{"id":"2880a342-9bca-485f-b154-9250f465e68f","arxiv_id":"1908.03403","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For double Danielewski surfaces the Makar-Limanov invariant is k[x], and under separability assumptions the surfaces are stably isomorphic without being isomorphic.","lead":"This paper studies a new family of affine surfaces built from two polynomial equations, called double Danielewski surfaces. It computes their Makar-Limanov invariant and shows they give new counterexamples to the Cancellation Problem.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.14's step 'dimensions equal, hence kernel of ι is xE' is not justified; injectivity of E/xE→A/xA must be shown directly.","rationale":"The reader's weakest assumption identifies the coprimality hypotheses as the load-bearing condition. My read agrees that those hypotheses are the key, but the specific soft spot is different: Theorem 3.14's proof of the key identity A^φ = E contains a non-sequitur. The dimension equality does not by itself imply that E/xE embeds into A/xA, because E/xE can be a reducible one-dimensional ring (a polynomial ring over a zero-dimensional base, with one line per irreducible factor of P(0,F)). The missing injectivity can be supplied by writing the induced map explicitly and using the same unit condition (19); so the theorem is very likely correct, and the reader's ACCEPT is not overthrown by a mathematical counterexample. However, because the cancellation claim rests on this proof step, a referee should require the injectivity argument to be written out rather than inferred from dimensions. Hence CONDITIONAL rather than outright ACCEPT.","tokens_in":13700,"tokens_out":40848,"duration_ms":433416,"concrete_test":"Verify the claimed injection directly. In Theorem 3.14, define Ψ : k[F,G,H]/(P(0,F), Q(0,G,F)) → k[Y,Z,W]/(P(0,Z), Q(0,Y,Z)) by F ↦ z~, G ↦ y~, H ↦ P'(0,z~)Q'(0,y~,z~)w~. Show Ψ is well-defined and, using (19), that P'(0,z~)Q'(0,y~,z~) is a unit, so Ψ is an isomorphism onto ι(E). If this isomorphism holds, the kernel of ι is xE and Theorem 3.14 is complete; if not, the stable-isomorphism conclusion lacks support.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The stable-isomorphism theorem rests on proving A^φ = E. The paper reduces this to showing that the kernel of ι : E → A/xA is exactly xE, and then asserts: 'dim(E/xE) = 1 = dim ι(E). Hence kernel of ι is xE.' Equal Krull dimensions do not imply injectivity of a surjective map of affine algebras. Here E/xE is a polynomial ring in one variable over the zero-dimensional ring k[F,G]/(P(0,F), Q(0,G,F)); if P(0,F) splits, this is a disjoint union of lines, and a nonzero kernel supported on one component would preserve dimension. The needed argument is available: the induced map sends F ↦ z~, G ↦ y~, H ↦ P'(0,z~)Q'(0,y~,z~)w~, and by (19) the multiplier is a unit, so E/xE → ι(E) is an isomorphism. As written, however, the proof skips this and uses a dimension criterion that is formally insufficient. This is a genuine gap in a load-bearing step, although not evidence that the theorem is false.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies double Danielewski surfaces B_{d,e}=k[X,Y,Z,T]/(X^dY-P(X,Z), X^eT-Q(X,Y,Z)), with deg_Z P=r>=2 and deg_Y Q=s>=2. It computes the Makar-Limanov invariant of these surfaces, showing that under the conditions of (4) one has ML(B)=k[x] (Theorem 3.8); it gives a complete isomorphism classification (Theorem 3.10); it characterizes automorphisms (Theorems 3.12-3.13); and it proves stable isomorphisms B_{d,e}[1] =~ B_{d,e-1}[1] under explicit coprimality conditions (Theorem 3.14), yielding new counterexamples to the Cancellation Problem (Corollary 3.15). The proofs rely on exponential maps, admissible filtrations, and the Derksen-Hadas-Makar-Limanov homogenization theorem.","tokens_in":103,"tokens_out":31033,"duration_ms":408427,"significance":"If the results hold, the paper provides a new two-dimensional family of affine surfaces over arbitrary fields that fail the cancellation property, together with an explicit classification of isomorphisms and automorphisms. The stable-isomorphism construction is explicit and parameter-free, and the ML-invariant computation is a genuine extension of the classical Danielewski theory. The main gap identified below concerns a formally insufficient step in the proof of Theorem 3.14, but the theorem can be repaired by a direct argument; the overall contribution is substantial and publishable in a good journal after revision.","major_comments":[{"comment":"The step 'dim(E/xE)=1=dim ι(E). Hence kernel of ι is xE' is not justified: for a surjective map of affine k-algebras of equal Krull dimension, a nonzero kernel supported on a single component need not lower the dimension. The intended conclusion follows directly from the explicit map: the induced map E/xE -> ι(E) sends F to z~, G to y~, and H to P'(0,z~)Q'(0,y~,z~)w~; by (19) the multiplier is a unit, so this map is an isomorphism. Please replace the dimension criterion with this explicit isomorphism, since the present argument is load-bearing for the proof that A^φ = E.","section":"Theorem 3.14"},{"comment":"The assertion 'Since A[1/x] = E[1/x][w]' is used to deduce E[1/x] = A^φ[1/x] and to compute the dimension of E, but it is not immediate. It can be verified from the displayed identities: from f = x^{d+e-1}w+z one obtains z ∈ E[1/x][w]; then from the formulas for g and h one obtains y and t in E[1/x][w]. Please include this verification, as the equality is load-bearing for the proof that A^φ = E.","section":"Theorem 3.14"},{"comment":"The proof uses without comment the equalities x^{d_1}_1 B ∩ k[x_1,z_1] = (x^{d_1}_1, P_1(x_1,z_1)) and the analogous equality for Q, which are not immediate from the definitions. These equalities can be derived from the normal form (2) established in Lemma 3.4, but the derivation should be supplied, since the matching of the parameters d_i and e_i in (9) and (13) rests on them.","section":"Theorem 3.10"}],"minor_comments":[{"comment":"In the statements of Lemmas 3.4 and 3.5, the graded ring is written as ⊕(B_n/B_{n+1}) and ⊕(D_n/D_{n+1}); with the given increasing filtrations these quotients are zero, and the intended expression is B_n/B_{n-1} and D_n/D_{n-1}, as used in the proofs.","section":"Lemmas 3.4 and 3.5"},{"comment":"In the sentence describing a homogeneous element of C in R, the range '0 ≤ j < r' should read '0 ≤ j < s' to match the representation obtained from the relation X^eT = Y^s.","section":"Lemma 3.6"},{"comment":"The citation to Lemma 2.1(iii) for the absence of nontrivial exponential maps on a non-normal one-dimensional ring appears to be a mismatched reference; the argument is better supported by Lemma 2.1(v) or by the standard fact that a polynomial ring over a field is normal.","section":"Lemma 3.6"}],"recommendation":"major_revision","confidential_remarks":"The mathematical claims appear sound and the counterexample family is interesting, but the proof of Theorem 3.14 contains a formally insufficient dimension argument and several compressed justifications that should be expanded before publication. The direct fix for the main gap is available, so I do not see the need for rejection. I would recommend a careful revision of §3.3 and a pass over the unproved intersection equalities in §3.2."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a good paper that deserves a serious referee. It constructs a genuinely new family of affine surfaces (the double Danielewski surfaces), computes their ML-invariant, classifies isomorphisms, describes automorphisms, and proves a stable isomorphism theorem yielding new counterexamples to the Cancellation Problem. The main theorems are not in the cited literature, and the proofs are mostly direct, with no fitted parameters or circular reasoning. The ML-invariant computation chains two filtrations and uses the standard Derksen-Hadas-Makar-Limanov homogenization theorem; that part reads cleanly. I agree with the reader's assessment.\n\nThe soft spots are localized. Theorem 3.10's proof is compressed, especially the steps showing d1=d2 and e1=e2 from the ideal equalities; they are believable but a referee should ask for a few more words. The larger issue is in Theorem 3.14. The paper wants to show A^phi=E, reduces to injectivity of E/xE -> A/xA, and then asserts 'dimensions equal, hence kernel is xE.' That implication is not valid as a general dimension argument: a surjection between two 1-dimensional rings can have nonzero kernel if the source is not a domain or not equidimensional. However, the paper has already displayed enough to fix it: the induced map sends F to z~, G to y~, H to P'(0,z~)Q'(0,y~,z~)w~, and by (19) the multiplier is a unit in A/xA. That gives an explicit inverse and proves injectivity. So the gap is a missing sentence, not a missing theorem. I would not call it a fatal flaw.\n\nThe citation pattern is normal; the self-citations are to background lemmas. The paper is aimed at affine algebraic geometers working on cancellation and locally nilpotent derivations. It is a substantial application of established tools rather than a new framework, but that's fine. A careful referee could usefully tighten Theorem 3.10 and patch Theorem 3.14 as above.\n\nMy recommendation: send it to review. It is the kind of paper where a knowledgeable referee can verify everything in a couple of days, and the result is worth having in the record.","headline":"A solid new family of cancellation counterexamples with complete proofs; one compressed step in the stable-isomorphism theorem needs a small fix but the argument goes through.","tokens_in":14416,"tokens_out":3492,"would_cite":true,"duration_ms":35591,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14R05","14R10","13A50","13B25","13A02","14R20"],"pacs":[],"model":"deepseek-v4-flash","headline":"A family of two-equation affine surfaces gives new counterexamples to the Cancellation Problem: non-isomorphic rings become isomorphic after adjoining one variable.","keywords":["Cancellation Problem","Double Danielewski surfaces","Makar-Limanov invariant","Exponential maps","Automorphism","Stable isomorphism","Affine surfaces"],"falsifier":"Take $d=e=2$, $P(0,Z)=Z^2+1$ and $Q(0,Y,Z)=Y^2+Z$ over a field of characteristic not $2$, then compute the Makar-Limanov invariant of $B_{2,2}$; if any exponential map fixes more than $k[x]$, Theorem 3.8 fails. Alternatively, construct the subring $E=k[x,f,g,h]$ inside $B_{2,2}[w]$ exactly as in the proof of Theorem 3.14 and check whether $E\\cong B_{2,1}$; if not, the cancellation conclusion for these data fails.","tokens_in":13510,"feed_emoji":"","tokens_out":17688,"duration_ms":169089,"temperature":0.7,"pith_summary":"The paper studies affine surfaces defined by two equations of the form $X^dY=P(X,Z)$ and $X^eT=Q(X,Y,Z)$, called double Danielewski surfaces. It shows that when $r=\\deg_Z P\\ge 2$ and $s=\\deg_Y Q\\ge 2$, or in two boundary cases, every nontrivial exponential map on the coordinate ring fixes exactly the polynomial subring $k[x]$, so the Makar-Limanov invariant is $k[x]$. It then classifies these rings up to isomorphism and describes their automorphisms as explicit polynomial substitutions. The main result is a stable-isomorphism theorem: under two coprimality conditions, $B_{d,e}[1]$ is isomorphic to $B_{d,e-1}[1]$, while $B_{d,e}$ and $B_{d,e+1}$ are not isomorphic; hence these surfaces are counterexamples to the Cancellation Problem. A curious reader should care because this gives a large, explicitly computable family of surfaces in which adjoining one variable erases a genuine difference.","feed_headline":"Two non-isomorphic surfaces become isomorphic after one variable","feed_subtitle":"A two-equation family of affine surfaces gives new counterexamples to the cancellation problem.","key_machinery":"The load-bearing object is the double Danielewski ring $B_{d,e}$, and the engine is the Makar-Limanov invariant—the common fixed subring of all nontrivial exponential maps from $B$ to $B[U]$. The invariant computation runs through two associated graded rings: first $D=k[X,Y,Z,T]/(X^dY-P(0,Z),\\,X^eT-Y^s)$, then $C=k[X,Y,Z,T]/(X^dY-Z^r,\\,X^eT-Y^s)$; every exponential map on $B$ induces homogeneous exponential maps on $D$ and $C$, where the invariant is shown to lie in $k[x]$. For the cancellation result, the key step is an element $v$ inside $A=B_{d,e}[w]$ formed from $w$, auxiliary polynomials $a,f,g,h$, and $x$; the coprimality hypotheses make a certain product a unit modulo $x$, which forces the exponential map to send $v$ to $v-U$ and identifies $A$ with a polynomial ring in $v$ over an invariant subring isomorphic to $B_{d,e-1}$.","core_discovery":"The central claim is that, for a fixed field $k$ and integers $d,e\\ge 1$, the coordinate ring $B_{d,e}=k[X,Y,Z,T]/(X^dY-P(X,Z),\\,X^eT-Q(X,Y,Z))$, with $r=\\deg_Z P\\ge 2$ and $s=\\deg_Y Q\\ge 2$, carries a complete structural theory. The Makar-Limanov invariant—the common fixed ring of all exponential maps, the algebraic counterpart of one-parameter additive group actions—is exactly $k[x]$ whenever $(r,s,e)$ satisfies condition (4) of Lemma 3.6. Two such rings are isomorphic only if their parameter quadruples $(d,e,r,s)$ agree and the polynomials are related by the explicit substitutions of Theorem 3.10, so no double Danielewski surface is isomorphic to any one-equation Danielewski surface. Under the additional coprimality assumptions on $P(0,Z)$, $Q(0,Y,Z)$ and their derivatives, $B_{d,e}$ and $B_{d,e+1}$ are non-isomorphic while $B_{d,e}[1]$ and $B_{d,e+1}[1]$ are isomorphic, giving the claimed counterexamples to the Cancellation Problem.","pith_inferences":["Because the proof is characteristic-free, the same construction yields dimension-two counterexamples to cancellation in positive characteristic, where the polynomial ring $k[3]$ itself is known not to cancel; this puts non-cancellation at the lowest possible dimension there.","The two-stage filtration through $D$ and $C$ looks like a general recipe for surfaces whose leading forms are monomial equations, so iterated versions of the construction with three or more equations might be analyzable by the same reduction.","The explicit isomorphism data in Theorem 3.10 can be read as an algorithm: search for the polynomial data $\\lambda,\\gamma,\\delta,f,g,h$ and check the two displayed congruences, turning isomorphism testing in the family into a polynomial-system problem.","The element $v$ constructed in Theorem 3.14 might serve as an explicit witness for the cancellation, giving a direct formula for the $B_{d,e-1}$ factor inside $B_{d,e}[w]$ rather than an existence proof."],"forward_implications":["For every admissible choice of $d,e,P,Q$, the rings $B_{d,e}$ and $B_{d,e+1}$ are non-isomorphic while their one-variable extensions are isomorphic, so the cancellation property fails for an infinite family of affine surfaces.","Iterating the stable isomorphism gives $B_{d,e}[1]\\cong B_{d,e'}[1]$ for all $e,e'\\ge 2$, so each fixed $d$ produces a chain of distinct rings that become the same after adjoining one variable.","Theorem 3.10 reduces isomorphism testing in this family to checking the parameter quadruple and solving polynomial substitution equations, making the classification explicit and algorithmic in spirit.","Corollary 3.11 shows the double surfaces are a genuinely new class: no member is isomorphic to any classical Danielewski surface.","Theorem 3.13 gives a short criterion for automorphisms: an endomorphism fixing $x$ and $k[x,z]$ is automatically an automorphism."],"supporting_citations":[{"why":"Supplies the original one-equation surfaces showing non-cancellation in dimension two; the present family is a two-equation extension of that construction.","marker":"[4]"},{"why":"Provides the homogenization theorem that lets an arbitrary exponential map on $B$ induce homogeneous exponential maps on the associated graded rings.","marker":"[5]"},{"why":"Supplies the standard lemma on exponential maps used throughout: invariant subrings are factorially closed, algebraically closed, and drop transcendence degree by one.","marker":"[3]"},{"why":"Cited for a lemma on exponential maps and for earlier positive-characteristic non-cancellation results that motivate the field-independent setting.","marker":"[12]"},{"why":"Establishes the stable-isomorphism phenomenon for classical Danielewski surfaces that Theorem 3.14 extends to the double surfaces.","marker":"[7]"},{"why":"Collects the standard facts on Danielewski surfaces and their Makar-Limanov invariants used for comparison and for the exceptional cases.","marker":"[8]"}],"fun_headline_variants":["Adding one variable makes non-isomorphic surfaces isomorphic","Cancellation problem fails: add a variable, surfaces match","Double Danielewski trick: non-isomorphism dies after one variable","New counterexamples: non-isomorphic after adding one variable","One variable added, surfaces become isomorphic—cancellation counterexample"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that after setting $X=0$, the polynomials $P(0,Z)$ and $P'(0,Z)$ together generate the whole ring $k[Z]$, and that $P(0,Z),Q(0,Y,Z),Q'(0,Y,Z)$ together generate $k[Y,Z]$; if that fails, the element $v$ used to prove the stable isomorphism is not known to exist and the argument collapses.","fun_headline_variants_meta":{"raw":{"variants":["Adding one variable makes non-isomorphic surfaces isomorphic","Cancellation problem fails: add a variable, surfaces match","Double Danielewski trick: non-isomorphism dies after one variable","New counterexamples: non-isomorphic after adding one variable","One variable added, surfaces become isomorphic—cancellation counterexample"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000202,"raw_usage":{"total_tokens":1323,"prompt_tokens":825,"completion_tokens":498,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":441,"completion_tokens_details":{"reasoning_tokens":414}},"tokens_in":441,"tokens_out":498,"duration_ms":5872,"temperature":1.0,"reasoning_tokens":414,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:15:54.643483+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take $d=e=2$, $P(0,Z)=Z^2+1$ and $Q(0,Y,Z)=Y^2+Z$ over a field of characteristic not $2$, then compute the Makar-Limanov invariant of $B_{2,2}$; if any exponential map fixes more than $k[x]$, Theorem 3.8 fails. Alternatively, construct the subring $E=k[x,f,g,h]$ inside $B_{2,2}[w]$ exactly as in the proof of Theorem 3.14 and check whether $E\\cong B_{2,1}$; if not, the cancellation conclusion for these data fails.","supporting_citations":[{"cited_title":"Danielewski, On a cancellation problem and automorphism groups of affine algebraic varieties , preprint 1989","cited_arxiv_id":null,"evidence_quote":"Supplies the original one-equation surfaces showing non-cancellation in dimension two; the present family is a two-equation extension of that construction."},{"cited_title":"Derksen, O","cited_arxiv_id":null,"evidence_quote":"Provides the homogenization theorem that lets an arbitrary exponential map on $B$ induce homogeneous exponential maps on the associated graded rings."},{"cited_title":"Crachiola, The hypersurface x +x2y +z2 +t3 = 0 over a field of arbitrary characteristic, Proc","cited_arxiv_id":null,"evidence_quote":"Supplies the standard lemma on exponential maps used throughout: invariant subrings are factorially closed, algebraically closed, and drop transcendence degree by one."},{"cited_title":"Gupta, On Zariski’s Cancellation Problem in positive characterist ic, Adv","cited_arxiv_id":null,"evidence_quote":"Cited for a lemma on exponential maps and for earlier positive-characteristic non-cancellation results that motivate the field-independent setting."},{"cited_title":"Fieseler, On complex affine surfaces with C+-actions, Comment","cited_arxiv_id":null,"evidence_quote":"Establishes the stable-isomorphism phenomenon for classical Danielewski surfaces that Theorem 3.14 extends to the double surfaces."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Collects the standard facts on Danielewski surfaces and their Makar-Limanov invariants used for comparison and for the exceptional cases."}],"review_version":1}