{"id":"aafda441-0ac5-482f-b2d8-8b223d10b51d","arxiv_id":"2508.18257","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For unions of subsets of k-planes, the paper proves packing dimension bounds analogous to known Hausdorff dimension results, and for hyperplanes proves an exact preservation of packing dimension under full-dimension extension.","lead":"This math paper proves new lower bounds on the packing dimension of unions of k-dimensional planes and shows that extending pieces of hyperplanes to full hyperplanes does not increase packing dimension. The proofs use a new notion of effective dimension on the space of planes, importing tools from algorithmic information theory into geometric measure theory.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The main hyperplane results (Theorems 5 and 6) depend on Proposition 27, whose proof invokes Lemma 29 without proof and contains a sign error in the reduction from (32); the gap appears repairable but must be closed.","rationale":"The reader's CONDITIONAL verdict is appropriate. Theorems 1-4 have detailed proofs and are supported by the machinery of Sections 2-4; the effective-dimension framework on Grassmannians is developed carefully, and Lemma 22, Corollary 23, and Lemma 24 are proved. The paper is honest about the omission of Lemma 29 and the condensation of Proposition 27. My own read confirms that the omitted lemma is the load-bearing pillar for Theorem 6: Proposition 27 is stated as a black box, and both hyperplane theorems are derived from it in a few lines. I found an additional sign error in the proof of Proposition 27, but a corrected sign appears to be sufficient, so this is a typographical and verification issue rather than a fatal flaw. No machine-checked proof or code is supplied, and the paper does not claim one. Therefore the correct disposition is to require a complete proof of Lemma 29 and a clean, corrected proof of Proposition 27 before acceptance; this is exactly the existing CONDITIONAL verdict. No change to the reader's verdict is needed.","tokens_in":31465,"tokens_out":17743,"duration_ms":154153,"concrete_test":"Fully write out the proof of Lemma 29 by adapting Lemma 6 of [25] to hyperplanes, then re-derive the displayed inequality in Proposition 27 after the application of Lemma 30 using (32) with the sign corrected; verify that condition (2) of Lemma 29 with δ=√ε holds for every n≥2. If the corrected derivation requires an extra geometric hypothesis, or if the modified Lemma 29 has a different constant, then Theorem 6 is not established by the current text.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Section 6.2, Lemma 29 is stated and then dismissed with 'This is just Lemma 6 in [25] modified for hyperplanes. Since the proof of this lemma is essentially the same as the proof in [25]; we omit it.' Proposition 27 is the sole bridge to Theorems 5 and 6, and its proof applies Lemma 29 at the critical enumeration step after (31)-(32). The hyperplane modification is not a purely cosmetic relabelling: the parameter (a,b) has n coordinates and the admissible perturbations (u,v) are constrained by u·x+v=a·x+b, so the line-based proof of Lemma 6 must be reworked to track this constraint. If Lemma 29 fails, Proposition 27 fails, and with it both Theorem 5 and Theorem 6. The condensed proof of Proposition 27 also has a sign issue: after applying Lemma 30 the text claims K^{A,D}_r(u,v) ≥ K^A_r(a,b)+(1−√ε)(r−t)+K^{A,a,b}_{r−t}(x)−(n−2)(r−t), but (32) gives K^{A,D}_t(a,b) ≥ K^{A,D}_r(a,b)−(1−√ε)(r−t), not plus. A corrected minus sign still appears sufficient, so this is likely a typo, but it further shows the technical core has not been independently checked.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops tools from algorithmic information theory to bound the packing dimension of unions of subsets of k-planes and of extensions of such subsets to full k-planes. It introduces effective dimension on the Grassmannian and affine Grassmannian, proves point-to-set principles and a symmetry-of-information result in that setting, and uses these to obtain Furstenberg-type bounds (Theorems 1 and 2), extension bounds (Theorem 3 and Proposition 4), and sharper hyperplane results (Theorems 5 and 6). The central new claims are an analog of Héra's Hausdorff-dimension bound for packing dimension, a generalization of Fraser's line result, a hyperplane extension equality, and a packing-dimension bound for unions of full-dimension subsets of hyperplanes that improves the general bound in the case k=n-1.","tokens_in":31688,"tokens_out":10249,"duration_ms":100156,"significance":"If correct, the paper would make a real contribution: Theorem 1 gives a packing-dimension analog of Héra's bound, Theorem 2 generalizes Fraser's line result to k-planes, and the Grassmannian effective-dimension machinery in Section 3 is a genuinely new and reusable tool. Theorems 5 and 6 are the headline results for hyperplanes, and Theorem 6's bound n-1+nt/((n-1)t+n) improves the general bound for all n and 0<t<n. The development is mostly careful and self-contained for Sections 3-5. However, the hyperplane results rest on Proposition 27, whose proof is supported by an unproved lemma (Lemma 29), a sign error in a displayed inequality, and an incorrect statement about projections in the proof of Lemma 30. These issues are load-bearing but appear repairable, so the paper merits revision rather than rejection.","major_comments":[{"comment":"Lemma 29 is stated without proof; the text explicitly says 'This is just Lemma 6 in [25] modified for hyperplanes. Since the proof of this lemma is essentially the same as the proof in [25]; we omit it.' This lemma is the enumeration step on which Proposition 27, and therefore Theorems 5 and 6, rest. The passage from lines to hyperplanes is not purely notational: the constraint u·x+v=a·x+b couples the parameters (u,v) to x, and the proof must control both the precision t=−log|(a,b)−(u,v)| and the family of hyperplanes through the point (x,a·x+b). Please include a complete proof, or a precise step-by-step reduction to [25, Lemma 6], rather than an omission.","section":"§6.2, Lemma 29"},{"comment":"After applying Lemma 30 and (32), the manuscript asserts K^{A,D}_r(u,v) ≥ K^A_r(a,b)+(1−√ε)(r−t)+K^{A,a,b}_{r−t}(x)−(n−2)(r−t)−O(log r). This sign is inconsistent with (32), which gives K^{A,D}_t(a,b) ≥ K^{A,D}_r(a,b)−(1−√ε)(r−t). The displayed inequality should contain a minus sign in front of (1−√ε)(r−t). As written, the subsequent deduction that condition (2) of Lemma 29 holds is not supported. Please correct the sign and re-verify the lower bound.","section":"§6.2, proof of Proposition 27"},{"comment":"The proof of (29) uses the claim 'Projection onto V is the composition of projection onto H1 and projection onto H2.' This is false in general: for two hyperplanes in R^n, the composition of the two projection operators has rank n−1 (not n−2) and does not agree with the orthogonal projection onto their intersection even approximately, as can be seen already in R^3 for H1=span(e1,e2) and H2=span(e1,e2+e3). Consequently the estimate ρ(V, ¯V) ≤ O(2^{-r}) is not justified. This step is essential for (29). Please replace it with a correct computation of the intersection projection from the hyperplane data (for example, by solving the two linear equations defining the hyperplanes) and re-derive the error bound.","section":"§6.2, proof of Lemma 30"}],"minor_comments":[{"comment":"Condition (1) should read 1/2 ≤ |q_i| ≤ 2 rather than |q|.","section":"§3.2, Lemma 14"},{"comment":"The notation K^A_{r,t}(u,v|u,v) in the first displayed equation of the proof appears to be a typo; as written it is unclear what conditioning on (u,v) at precision t means in the self-referential identity. Please clarify or correct.","section":"§6.2, proof of Lemma 30"},{"comment":"The pointwise inequality (20) is asserted for every r sufficiently large; the passage from the two displayed facts to n(r−c_r)+K^A_{c_r}(x,a,b) ≥ K^A_{r,t}(x, a·x+b) omits an O(log r) term. Please make the error terms explicit so that the claimed constant C is independent of x.","section":"§6.1, proof of Theorem 5"},{"comment":"In the sentence 'This is a packing dimension analog of a result of Héra [15], who proved the same bound but with both instances of packing dimension replaced by Hausdorff dimension,' the word 'both' is inaccurate because the bound involves packing dimension only in the hypothesis on the set of planes and the conclusion on the union; please rephrase.","section":"§1.1"}],"recommendation":"major_revision","confidential_remarks":"The paper's Section 6.2 is the main concern: Lemma 29 is omitted, the proof of Proposition 27 contains a sign error, and the proof of Lemma 30 contains a false statement about composition of projections. These problems are technical and likely repairable, but they mean the paper's headline hyperplane theorems are not currently established by the written proof. The rest of the paper (Sections 3-5) appears solid and valuable. I recommend major revision and a careful re-verification of the hyperplane arguments."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a real contribution to the packing-dimension side of the Furstenberg/Kakeya program. Theorem 1 is the packing-dimension analog of Hera's bound, Theorem 2 extends Fraser's lines result to k-planes, and Theorem 5 gives a clean equality for hyperplane extensions. The effective dimension on the Grassmannian and affine Grassmannian in Section 3 is genuinely new infrastructure, with point-to-set principles included. The paper is also honest about what is imported and what is new. The soft spot is concentrated in Section 6.2. Proposition 27 is the sole bridge to Theorems 5 and 6, and its proof is condensed. Lemma 29 is stated as a modified version of Lemma 6 from [25] and its proof is omitted; the modification is not purely cosmetic, since hyperplanes in R^n carry the constraint u·x+v=a·x+b, and the line-based proof needs real reworking. Lemma 30 gets a proof sketch that reduces to the two-dimensional case, which is plausible but not fully checked. There is also a likely sign error: after applying Lemma 30, the text has K^{A,D}_r(u,v) ≥ K^A_r(a,b)+(1−√ε)(r−t)+..., while (32) gives K^{A,D}_t(a,b) ≥ K^{A,D}_r(a,b)−(1−√ε)(r−t), so the sign should be minus. The corrected inequality still appears sufficient, so this is probably a typo, but it is a symptom of a technical core that has not been independently verified. Similarly, the definition of d_r in the proof of Proposition 27 looks garbled. None of this is obviously fatal; the strategy is coherent and the two-dimensional case is already in print. The citation pattern is fine: the self-citations to [5] are prior published work, and the differences are acknowledged. No circularity. Who this is for: anyone working on packing-dimension versions of Furstenberg-type problems, and anyone who wants a working notion of effective dimension for spaces of planes. It deserves a serious referee. The referee should ask for a full proof of Lemma 29 and an expanded, corrected proof of Proposition 27 before acceptance, but the main results are new and the framework is valuable. I would bring it to a reading group.","headline":"Genuinely new packing-dimension results for unions and extensions of k-planes, anchored by a useful effective-dimension framework on Grassmannians; the hyperplane theorems rest on an under-verified Proposition 27 that needs a full proof.","tokens_in":687,"tokens_out":1128,"would_cite":true,"duration_ms":37165,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["28A78","28A80","68Q30"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper shows that extending every hyperplane that meets a set in full Hausdorff dimension to the whole hyperplane does not increase packing dimension, and proves an improved lower bound for unions of such hyperplane slices.","keywords":["packing dimension","Hausdorff dimension","k-planes","affine Grassmannian","effective dimension","Kolmogorov complexity","Furstenberg-type sets","hyperplane extensions"],"falsifier":"Exhibit, for some $n$ and $0<t<n$, a family $\\mathcal{P}\\subseteq A(n,n-1)$ with packing dimension $t$ and full Hausdorff-dimensional subsets whose union $F$ satisfies $\\dim_P(F)<n-1+\\frac{nt}{(n-1)t+n}$; such a construction would refute Theorem 6. Alternatively, a set $E$ whose full-hyperplane-slice extension has $\\dim_P(F)>\\dim_P(E)$ would refute Theorem 5.","tokens_in":31198,"feed_emoji":"📐","tokens_out":12381,"duration_ms":114178,"temperature":0.7,"pith_summary":"Fractal size can be measured in several ways; packing dimension records the largest scales at which a set looks big, in contrast to Hausdorff dimension, which records the smallest scales. The paper asks how large a union of subsets of $k$-planes must be when the family of planes has packing dimension $t$ and each subset has dimension at least $s$, and also how much packing dimension can grow when such subsets are fattened to entire planes. Using a new effective dimension on the spaces of $k$-planes, it proves bounds for general $k$ and two stronger hyperplane results: equality under full-dimensional hyperplane extension, and an improved union bound for full-dimensional hyperplane slices.","feed_headline":"Full hyperplane slices cannot inflate packing dimension","feed_subtitle":"For hyperplane families of packing dimension t, the union of full slices has packing dimension at least $n-1+nt/((n-1)t+n)$.","key_machinery":"The machinery is effective dimension defined on the Grassmannian $G(n,k)$ and the affine Grassmannian $A(n,k)$, meaning Kolmogorov complexity at precision $r$ of rational orthogonal-projection matrices that approximate a plane. A point-to-set principle on these computable metric spaces converts per-point complexity bounds into classical packing-dimension bounds. The load-bearing pieces are Lemma 22, which shows a $k$-plane is computable from approximations of $k+1$ points on it; Corollary 23, which gives the orthogonal complement; Lemma 24, which bounds the complexity of points on a known plane by $kr+O(\\log r)$; and Lemma 30, which reduces the hyperplane intersection estimate to a two-dimensional statement.","core_discovery":"The paper's central claim is that a single pointwise algorithmic inequality drives both hyperplane theorems. For a hyperplane written as $(x_1,\\dots,x_{n-1})\\mapsto a\\cdot x+b$, if $x$ has effective dimension at least $n-1-\\varepsilon/4$ relative to $(a,b)$, then for every precision $r$, $$K^A_r(x,a\\cdot x+b)\\ge (n-1)r+K^A_{c_r}(a,b)+(r-c_r)-C\\sqrt{\\varepsilon r},$$ where $c_r$ is the precision that maximizes $K^A_t(a,b)-t$ over $t\\le r$. Feeding this inequality into the point-to-set principle yields the extension equality $\\dim_P(F)=\\dim_P(E)$ and the improved hyperplane union bound $\\dim_P(F)\\ge n-1+\\frac{nt}{(n-1)t+n}$.","pith_inferences":["Beyond the paper's claims, the same effective-dimension machinery should apply to other families of geometric objects with computable rational approximations, such as spheres or algebraic submanifolds, yielding analogous union bounds.","The paper leaves open whether the gap between its packing-dimension bound and the known Hausdorff-dimension bound for hyperplane unions is an artifact of the proof; constructing sharp examples for the packing case would settle this.","The proof's use of oracle-dependent precision plateaus suggests that the improvement in Theorem 6 comes from exploiting periods where the complexity of the parameter $(a,b)$ grows slowly; varying the oracle construction may yield further refinements."],"forward_implications":["If Theorem 5 is correct, the packing dimension of a set is unchanged when all hyperplanes meeting it in Hausdorff dimension $n-1$ are extended to full hyperplanes, so full-dimensional slices cannot inflate packing dimension.","If Theorem 6 is correct, the general union bound $n-1+t/n$ for hyperplanes is improved to $n-1+\\frac{nt}{(n-1)t+n}$ for every $n$ and every $0<t<n$.","For general $k$, the Hausdorff-slice and packing-slice versions of the union problem have different bounds, reflecting that the scales on which the family of planes is large need not align with the scales on which the slices are large.","For $k$-planes, extending all positive-measure slices cannot push packing dimension above $2\\dim_P(E)-k$; for lines, the same holds even when only full Hausdorff-dimensional slices are extended."],"supporting_citations":[{"why":"It establishes the point-to-set principle that converts effective dimension bounds into classical Hausdorff and packing dimension bounds.","marker":"[22]"},{"why":"It extends the point-to-set principle to computable metric spaces, which the paper applies to the Grassmannian and affine Grassmannian.","marker":"[23]"},{"why":"It supplies the symmetry-of-information lemma, the two-dimensional intersection lemma used to prove Lemma 30, and the lemma that Lemma 29 adapts to hyperplanes.","marker":"[25]"},{"why":"It contains the line-extension result and proof template that Theorem 5 generalizes, as well as the comparison point for Theorem 6.","marker":"[5]"},{"why":"It provides the Hausdorff-dimension union bound of which Theorem 1 is the packing analog, together with sharpness examples.","marker":"[15]"},{"why":"It gives the packing-dimension line result, and the counterexample showing the pure packing version fails, that Theorem 2 generalizes.","marker":"[11]"},{"why":"It gives the Hausdorff union and extension bounds for affine subspaces that the hyperplane results are packing analogs of.","marker":"[14]"},{"why":"It contains the positive-measure hyperplane extension result generalized by Theorem 3.","marker":"[10]"}],"fun_headline_variants":["Packing dimension resists hyperplane unions","Hyperplane extension keeps packing dimension fixed","Algorithmic tools tighten hyperplane packing bounds","New bound for unions of hyperplane slices","Effective dimension improves hyperplane union results"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The strongest hyperplane theorems rest on Lemma 29, a stated-but-unproved adaptation of a two-dimensional lemma; the author says the proof is essentially the same and omits it, so the main results' support depends on that adaptation being valid.","fun_headline_variants_meta":{"raw":{"variants":["Packing dimension resists hyperplane unions","Hyperplane extension keeps packing dimension fixed","Algorithmic tools tighten hyperplane packing bounds","New bound for unions of hyperplane slices","Effective dimension improves hyperplane union results"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000628,"raw_usage":{"total_tokens":2857,"prompt_tokens":849,"completion_tokens":2008,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":465,"completion_tokens_details":{"reasoning_tokens":1944}},"tokens_in":465,"tokens_out":2008,"duration_ms":15286,"temperature":1.0,"reasoning_tokens":1944,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T16:57:38.349613+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit, for some $n$ and $0<t<n$, a family $\\mathcal{P}\\subseteq A(n,n-1)$ with packing dimension $t$ and full Hausdorff-dimensional subsets whose union $F$ satisfies $\\dim_P(F)<n-1+\\frac{nt}{(n-1)t+n}$; such a construction would refute Theorem 6. Alternatively, a set $E$ whose full-hyperplane-slice extension has $\\dim_P(F)>\\dim_P(E)$ would refute Theorem 5.","supporting_citations":[{"cited_title":"Lutz and Neil Lutz, Algorithmic information, plane Kakeya sets, and conditional dimension, ACM Trans","cited_arxiv_id":null,"evidence_quote":"It establishes the point-to-set principle that converts effective dimension bounds into classical Hausdorff and packing dimension bounds."},{"cited_title":"Lutz, Neil Lutz, and Elvira Mayordomo, Extending the reach of the point-to-set principle, Information and Computation 294 (2023), 105078","cited_arxiv_id":null,"evidence_quote":"It extends the point-to-set principle to computable metric spaces, which the paper applies to the Grassmannian and affine Grassmannian."},{"cited_title":"Stull, Bounding the dimension of points on a line , Inform","cited_arxiv_id":null,"evidence_quote":"It supplies the symmetry-of-information lemma, the two-dimensional intersection lemma used to prove Lemma 30, and the lemma that Lemma 29 adapts to hyperplanes."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It contains the line-extension result and proof template that Theorem 5 generalizes, as well as the comparison point for Theorem 6."},{"cited_title":"2, 903–923","cited_arxiv_id":null,"evidence_quote":"It provides the Hausdorff-dimension union bound of which Theorem 1 is the packing analog, together with sharpness examples."},{"cited_title":"On variants of the Furstenberg set problem","cited_arxiv_id":"2409.03678","evidence_quote":"It gives the packing-dimension line result, and the counterexample showing the pure packing version fails, that Theorem 2 generalizes."},{"cited_title":"Fractal Geom","cited_arxiv_id":null,"evidence_quote":"It gives the Hausdorff union and extension bounds for affine subspaces that the hyperplane results are packing analogs of."},{"cited_title":"Falconer and Pertti Mattila, Strong marstrand theorems and dimensions of sets formed by subsets of hyperplanes , J","cited_arxiv_id":null,"evidence_quote":"It contains the positive-measure hyperplane extension result generalized by Theorem 3."}],"review_version":2}