{"id":"4c4a9584-899e-4949-acd1-bc06e01ba440","arxiv_id":"1908.05083","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For the subgroups K'AN of SO(p,q) acting linearly on R^{p,q}, every orbit outside a p-dimensional degenerate subspace W^p is independent of the choice of K', while orbits on W^p depend on K', and the nilpotent factor N alone never acts with codimension-one orbits.","lead":"The paper works out the orbit structure of a family of symmetry groups acting on pseudo-Euclidean spaces, flat spaces with an indefinite metric where nonzero light-like vectors have zero length. It shows that two different subgroups can produce the same orbit decomposition away from a degenerate subspace but different ones on it, and that the nilpotent factor of the Iwasawa decomposition never acts with codimension-one orbits.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central orbit classification rests on the explicit model of n in Proposition 2.1, whose proof is not self-contained (broken '(??)' reference, 'straightforward computation'). If equations (4) are wrong, the dimension counts behind Theorems 3.1 and 5.6, and hence Proposition 5.7, collapse.","rationale":"The proof strategy of the paper is to compute, for a generic point x, the stabilizer Lie algebra of n or k0⊕a⊕n by explicitly solving the linear equation Xx = 0 using the coordinate model (4). All conclusions about orbit dimensions and the 'exactly two orbits' statements in Theorem 5.6, and the orbit-equivalence/non-equivalence of Proposition 5.7, are downstream of that computation. The model itself is asserted rather than derived: Proposition 2.1's proof says 'straightforward computation' and contains an unresolved '(??)' reference; the root-space equations (5)–(6) are not derived; and the step from (6) to the basis for the short root spaces is not shown. This is not a disagreement with the consensus; it is a verification gap in the central computation. I did find the dimension arithmetic internally consistent: dim(k0+a+n) = [p(p-1)+q(q+1)]/2 matches, and the solved-variable counts in the three cases sum correctly. The manuscript also contains a clearly false count (A has 3^q orbits on the span of the w_j, not 2^{2q}-1) and a garbled display (11), which lower confidence but are not load-bearing. The proposed test—an independent root-space computation, or a small computer-algebra check—would settle whether Proposition 2.1 is correct. Until then, the conditional verdict is appropriate.","tokens_in":19915,"tokens_out":13965,"duration_ms":128030,"concrete_test":"Independently re-derive Proposition 2.1: for the a of (2)–(3), compute the root spaces g_alpha = {X : [H,X] = alpha(H)X for all H in a} for alpha = f_l and f_i ± f_j, using the lexicographic positive system stated in the paper. Check that dim g_{f_l} = p-q, dim g_{f_i±f_j} = 1, and that the direct sum is exactly the set described by (4). A small computer-algebra check, for example with (p,q) = (4,2) or (5,3) and explicit matrices, would settle it. If (4) is reproduced, the main concern is resolved; if not, the orbit classification in Section 5 is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The single most load-bearing assumption is Proposition 2.1: the explicit coordinate description (4) of n is the input for every linear system in the paper. The stabilizer computations in Theorem 3.1 (system (7)) and Proposition 5.1 (system (18)) solve for entries of X using exactly the relations (4); the invariance claims in Remark 3.2 that underlie Corollaries 5.3–5.5 also use this model. If (4) omits a root space, includes an extra relation, or misassigns an entry, the solved-variable counts change and the orbit dimensions p+q-k, q, p-1, l-1 would be wrong. The proof of Proposition 2.1 does not provide the computation: it cites a broken '(??)' cross-reference and says 'by a straightforward computation one gets'. The root-space equations (5) and (6) are displayed without derivation, and the transition from (6) to the claimed basis for the short root spaces is asserted. Although the claims may be true, the central theorem is not independently verifiable as written. A secondary gap is the unsupported 'either/or' elimination in Claim 1 of Theorem 3.1 and Case 1 of Proposition 5.1; the variables being solved are largely distinct, so this is probably fixable, but it should be stated. The incorrect A-orbit count (2^{2q}-1 instead of 3^q) does not affect the main dichotomy but signals that the manuscript is not in a verified state.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies cohomogeneity one isometric linear actions on pseudo-Euclidean space R^{p,q} by subgroups of a maximal parabolic subgroup Q = K0AN, where K0 = SO(p-q), A is a maximal split torus, and N is the nilpotent factor of the Iwasawa decomposition. It claims three main results: (1) the action of N on R^{p,q} is not of cohomogeneity one (Theorem 3.1); (2) for every subgroup K' of K0, the orbits of K'AN on the light cone, on pseudo-spheres, and on pseudo-hyperbolic spaces are explicitly classified, with orbit dimensions that depend only on the position of the point relative to the hyperplanes Pi and Pj (Theorem 5.6 and Corollaries 5.3-5.5); (3) there exist cohomogeneity one actions that are orbit-equivalent on the complement of a p-dimensional degenerate subspace W^p and not orbit-equivalent on W^p (Proposition 5.7). All proofs are direct matrix computations from an explicit coordinate model of the nilpotent subalgebra n given in Proposition 2.1.","tokens_in":19964,"tokens_out":12235,"duration_ms":121139,"significance":"If the computations are correct, the paper gives an explicit classification of orbits for a natural family of noncompact, generally nonproper actions, generalizing the q = 1 results of Berndt-Diaz-Ramos-Vanaei in [7] and exhibiting a precise phenomenon of orbit-independence of the compact factor K'. This is a useful contribution to the study of cohomogeneity one actions in indefinite signature. The assumptions are not fitted to data and no parameters are introduced ad hoc; the methods are elementary but potentially verifiable. However, the central claims are not independently verifiable as written because the explicit model of n in Proposition 2.1 is incomplete and its proof is not supplied. The reader's assessment that the dimension counts are internally consistent once (4) is assumed is accurate, but internal consistency of the counts does not establish the correctness of the parametrization, which is load-bearing for every subsequent statement.","major_comments":[{"comment":"The 'if and only if' characterization of n in (4) is incomplete. As displayed, it lists relations between entries of A, B, and D, but does not state that all entries not appearing in these relations vanish. For p=3, q=1 the system (4) imposes only two independent constraints on so(3,1), leaving four free parameters, while the paper's own next sentence asserts dim N = q(p-1) = 2. Thus the displayed parametrization contradicts the dimension of N it is supposed to justify. Since every later stabilizer computation, including systems (7), (18), and the invariance claims of Remark 3.2, uses this model of n, this is a load-bearing error that must be corrected.","section":"§2, Proposition 2.1, equations (4)"},{"comment":"The proof of Proposition 2.1 is not self-contained: it says the root spaces are obtained 'by a straightforward computation', cites a broken cross-reference '(??)', and passes from equations (6) to the claimed bases purely by assertion. This matters because completeness of the root space description is exactly what the rest of the paper needs. Either a full derivation should be provided, or the explicit matrix form of the restricted root spaces should be quoted from a source where it is proved.","section":"§2, proof of Proposition 2.1"},{"comment":"The stabilizer dimension counts rely on an unsupported elimination argument. In Claim 1 of Theorem 3.1 and again in Claim 1 of Proposition 5.1, when x_{p-k+1} is nonzero the text says one may use 'either' the equation expressing A_{p-k+1,j} 'or' the one expressing D_{p-j+1,k}, and then counts the eliminated variables. The rank of the relevant submatrix of (7) and of (18) is not computed, and the two branches are not shown to produce the same rank. The final orbit dimensions p+q-k and p+q-(k+1) depend on these counts, so the proof should state the rank and justify the number of independent linear conditions explicitly.","section":"§3, Theorem 3.1, Claim 1; §5, Proposition 5.1, Claim 1"}],"minor_comments":[{"comment":"The count 'A has 2^{2q}-1 orbits on ⊕Rw_j' is incorrect. The action of R_+^q by coordinate-wise scaling has orbits classified by the sign pattern of (r_1,...,r_q), including zeros, so there are 3^q orbits, not 2^{2q}-1; the listed binomial coefficients C(2q,k) are also inconsistent with this count.","section":"§5, cylinder discussion before Corollary 5.5"},{"comment":"The abstract contains the typo 'on $\\mathbb{W}p$'; this should be '$\\mathbb{W}^p$'.","section":"Abstract"},{"comment":"The dangling cross-reference '(??)' in the proof of Proposition 2.1 should be removed or replaced with an actual equation number.","section":"§2, Proposition 2.1 proof"},{"comment":"The reference list has formatting problems: the entry '[PV]' is unnumbered, and the stray 'item[[N]]' before [12] should be cleaned up.","section":"References"},{"comment":"Equation (11) introduces the substitutions k=q-j and l=q-i without defining them before use; this makes the displayed formula hard to read and should be clarified.","section":"§3, Remark 3.2, equation (11)"},{"comment":"The typo 'Proof of Calim 1' should be corrected to 'Claim 1'.","section":"§5, Proposition 5.1, Claim 1"}],"recommendation":"major_revision","confidential_remarks":"The central orbit classification is not in a verifiable state because the explicit model of n in Proposition 2.1 is incomplete and its proof is a reference to a broken cross-reference. The orbit-equivalence phenomenon described in Proposition 5.7 is potentially interesting, but the manuscript needs a full rewrite of Section 2 and a rigorous rank computation for the elimination arguments before it can be considered further. I see no indication of deliberate misrepresentation; the issues appear to be technical gaps that may be fixable within the paper's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real news: this paper extends Berndt--Diaz-Ramos--Vanaei's q=1 work to general q, and the main orbit-dimension dichotomy is plausible. I spot-checked the central counts--dim(k0+a+n), the elimination counts in Theorem 3.1 and Proposition 5.1, the orbit dimensions p+q-k, q, p-1, l-1--and they are internally consistent. The three-case partition of R^{p,q} is exhaustive, and the punchline of Theorem 5.6/Proposition 5.7, orbit-equivalence off the degenerate W^p and dependence on K' on the cylinder inside W^p, is a real structural observation, not a restatement of the q=1 literature. Credit where due: that is new work, and the broad strategy is sound.\n\nNow the soft spots, in proportion. The load-bearing Proposition 2.1, the explicit model of the nilpotent subalgebra n, is not proven: it says \"by a straightforward computation one gets\" and cites a broken cross-reference \"(??)\". Every later linear system--(7), (18), the invariance claims in Remark 3.2--assumes equations (4) are complete and correct. I think they probably are, and the paper's internal consistency supports that, but as posted the central theorem is not independently verifiable. That is a real gap, not a cosmetic one. The \"either/or\" branch choices in Claim 1 of Theorem 3.1 and Case 1 of Proposition 5.1 also skip the independence check; that is probably fixable, but it should be stated.\n\nOther issues are minor or textual but signal lack of polish: the A-orbit count on the span of the w_i is given as 2^{2q}-1 with binomial coefficients, which is simply wrong (the correct number is 3^q, one per sign pattern); display (11) is garbled; Remark 2.2 refers to figures that are absent; the introduction says the N-action is \"of cohomogeneity two\" which is stronger than what Theorem 3.1 proves; and the bibliography has malformed entries. None of these by itself kills the main dichotomy, but together they make the manuscript unreliable as a citable source in its current form.\n\nWho is this for? Specialists in cohomogeneity one actions on indefinite flat spaces. They will get the core idea quickly and will want to know whether the orbit classification is correct. It deserves a serious referee, but only after the authors supply the missing computation in Proposition 2.1 and clean up the text. I would send it to peer review with a clear request for major revision, and I would not cite it in its present state.","headline":"Genuine extension of the q=1 Minkowski classification to all p>q, with dimension counts that mostly check out, but the posted version leans on an unverified parametrization of the nilpotent algebra and has enough textual corruption that it needs major revision before it is refereeing-ready.","tokens_in":20830,"tokens_out":1107,"would_cite":false,"duration_ms":12881,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["57S25","53C30"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that the subgroups K'AN of a maximal parabolic subgroup of SO(p,q) all have the same orbit decomposition on R^{p,q} minus the degenerate p-plane W^p, while their orbit structures on W^p differ.","keywords":["Cohomogeneity one","Isometric action","Pseudo-Euclidean space","Iwasawa decomposition","Nilpotent factor","Orbit equivalence","Maximal parabolic subgroup","Nullcone"],"falsifier":"For a small concrete case, say $p=4,q=2$, write out the stabilizer equations (7) using the parameterization (4) for a point with $x_4+x_5\\neq 0$ and compute the rank of the stabilizer of $N$. If the orbit dimension is not $p+q-2=4$, Theorem 3.1 fails; similarly, for $K_0AN$, checking whether the stabilizer rank matches the stated dimension $p+q-1$ would test Proposition 5.1 and hence the orbit classification.","tokens_in":19412,"feed_emoji":"📐","tokens_out":12155,"duration_ms":113982,"temperature":0.7,"pith_summary":"This paper studies isometric actions on pseudo-Euclidean space $\\mathbb{R}^{p,q}$ that have an orbit of codimension one, the cohomogeneity one case. It proves that the nilpotent factor $N$ of an Iwasawa decomposition of $SO(p,q)$ never acts with cohomogeneity one: every orbit has dimension at most $p+q-2$. For the larger groups $K'AN$, with $K'$ any subgroup of $K_0\\simeq SO(p-q)$, it classifies the orbits and finds a sharp division: outside the $p$-dimensional degenerate subspace $W^p$ every choice of $K'$ produces exactly the same orbit decomposition, while on $W^p$ the orbit structures differ according to the orbits of $K'$ on a sphere. If correct, this yields cohomogeneity one actions in indefinite signature that are orbit-equivalent on a dense open set yet not globally, extending the known $q=1$ Minkowski-space case to all $q$.","feed_headline":"Orbit structures agree outside a degenerate plane, differ on it","feed_subtitle":"On pseudo-Euclidean space, varying the compact factor changes orbits only inside that plane.","key_machinery":"The load-bearing object is the explicit coordinate model of the nilpotent subalgebra $n$ in Proposition 2.1, equations (4), coming from the Iwasawa decomposition $so(p,q)=k\\oplus a\\oplus n$. It writes every element of $n$ through three families of linear relations among the blocks $A,B,D$ of a matrix in $so(p,q)$, and it converts the stabilizer condition $Xx=0$ into the linear systems (7) and (18). Counting how many variables those systems eliminate gives the orbit dimensions $p+q-k$, $p+q-(k+1)$, $p-1$, $q$, and $l-1$ in Theorem 3.1 and Proposition 5.1. The basis in Remark 3.2 then shows the nilpotent action fixes $w_1=e_p-e_{p+1}$ and translates the span of the $w_i$, which is what makes the cylinder $W^p\\cap S^{p-1,q}(r)=S^{p-q-1}(r)\\times \\mathbb{R}^q$ behave differently from the rest of the space.","core_discovery":"The headline result is Proposition 5.7: for $p>q+1>2$ there exist cohomogeneity one isometric actions on $\\mathbb{R}^{p,q}$ that are orbit-equivalent on the complement of a $p$-dimensional degenerate subspace $W^p$ and not orbit-equivalent on $W^p$. The actions are the natural linear actions of $G=K'AN$, where $K'\\subseteq K_0=SO(p-q)$, $A$ is the abelian part and $N$ the nilpotent part of the Iwasawa decomposition used to build a maximal parabolic subgroup. Theorem 5.6 describes every orbit: off $W^p$ the orbits are exactly the connected components of the hyperplane strata $\\bigcap_{i=0}^{p+q-m-1}\\Pi_i \\smallsetminus \\bigcap_{i=0}^{p+q-m}\\Pi_i$ inside the nullcone, a pseudo-sphere $S^{p-1,q}(r)$, or a pseudo-hyperbolic space $H^{p,q-1}(r)$, with $m\\in\\{p,\\ldots,p+q-1\\}$, and these do not depend on $K'$. On $W^p$ the relevant part of a pseudo-sphere is the cylinder $S^{p-q-1}(r)\\times \\mathbb{R}^q$, and the orbit through $y+\\sum_{j=1}^q r_j w_j$ is $K'(y)\\times \\bigoplus_{j=1}^q \\mathbb{R} w_j$, which changes with $K'$; for $K'=K_0$ the cylinder is a single orbit. Thus the paper establishes a concrete family of actions whose orbit maps agree on the complement of a degenerate subspace and disagree exactly on it.","pith_inferences":["Inference: because all $K'AN$ actions coincide off $W^p$, the orbit decomposition of the complement can be treated as an $AN$-orbit space; the compact factor $K'$ only decides how the cylinder $W^p\\cap S^{p-1,q}(r)$ is subdivided.","Inference: the same mechanism may occur for other parabolic subgroups: degenerate subspaces spanned by lightlike directions could be the only loci where additional compact symmetries change orbit decompositions.","Inference: the dimension formulas suggest a combinatorial invariant—the first index $k$ with $x_{p-k+1}+x_{p+k}\\neq 0$ fixes the orbit dimension—which could yield the full orbit space as a cone over a stratified sphere.","Inference: a direct computer-algebra check of the stabilizer ranks for small $p,q$, say $p=4,q=2$, would test whether the linear eliminations are independent and whether the stated counts $p+q-k$ and $p+q-(k+1)$ are correct."],"forward_implications":["The nilpotent factor $N$ alone never acts with cohomogeneity one on $\\mathbb{R}^{p,q}$: its largest orbits have dimension $p+q-2$, one less than codimension one.","For every $K'\\subseteq K_0$, the group $K'AN$ acts with cohomogeneity one, and on $\\mathbb{R}^{p,q}\\setminus W^p$ all such actions have the same orbit decomposition.","Off $W^p$, the orbits are exactly the connected components of the hyperplane strata $\\bigcap_{i=0}^{p+q-m-1}\\Pi_i\\setminus\\bigcap_{i=0}^{p+q-m}\\Pi_i$ inside the nullcone, pseudo-spheres, and pseudo-hyperbolic spaces, with two orbits for each dimension $m\\in\\{p,\\ldots,p+q-1\\}$.","On the cylinder $W^p\\cap S^{p-1,q}(r)=S^{p-q-1}(r)\\times\\mathbb{R}^q$, the orbit through $y+\\sum r_j w_j$ is $K'(y)\\times\\bigoplus \\mathbb{R}w_j$, so different subgroups $K'$ give genuinely different orbit decompositions there.","When $K'=K_0=SO(p-q)$, the cylinder becomes a single orbit, so the full group $K_0AN$ is transitive on it; smaller $K'$ produce finer decompositions."],"supporting_citations":[{"why":"The Minkowski-space cohomogeneity-one orbit classification that Theorem 5.6 and Proposition 5.7 generalize.","marker":"[7]"},{"why":"Supplies the Iwasawa decomposition and restricted-root space facts used to parameterize the nilpotent algebra $n$.","marker":"[9]"},{"why":"Introduced cohomogeneity-one actions and gave the orbit-space classification for compact groups.","marker":"[12]"},{"why":"Provides the nullcone, pseudo-sphere, and pseudo-hyperbolic space models used to identify the $SO(p,q)$ orbits.","marker":"[13]"},{"why":"Gives the orbit-space theorem for proper cohomogeneity-one actions that frames the indefinite nonproper setting.","marker":"[5]"}],"fun_headline_variants":["Orbits match outside a degenerate plane, diverge inside it","Orbits agree off a degenerate subspace, differ on it","Orbit equivalence breaks exactly on a degenerate plane","Orbits coincide off a degenerate plane, not on it"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The explicit coordinate description of the nilpotent subalgebra $n$ in Proposition 2.1, equations (4), is complete and correct, and the linear eliminations used to count stabilizer dimensions are independent; if either of these fails, the orbit dimension counts and the whole classification collapse.","fun_headline_variants_meta":{"raw":{"variants":["Orbits match outside a degenerate plane, diverge inside it","Orbits agree off a degenerate subspace, differ on it","Orbit equivalence breaks exactly on a degenerate plane","Orbits coincide off a degenerate plane, not on it"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000554,"raw_usage":{"total_tokens":2694,"prompt_tokens":1053,"completion_tokens":1641,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":669,"completion_tokens_details":{"reasoning_tokens":1575}},"tokens_in":669,"tokens_out":1641,"duration_ms":12122,"temperature":1.0,"reasoning_tokens":1575,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:27:16.693813+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a small concrete case, say $p=4,q=2$, write out the stabilizer equations (7) using the parameterization (4) for a point with $x_4+x_5\\neq 0$ and compute the rank of the stabilizer of $N$. If the orbit dimension is not $p+q-2=4$, Theorem 3.1 fails; similarly, for $K_0AN$, checking whether the stabilizer rank matches the stated dimension $p+q-1$ would test Proposition 5.1 and hence the orbit classification.","supporting_citations":[{"cited_title":"Berndt, J","cited_arxiv_id":null,"evidence_quote":"The Minkowski-space cohomogeneity-one orbit classification that Theorem 5.6 and Proposition 5.7 generalize."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Iwasawa decomposition and restricted-root space facts used to parameterize the nilpotent algebra $n$."},{"cited_title":"Mostert, On a compact Lie group acting on a manifold, Ann","cited_arxiv_id":null,"evidence_quote":"Introduced cohomogeneity-one actions and gave the orbit-space classification for compact groups."},{"cited_title":"O’Neill, Semi-Riemannian geometry with applications to relativity, Pure and Applied Mathematics 103 1st ed","cited_arxiv_id":null,"evidence_quote":"Provides the nullcone, pseudo-sphere, and pseudo-hyperbolic space models used to identify the $SO(p,q)$ orbits."},{"cited_title":"Berard-Bergery, Sur de nouvells vari ´et´e riemanniennes d’Einstein,Inst","cited_arxiv_id":null,"evidence_quote":"Gives the orbit-space theorem for proper cohomogeneity-one actions that frames the indefinite nonproper setting."}],"review_version":1}