{"id":"f650c89f-4646-47b8-9cc2-5d4ec237ea2e","arxiv_id":"1908.07891","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Conservative Anosov diffeomorphisms with one-dimensional dominated splittings can be smoothly deformed to realize any strictly majorized Lyapunov spectrum.","lead":"This paper proves that for Anosov diffeomorphisms with a simple dominated splitting, any Lyapunov spectrum satisfying a strict majorization condition can be realized by a smooth volume-preserving deformation. This resolves a central flexibility question in smooth ergodic theory and shows that, on tori, all hyperbolic simple spectra occur.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proof gap in Prop 2.1: condition (iv) of damping perturbations does not follow from Z4⊂U with the stated σ/2 threshold; averaged gaps can be 0, not σ/2.","rationale":"The paper's central claim is Theorem 1.5: from any conservative Anosov diffeomorphism with simple dominated splitting and a strictly majorized target spectrum, a continuous path in Di^r_m(M) can reach the target while staying Anosov with simple dominated splitting. The proof reduces this to Proposition 2.1, whose proof constructs multi-parametric damping perturbations. Our stress-test focused on the internal logic of that proof rather than on scope limitations. The reader's weakest-assumption identification (simple dominated splitting) is a legitimate restriction, but it is an explicit hypothesis, not a flaw. The load-bearing concern we found is a concrete error in the verification that the constructed map f_t satisfies condition (iv) of the damping perturbation definition. The set U is defined with threshold σ/2, but the worst-case averaged gap under that threshold is 0, not σ/2, so the claimed implication fails exactly at the boundary g_u(λ(f)) = σ. Since condition (iv) is essential for Proposition 4.2 (which establishes the Anosov property and the simple dominated splitting of f_t), the central construction is not fully certified as written. The error is minor and easily corrected by sharpening the threshold to σ/4 or by assuming g_u ≥ 2σ; in the actual application of Theorem 1.5, σ is taken as half the true gap, so the iterates have margin. We therefore do not reject the paper, but we recommend conditional acceptance pending the threshold correction. The concrete test isolates the failure mode and confirms the proposed fix, so the concern is actionable and not speculative.","tokens_in":33345,"tokens_out":41891,"duration_ms":365566,"concrete_test":"Verify the claim by taking d=3, u=1, and λ(f) = (σ, −σ, −2σ), so g_1(λ) = σ. Pick a Lyapunov metric with χ_2 = −σ + 0.49σ and χ_3 = −2σ + 0.49σ on a point x (and similarly on its forward orbit), so |χ_j − λ_j| < σ/2. Then the forward average over N−1 iterates has \\barχ_2 − \\barχ_3 ≈ 0.02σ < σ/2, violating condition (iv). Check the proof of Lemma 6.1 and confirm that Z4 ⊂ U does not rule out this scenario. If the threshold is changed to σ/4, the averaged gap becomes ≥ σ − 2(σ/4) = σ/2, and the proof works, confirming the error is repairable.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The threshold mismatch is in the proof of Proposition 2.1 (Section 6). The paper defines R_j := {x : |χ_j(x) − λ_j(f)| ≥ σ/2} and U := M \\setminus ∪_{j,n} f^n(R_j). It then claims in Lemma 6.1 that for the constructed f_t, condition (iv) of the damping-perturbation definition follows from Z4 ⊂ U. However, condition (iv) requires g_u(1/(N−1)∑_{n=1}^{N−1} χ(f^n x)) ≥ σ/2 for x ∈ Z. If x ∈ U, each |χ_j(f^n x) − λ_j| < σ/2, so the j-th average \\barχ_j satisfies |\\barχ_j − λ_j| < σ/2. Therefore \\barχ_j − \\barχ_{j+1} > (λ_j − λ_{j+1}) − σ. Since g_u(λ(f)) ≥ σ only gives λ_j − λ_{j+1} ≥ σ, this lower bound is 0, not σ/2. Example: d=3, u=1, λ(f)=(σ, −σ, −2σ), with χ_2 = −σ + 0.49σ and χ_3 = −2σ + 0.49σ gives average gap < σ/2. Thus condition (iv) can fail, so Lemma 6.1's verification is incorrect. Because Lemma 6.1 feeds into Proposition 4.2, which guarantees each f_t is Anosov with simple dominated splitting, this gap threatens the core construction of Theorem 1.5. The fix is straightforward: replace σ/2 by σ/4 in the definition of R_j (or assume g_u(λ(f)) ≥ 2σ in Proposition 2.1), but as written the proof is incomplete.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves flexibility results for Lyapunov spectra of conservative Anosov diffeomorphisms admitting simple dominated splittings. Theorem 1.5 states that any vector ξ strictly majorized by the Lyapunov spectrum of such an f, and satisfying the sign and strict-gap conditions (a), can be realized as the Lyapunov spectrum of f1 at the end of a continuous path in Di^r_m(M) of conservative Anosov diffeomorphisms with simple dominated splitting. Corollary 1.6 shows that every strictly ordered hyperbolic list of nonzero numbers summing to zero is realized on T^d, and Theorem 1.7 provides a full majorization characterization on T^3. The proofs introduce several new tools: Lyapunov metrics with L1 estimates, Lyapunov charts, damping perturbations, and a multiparameter model deformation, combined with a tower construction and a topological intermediate-value argument.","tokens_in":33737,"tokens_out":25176,"duration_ms":187969,"significance":"Assuming the results are correct, the paper is an important contribution to the flexibility program in smooth dynamics. It resolves the flexibility question for a natural class of hyperbolic systems and identifies majorization as the exact obstruction, complementing known rigidity results. The proof of Theorem 1.5 is constructive and largely self-contained, with explicit quantitative control on the perturbations; the paper also makes a convincing case that the method may extend to broader settings. The exposition is clear and includes useful discussion of the history and open problems.","major_comments":[{"comment":"The verification of condition (iv) in the definition of damping perturbations (Section 4.1) is incomplete. For x in Z4 ⊂ U, the definition of U gives |χ_j(f^n x) − λ_j(f)| < σ/2 for every j and every n in the relevant range, but this only implies that the averaged vector \\barχ = (1/(N−1))∑_{n=1}^{N−1} χ(f^n x) satisfies \\barχ_j − \\barχ_{j+1} > (λ_j(f) − λ_{j+1}(f)) − σ. Since the hypothesis g_u(λ(f)) ≥ σ only gives λ_j(f) − λ_{j+1}(f) ≥ σ, the lower bound is 0, not σ/2 as required by condition (iv). For example, with d=3, u=1, λ(f) = (σ,−σ,−2σ) and pointwise values χ_2 = −1.49σ, χ_3 = −1.51σ on the orbit segment (each within σ/2 of the corresponding λ_j), the averaged gap is 0.02σ < σ/2. Thus condition (iv) can fail, and since this condition is used in the proof of Lemma 4.3 and Proposition 4.2, the conclusion that each f_t is Anosov with simple dominated splitting is not established. The gap is fixable by changing the threshold σ/2 to σ/4 in the definition of R_j (or assuming g_u(λ(f)) ≥ 2σ), but as written the proof is incomplete.","section":"Section 6, Lemma 6.1"},{"comment":"The proof of Proposition 7.4 is only a sketch, stated as 'we mimic the proof of Proposition 2.1 (but with a2 = 0)'. Since Proposition 7.4 is a load-bearing component of the 'if' direction of Theorem 1.7, the authors should provide a full proof. In particular, the degenerate case a2 = 0 requires showing that \\hatλ_2(f_t) remains exactly equal to \\hatλ_2(f) despite the error terms in (6.16); the intended argument presumably uses the preservation of the foliation F^12, but this is not carried out. The verification of the damping-perturbation conditions for the foliated charts also needs to be written in detail.","section":"Section 7.2, Proposition 7.4"}],"minor_comments":[{"comment":"The set denoted \\bar Z in the proof of Lemma 4.3 is not defined; it presumably denotes the closure of the support Z, and this should be stated explicitly.","section":"Section 4.2 (proof of Lemma 4.3)"},{"comment":"In the paragraph introducing the wedge product, 'mutilinear' should be 'multilinear'.","section":"Section 5"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern about condition (iv) is valid and affects the main proof of Theorem 1.5, but the fix is local (change σ/2 to σ/4 in the definition of R_j). I recommend major revision rather than rejection. The authors should also expand the proof of Proposition 7.4, which is currently only a sketch."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bochi–Katok–Rodriguez Hertz prove a global flexibility theorem for Lyapunov spectra of conservative Anosov diffeomorphisms with simple dominated splittings. The main result – Theorem 1.5 – shows that any strictly majorized ordered vector satisfying the gap conditions is realized as the Lyapunov spectrum of some conservative Anosov diffeomorphism in the same component. Corollary 1.6 gives the weak flexibility conjecture on tori for simple spectra. This is a genuine advance over prior local results (Baraviera–Bonatti, Hu–Jiang–Jiang). The multiparameter damping perturbation method is new, and the majorization characterization is clean. The proof of the main theorem is detailed and mostly self-contained, and the paper honestly states its limitations: no C1 bounds on the deformation, and boundary cases with equal exponents are not handled.\n\nI want to flag one issue that a referee should examine. In the proof of Proposition 2.1 (Section 6), Lemma 6.1 verifies that the constructed f_t is a damping perturbation. Condition (iv) of the damping perturbation requires g_u of the averaged expansion vector to be at least sigma/2 on the support. The proof says this follows from Z4 being contained in U, where U is defined using R_j = { |chi_j - lambda_j| >= sigma/2 }. That implication is incorrect. From x in U we only get |average chi_j - lambda_j| < sigma/2, so the gap between consecutive averaged exponents is bounded below only by (lambda_j - lambda_{j+1}) - sigma >= 0, not sigma/2. For instance, with lambda = (sigma, -sigma, -2sigma), averages near -1.49sigma and -1.51sigma give a gap of 0.02sigma. So condition (iv) can fail, and Lemma 6.1 is wrong as written. This is not fatal: replacing sigma/2 by sigma/4 in the definition of R_j (or assuming g_u(lambda(f)) >= 2sigma in Proposition 2.1) repairs the verification. The rest of the construction is unaffected, but the published version should fix this.\n\nThe other soft spot is Proposition 7.4, which is only sketched; it is plausible but deserves a full proof. The citation pattern is fine; self-citations are for standard background.\n\nOverall this is a serious paper that deserves a serious referee. The main theorem is very likely true, and the fix is routine. I recommend acceptance with minor revisions.","headline":"Strong new flexibility theorem for Lyapunov exponents of Anosov diffeomorphisms; the main proof has a small but real gap in Lemma 6.1 that is easy to patch.","tokens_in":34171,"tokens_out":5197,"would_cite":true,"duration_ms":43236,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37D30","37D25","37C40"],"pacs":[],"model":"deepseek-v4-flash","headline":"Strict majorization is the only obstruction to prescribing Lyapunov spectra of Anosov maps.","keywords":["Lyapunov exponents","flexibility","Anosov diffeomorphisms","dominated splitting","majorization","volume-preserving diffeomorphisms","Lyapunov metrics","smooth dynamics"],"falsifier":"In $\\mathbb{T}^3$, take any hyperbolic matrix $L \\in GL(3,\\mathbb{Z})$ with three distinct real eigenvalues and search numerically (for instance by high-precision integration of the derivative cocycle over $C^2$ perturbations of $F_L$) for a conservative Anosov diffeomorphism homotopic to $F_L$ whose top Lyapunov exponent strictly exceeds $\\lambda_1(L)$; Theorem 1.7 says no such map with simple dominated splitting exists, so finding one, or finding any spectrum not majorized by $\\lambda(L)$, would falsify the necessity part.","tokens_in":33193,"feed_emoji":"🌀","tokens_out":7939,"duration_ms":520089,"temperature":0.7,"pith_summary":"This paper establishes that Lyapunov spectra of volume-preserving Anosov diffeomorphisms with one-dimensional dominated splittings are as flexible as majorization allows: any strictly ordered list of numbers with the same sign pattern that is strictly majorized by the current spectrum can be reached continuously (Theorem 1.5). On the torus, this yields a complete statement: every list of strictly ordered nonzero numbers summing to zero is the simple Lyapunov spectrum of some smooth conservative Anosov diffeomorphism (Corollary 1.6). The result pins down majorization as the exact obstacle in dimension three with a fixed homotopy class (Theorem 1.7). The paper also lays out the broader flexibility program, the conjecture that dynamical invariants take all values allowed by general constraints, as the motivation for these theorems.","feed_headline":"Any zero-sum ordered list is a torus Lyapunov spectrum","feed_subtitle":"New proof shows smooth volume-preserving Anosov maps realize every strictly ordered spectrum allowed by majorization.","key_machinery":"The load-bearing object is the pair formed by a simple dominated splitting, a continuous, uniformly contracted and expanded decomposition of the tangent bundle into $d$ one-dimensional subbundles, and the Lyapunov metrics built in Proposition 3.1 by geometric averaging of $N$-step expansion rates; these make each Lyapunov exponent equal to the integral of a single pointwise expansion function that is $L^1$-close to a constant. On top of these metrics, the paper constructs damping perturbations whose support is a tower of small Lyapunov balls with long first return time, and model deformations on the unit ball given by composing rotations of coordinate planes with angle modulated by a bump function. Lemma 5.1, the computational heart, shows that the $j$-th principal minor of the deformation's derivative factors through a single coordinate rotation, so its averaged logarithm is exactly $-Q(t_j)$, a function of the $j$-th parameter alone; formulas (5.9) and (6.16) then transcribe this into an independent, controlled drop of each summed exponent. The proof closes with a topological cube argument (Lemma 2.3) ensuring that the image of the parameter cube covers a whole small box of spectra, which allows the iterative stepping from $\\lambda(f)$ to $\\xi$.","core_discovery":"The paper proves that for a conservative (volume-preserving) Anosov diffeomorphism whose derivative splits into one-dimensional invariant bundles with uniform domination, the Lyapunov spectrum is flexible exactly in the region allowed by the majorization partial order: any strictly ordered list of numbers with the same sign pattern (unstable index) and strictly majorized by the current spectrum can be reached along a continuous path of conservative Anosov diffeomorphisms with simple dominated splitting (Theorem 1.5). On the torus $\\mathbb{T}^d$ this yields the clean statement (Corollary 1.6) that every vector of strictly ordered nonzero numbers summing to zero is the simple Lyapunov spectrum of some $C^8$ conservative Anosov diffeomorphism. For $\\mathbb{T}^3$ with a fixed homotopy class of a linear Anosov automorphism with simple spectrum, the converse also holds: the spectra realizable by diffeomorphisms with simple dominated splitting are exactly those majorized by the linear automorphism's spectrum (Theorem 1.7). Thus majorization, the condition that the target spectrum is obtained from the initial one by a mixing process, is shown to be the precise obstruction in the one-dimensional-splitting setting.","pith_inferences":["Inference: Because the perturbation effect depends only on the minimal gap $\\sigma$ and not on the map $f$, the same scheme should yield uniform flexibility for open sets of conservative Anosov diffeomorphisms with uniformly bounded gap, a useful ingredient for attacking the general weak flexibility conjecture.","Inference: Realizing repeated Lyapunov exponents would require leaving the simple-dominated class, since the construction's independence of parameters relies on one-dimensional bundles; averaging over symmetric spaces instead of norms would move whole blocks of exponents, suggesting a blockwise majorization theory for non-simple spectra.","Inference: In higher-dimensional tori the necessity argument would need a higher-rank analogue of quasi-isometric strong unstable foliations; absent such a tool, Theorem 1.7's characterization suggests that majorization by the linear model is the right conjecture for all homotopy classes.","Inference: A numerical search in $\\mathbb{T}^3$ for conservative Anosov maps homotopic to a linear automorphism $F_L$ whose top exponent exceeds $\\lambda_1(L)$ would either confirm the majorization barrier or reveal a genuinely new phenomenon of non-simple-dominated Anosov diffeomorphisms."],"forward_implications":["Every strictly ordered list of nonzero numbers summing to zero is realized on $\\mathbb{T}^d$ by a $C^8$ conservative Anosov diffeomorphism (Corollary 1.6).","On tori, weak flexibility holds for simple spectra: if the target list contains zero, one takes the product of a realized Anosov map on $\\mathbb{T}^{d-1}$ with an irrational rotation, obtaining ergodicity from mixing.","In the homotopy class of a fully hyperbolic linear automorphism of $\\mathbb{T}^3$, majorization by the linear spectrum is both necessary and sufficient for realization with simple dominated splitting (Theorem 1.7).","The path of diffeomorphisms can be chosen continuous with spectra following any monotone-in-majorization path from $\\lambda(f)$ to $\\xi$ (Remark 2.4).","Crossing a boundary component where the smallest unstable or largest stable exponent vanishes leads out of Anosov into partially hyperbolic maps, so the flexible region naturally borders on partially hyperbolic dynamics (Section 1.7.1)."],"supporting_citations":[{"why":"Supplies the local perturbation method that mixes adjacent Lyapunov exponents and the Jensen-inequality estimate used in Lemma 5.2.","marker":"[6]"},{"why":"Adapts the metric so domination and hyperbolicity are visible in a single iterate, used throughout Sections 3 and 4.","marker":"[23]"},{"why":"Provides the torus structure theory: topological conjugacy of Anosov diffeomorphisms to automorphisms and the entropy inequality (1.3).","marker":"[28]"},{"why":"Establishes quasi-isometry of strong unstable foliations on $\\mathbb{T}^3$, the key fact in the necessity part of Theorem 1.7.","marker":"[15]"},{"why":"Sets up the majorization order and the doubly-stochastic characterization used in the statements of the results.","marker":"[32]"},{"why":"Gives the explicit polynomial construction used in Lemma 7.1 to find linear automorphisms with simple spectra.","marker":"[43]"},{"why":"Provides the cone-invariance criterion that converts forward-invariant cone fields into dominated splittings, used in Proposition 4.2.","marker":"[41]"},{"why":"Supplies absolute continuity of the strong unstable foliation, used in Lemma 7.2 to lower-bound the growth of the top exponent.","marker":"[36]"}],"fun_headline_variants":["Majorization bounds torus Lyapunov spectra","Torus spectra: any zero-sum order works","Flexible Lyapunov spectra on tori","Conservative Anosov maps realize all majorized spectra","Zero-sum spectra for torus diffeomorphisms"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof rests on the Anosov map having a simple dominated splitting, meaning the tangent bundle splits into one-dimensional invariant directions with uniformly separated expansion rates, and this assumption fails as soon as any of the invariant directions has dimension two or more.","fun_headline_variants_meta":{"raw":{"variants":["Majorization bounds torus Lyapunov spectra","Torus spectra: any zero-sum order works","Flexible Lyapunov spectra on tori","Conservative Anosov maps realize all majorized spectra","Zero-sum spectra for torus diffeomorphisms"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000228,"raw_usage":{"total_tokens":1406,"prompt_tokens":808,"completion_tokens":598,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":424,"completion_tokens_details":{"reasoning_tokens":525}},"tokens_in":424,"tokens_out":598,"duration_ms":151953,"temperature":1.0,"reasoning_tokens":525,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:53:03.037483+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"In $\\mathbb{T}^3$, take any hyperbolic matrix $L \\in GL(3,\\mathbb{Z})$ with three distinct real eigenvalues and search numerically (for instance by high-precision integration of the derivative cocycle over $C^2$ perturbations of $F_L$) for a conservative Anosov diffeomorphism homotopic to $F_L$ whose top Lyapunov exponent strictly exceeds $\\lambda_1(L)$; Theorem 1.7 says no such map with simple dominated splitting exists, so finding one, or finding any spectrum not majorized by $\\lambda(L)$, would falsify the necessity part.","supporting_citations":[{"cited_title":"– Removing zero Lyapunov exponents","cited_arxiv_id":null,"evidence_quote":"Supplies the local perturbation method that mixes adjacent Lyapunov exponents and the Jensen-inequality estimate used in Lemma 5.2."},{"cited_title":"– Adapted metrics for dominated splittings","cited_arxiv_id":null,"evidence_quote":"Adapts the metric so domination and hyperbolicity are visible in a single iterate, used throughout Sections 3 and 4."},{"cited_title":"– Introduction to the modern theory of dynamical sys- tems","cited_arxiv_id":null,"evidence_quote":"Provides the torus structure theory: topological conjugacy of Anosov diffeomorphisms to automorphisms and the entropy inequality (1.3)."},{"cited_title":", – Dynamical coherence of partially hyperbolic diffeomorphisms of the 3-torus","cited_arxiv_id":null,"evidence_quote":"Establishes quasi-isometry of strong unstable foliations on $\\mathbb{T}^3$, the key fact in the necessity part of Theorem 1.7."},{"cited_title":"– Inequalities: theory of majorization and its applications","cited_arxiv_id":null,"evidence_quote":"Sets up the majorization order and the doubly-stochastic characterization used in the statements of the results."},{"cited_title":"– On the fractional parts of the powers of a number","cited_arxiv_id":null,"evidence_quote":"Gives the explicit polynomial construction used in Lemma 7.1 to find linear automorphisms with simple spectra."},{"cited_title":"– A (short) survey on dominated splittings","cited_arxiv_id":null,"evidence_quote":"Provides the cone-invariance criterion that converts forward-invariant cone fields into dominated splittings, used in Proposition 4.2."},{"cited_title":"B.; Sinai, Ya","cited_arxiv_id":null,"evidence_quote":"Supplies absolute continuity of the strong unstable foliation, used in Lemma 7.2 to lower-bound the growth of the top exponent."}],"review_version":1}