Even Minkowski uniqueness as a consequence of logarithmic Brunn–Minkowski
Weighted surface-area data describe a convex body through its supporting directions. This note records two distinct uniqueness statements for origin-symmetric bodies. Smooth strictly positive even data give a unique smooth, positively curved solution for 0 ≤ p < 1. Equal weighted measures determine arbitrary full-dimensional origin-symmetric bodies for 0 < p < 1.
These are attributed consequences of OpenAI's pinned logarithmic Brunn–Minkowski inequality and established transfer, existence and regularity results. The note supplies an explicit nonsmooth first-variation account and a local smooth endpoint calculation.
Original abstract
We record consequences of OpenAI's recently released origin-symmetric logarithmic Brunn–Minkowski inequality. For every ambient dimension n ≥ 2 and 0 ≤ p < 1, every smooth strictly positive even density on Sⁿ⁻¹ has exactly one smooth positive even support-function solution of h¹⁻ᵖ det(∇²h+hI)=f, with positive curvature-radius matrix. For 0 < p < 1, identical Lₚ surface-area measures determine arbitrary full-dimensional origin-symmetric convex bodies. We give the nonsmooth first-variation and strictness argument explicitly and a local version of the smooth endpoint transfer of He–Liu. Existence and regularity are established inputs. These are consequences of the new inequality and known reductions, not a new proof of logarithmic Brunn–Minkowski or a claim of first public announcement. Equal-volume boxes exclude arbitrary-measure uniqueness at p=0.
What the prescribed measure records
For a full-dimensional origin-symmetric convex body K = −K, the support function hK(u) = maxx∈K x·u is the distance of its supporting hyperplane in direction u from the origin. This function is positive and even.
Ordinary surface-area measure SK records boundary area by outer-normal direction. The Lp surface-area measure weights it by the support function:
dSK,p = hK1−p dSK.
This is the unnormalized measure. It is not divided by volume or total mass. For a smooth body, prescribing its density relative to ordinary, unnormalized spherical area dω becomes a Monge–Ampère equation for h.
Two uniqueness statements, with different endpoint scopes
In both statements, n is the ambient dimension and n ≥ 2.
- Smooth positive data, 0 ≤ p < 1. For every strictly positive even f ∈ C∞(Sn−1), there exists exactly one positive even smooth support function h satisfying
h1−p det Q(h) = f, Q(h) = ∇²h+hI > 0.
The Hessian is spherical; the determinant and positive-definiteness condition are on the (n−1)-dimensional tangent space. Equivalently, h supports a smooth origin-symmetric body with strictly positive Gauss curvature.
- Arbitrary bodies, 0 < p < 1 only. If K,L are full-dimensional origin-symmetric convex bodies and SK,p = SL,p, then K = L. The bodies need not be smooth; the measures need not have smooth densities.
No volume-one condition is imposed. Dilating by c > 0 gives ScK,p = cn−pSK,p. Since n−p > 0, fixed nonzero data determine scale. Evenness fixes the center and removes translations. At p = 0, a smooth solution has nV(K) = ∫f dω.
The admissible support-function class, positivity, evenness and Q > 0 are part of the smooth theorem. It is not an unconstrained uniqueness claim for arbitrary functions.
The logarithmic Brunn–Minkowski input
The new inequality is credited to OpenAI's pinned manuscript. For origin-symmetric convex bodies and 0 ≤ t ≤ 1, it states
V(𝓦[hK1−thLt]) ≥ V(K)1−tV(L)t.
Here 𝓦[a] is the intersection of the half-spaces x·u ≤ a(u). The bound a need not itself be a support function. The deduction uses this volume inequality without an equality classification for logarithmic Brunn–Minkowski.
Why the positive-p statement covers nonsmooth bodies
In the stated range 0 < p < 1, the note writes out the nonsmooth Wulff first variation and the established logarithmic Minkowski transfer. Jensen's inequality then gives
Vp(K,L) = (1/n)∫hLphK1−p dSK ≥ V(K)1−p/nV(L)p/n.
At equality, strict Jensen makes the support ratio constant only SK-almost everywhere. Ordinary Minkowski equality then gives homothety and translation; origin symmetry removes the translation. Pointwise equality of support functions is not inferred directly from a possibly singular measure.
Equal unnormalized Lp measures give Vp(K,L) = V(L) and the reverse identity. Applying the mixed inequality in both orders, with n−p > 0, forces equal volumes and then identical bodies. The transfer machinery is credited to Böröczky–Lutwak–Yang–Zhang and the other cited authors.
The smooth endpoint uses only a local path
For two admissible solutions h0,h1 at p = 0, put M = ∫f dω. Both bodies have volume M/n. The logarithmic Minkowski transfer makes each a global minimizer of
𝓕(K) = (1/M)∫f log hK dω − (1/n)log V(K).
Their weighted logarithmic integrals agree. Set w = log(h1/h0) and qt = h0etw. Positive definiteness is open on the compact sphere, so Q(qt) > 0 for t in a neighborhood of zero. Only this local smooth support-function path is used.
For sufficiently small t ≥ 0, the volume inequality and minimality force constant volume and, by arbitrary smooth even variations, qt det Q(qt) = f. Differentiate from the right at zero. Writing h = h0, Q = Q(h) and A = Q(hw) gives
0 = −tr((Q−1A)²) − (n+1)w² + 2hQ−1(∇w,∇w).
The coefficient is n+1 with n the ambient dimension. The trace is nonnegative because Q−1A is similar to a real symmetric matrix. At both extrema of w the gradient vanishes, forcing both extrema to be zero. Thus h0 = h1, including in dimension two. This is a local version of the smooth endpoint mechanism credited to He–Liu.
Existence and regularity are established inputs
The note applies the symmetry-preserving existence theorem of Bianchi–Böröczky–Colesanti–Yang with the group {I,−I}. Full dimension and origin symmetry put the origin in the interior, making the support function positive. The global regularity result of Bianchi–Böröczky–Colesanti supplies the boundary/support-chart bridge; Chou–Wang supplies higher support regularity. Smooth positive data then give Q > 0 and smoothness. In dimension two the equation h″+h = fhp−1 gives a direct bootstrap.
The base inequality, transfers, existence and regularity remain credited to their sources. The note makes no new inequality, transfer or first-public-announcement claim.
Why arbitrary-measure uniqueness stops at p = 0
Let B(a) = ∏i=1n[−ai,ai], with all ai > 0. Its ordinary surface-area atom at each ±ei has mass V(B)/(2ai), while the support there is ai. Consequently every atom of SB,0 has mass V(B)/2.
For s > 0, s ≠ 1, the boxes B(1,…,1) and B(s,s−1,1,…,1) are distinct, have equal volume and have identical unnormalized L0 measures. Their measures are atomic. They are outside the smooth strictly positive density class, so this boundary example does not contradict the smooth endpoint theorem.
Degenerate bodies, non-even data and arbitrary functions outside the admissible positive support class are not covered. No result for those omitted classes is asserted.
Original files and manuscript
- Manuscript (PDF, 79,097 bytes) · Zenodo original PDF
- Source and verification package (ZIP, 123,632 bytes) · Zenodo original ZIP
- README.md (6,229 bytes) · Zenodo original README
Open the five-page PDF · Download PDF · Zenodo record
These are unchanged published originals in the record's order. The manuscript and declared publication date are 6 October 2026; record metadata reports creation on 7 October 2026 UTC and updates on 10 October 2026. The hosting page displays edition v1; no author-supplied version number is stated in the record metadata.
Credited sources
- OpenAI, The logarithmic Brunn–Minkowski conjecture, pinned at commit
adc7f1241b42e322a6451854ab7e4b4c146bf78a: the base origin-symmetric volume inequality. - K. J. Böröczky, E. Lutwak, D. Yang and G. Zhang, The log-Brunn–Minkowski inequality (2012): the established logarithmic and Lp transfers.
- W. He and J. Liu, On the uniqueness of even Lp-Minkowski problem, v1 (2025), Theorem 4 and Section 3: the smooth endpoint mechanism.
- G. Bianchi, K. J. Böröczky, A. Colesanti and D. Yang, The Lp-Minkowski problem for −n < p < 1, Theorem 1.7 and Proposition 7.3: symmetry-preserving existence.
- G. Bianchi, K. J. Böröczky and A. Colesanti, Smoothness in the Lp Minkowski problem for p < 1: global regularity; K.-S. Chou and X.-J. Wang, The Lp-Minkowski problem and the Minkowski problem in centroaffine geometry, Proposition 1.2: higher support regularity.
- A. Stancu, A PDE approach to geometric inequalities (2018): the earlier announcement credited in the manuscript.
The manuscript gives the broader bibliography and source qualifications. Its contribution is an explicit consequence/exposition account, with no priority or formal-verification certificate.
Citation and rights
Kriebel, Alec. (2026). Even Minkowski uniqueness as a consequence of logarithmic Brunn–Minkowski [Unrefereed preprint; Zenodo display edition v1]. Zenodo. doi:10.5281/zenodo.23203081. Download BibTeX.
The owned contribution is licensed CC BY 4.0. Cited third-party works retain their own rights, as described in the original README.
Status and collaboration
The mathematical basis is the cited analytic inequality and the deductions in the manuscript. Finite exact-arithmetic checks support identities and boundary behavior; they do not replace the universal proofs. No reproduced Lean kernel/axiom/Comparator certification or formalization of this follow-on theorem is claimed.
AI tools were used extensively in research, drafting and verification. This is an unrefereed, AI-assisted preprint; automated adversarial reviews are not conventional human peer review. The author is responsible for the claims.
Independent verification of the nonsmooth variation, local endpoint argument or existence/regularity interfaces, substantive expository work, and development of a journal version would be useful contributions. Collaboration and authorship would be based on actual contributions.