total_tokens":1948,"prompt_tokens":1016,"completion_tokens":932,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":632,"completion_tokens_details":{"reasoning_tokens":867}},"tokens_in":632,"tokens_out":932,"duration_ms":9252,"temperature":1.0,"reasoning_tokens":867,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:51:18.750210+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On a double complex $C^{\\bullet,\\bullet}$ filtered by rows, the spectral sequence's first differential must equal $d_h + (-1)^p d_v$; compute $d^{p,q}_1 = \\operatorname{emb}\\cdot B^{p+q}\\cdot \\operatorname{proj}$ for a small example, say the double complex associated to a square of abelian groups, and compare the result with the classical formula. Any mismatch would falsify Section 2.7's construction.","supporting_citations":[],"review_version":1}