Mixed Area Measures in Convex Geometry

• Mixed area measures are finite Borel measures that localize mixed volumes to boundary-normal directions and underpin support and inverse geometric problems.
• They arise via polarization of classical surface area measures and are characterized by symmetry, translation-invariance, and 1-homogeneity across all arguments.
• Recent studies extend their framework to weighted, fractional, and translative contexts, yielding new projection formulas and density representations.

Mixed area measures are finite Borel measures on the unit sphere that localize mixed volumes to boundary-normal directions. For convex bodies $K_1,\dots,K_{n-1}\in\mathcal K^n$, the measure $S(K_1,\dots,K_{n-1},\cdot)$ is characterized by the mixed-volume localization formula

$$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$

valid for every convex body $L$. In contemporary convex-geometric research, mixed area measures serve simultaneously as the boundary-measure counterpart of mixed volumes, as the local objects governing Christoffel- and Minkowski-type inverse problems, and as generators of large classes of translation invariant area measures. Recent work has focused on support characterizations, Kubota-type projection formulas, inverse problems with anisotropic reference bodies, weighted and fractional analogues, and their role in the representation-theoretic theory of area measures (Hug et al., 2023, Hug et al., 2024, Knoerr, 29 May 2026).

1. Definition, polarization, and elementary structure

Mixed area measures arise by polarization of the surface area measure. One formulation is

$$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$

which makes the map symmetric and multilinear in its arguments (Knoerr, 29 May 2026). An equivalent characterization, emphasized in several papers, is the localization identity above, which expresses mixed volume as an integral of a support function against a measure on $S^{n-1}$ (Brauner et al., 13 Aug 2025).

The standard shorthand

$$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$

identifies the classical $j$-th area measures as mixed area measures with repeated Euclidean unit balls (Hug et al., 2024). In the disk-reference setting, the anisotropic first area measure is defined by

$$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$

and for $C=B^n$ this recovers the classical first area measure $S(K_1,\dots,K_{n-1},\cdot)$0 (Brauner et al., 13 Aug 2025).

Several structural properties recur across the literature. The mixed area measure is symmetric in the arguments, translation-invariant in each argument, $S(K_1,\dots,K_{n-1},\cdot)$1-homogeneous in each argument, centered, and non-negative (Brauner et al., 13 Aug 2025). For point masses, one has the support-face identity

$S(K_1,\dots,K_{n-1},\cdot)$2

which ties atoms of the measure to mixed volumes of support faces in direction $S(K_1,\dots,K_{n-1},\cdot)$3 (Hug et al., 2023). In the polytope case, if $S(K_1,\dots,K_{n-1},\cdot)$4, then

$S(K_1,\dots,K_{n-1},\cdot)$5

so the measure becomes an explicit sum over facet normals (Hug et al., 2023).

These formulas show that mixed area measures encode not only first variation but also directional support-face geometry. In the classical theory they are the local counterparts of mixed volumes; in later developments they become the natural measure-valued data in inverse problems and support theorems.

2. Support, extreme normals, and Schneider’s conjecture

A central problem is to describe the support of $S(K_1,\dots,K_{n-1},\cdot)$6. Schneider’s conjecture predicts that

$S(K_1,\dots,K_{n-1},\cdot)$7

where $S(K_1,\dots,K_{n-1},\cdot)$8 is the set of extreme normal vectors determined by touching-space data (Hug et al., 2023).

For an $S(K_1,\dots,K_{n-1},\cdot)$9-tuple $$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$0, a unit vector $$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$1 is $$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$2-extreme if the touching spaces $$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$3 contain one-dimensional subspaces with linearly independent directions. Equivalently,

$$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$4

The recent support theorem proves

$$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$5

for every $$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$6-tuple $$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$7 of polyoids, and also for smooth convex bodies provided at least one body is smooth and strictly convex (Hug et al., 2023).

Polyoids enter here as limits of Minkowski sums of $$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$8-topes, with a generating-measure representation

$$V(L,K_1,\ldots,K_{n-1})=\frac{1}{n}\int_{S^{n-1}} h_L(u)\,S(K_1,\ldots,K_{n-1};du),$$9

and the support theorem extends earlier cases covering polytopes, zonoids, triangle bodies, and the special unit-ball setting. The stated motivation is that earlier work of Yair Shenfeld and Ramon van Handel characterized equality in Alexandrov–Fenchel by equality of support functions on the support of a mixed area measure, so the support problem is structurally decisive for equality theory (Hug et al., 2023).

A special support theorem is available for the mixed measures

$L$0

where $L$1 is the $L$2-dimensional Euclidean unit ball in a hyperplane $L$3. In that setting,

$L$4

confirming Schneider’s conjecture in this case as well (Hug et al., 2024). An additional consequence is the nested-support inclusion

$L$5

for $L$6, while the case $L$7 fails in general (Hug et al., 2024).

3. Projection formulas and Kubota-type representations

Recent work has shown that mixed area measures with lower-dimensional reference bodies admit precise projection formulas. For a fixed line $L$8, the Grassmannian $L$9 of $$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$0-dimensional subspaces containing $$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$1, and measurable $$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$2, one has the Kubota-type formula

$$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$3

Here $$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$4 denotes the mixed area measure in the ambient subspace $$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$5 (Hug et al., 2024).

This formula shows that the mixed area measure in $$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$6 with $$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$7 copies of $$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$8 is an average of lower-dimensional mixed area measures of orthogonal projections onto subspaces containing $$S(K_1,\dots,K_{n-1})=\frac{1}{(n-1)!}\frac{\partial^{n-1}}{\partial\lambda_1\dots\partial\lambda_{n-1}}\Big|_0 S_{n-1}\!\left(\sum_{j=1}^{n-1}\lambda_j K_j\right),$$9. The resulting support description becomes

$S^{n-1}$0

so support can be read off from extreme unit normals of projected bodies (Hug et al., 2024).

A related Kubota-type formula underlies the disk Christoffel problem. If $S^{n-1}$1 is an $S^{n-1}$2-dimensional disk with axis parallel to $S^{n-1}$3, then

$S^{n-1}$4

This reduces the $S^{n-1}$5-dimensional mixed problem to a family of planar first area measures $S^{n-1}$6 (Brauner et al., 13 Aug 2025).

These formulas generalize the classical Cauchy–Kubota philosophy from intrinsic volumes to measure-valued localizations. They also explain why many inverse problems for mixed area measures become tractable after reducing to lower-dimensional sections or projections.

4. Christoffel and Christoffel–Minkowski problems

The mixed Christoffel problem asks when a finite Borel measure on $S^{n-1}$7 is a mixed area measure of a convex body with prescribed reference bodies. In the disk case, the problem is to characterize $S^{n-1}$8 for which

$S^{n-1}$9

for some convex body $$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$0, where $$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$1 is an $$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$2-dimensional disk (Brauner et al., 13 Aug 2025).

For a non-negative, centered, finite Borel measure $$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$3 with $$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$4, existence of such a $$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$5 is equivalent to three conditions: $$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$6 is absolutely continuous with a continuous density; for almost every $$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$7,

$$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$8

and there exists a support function $$S_j(K,\cdot)=S\bigl(K[j],B^{n-1}[n-1-j],\cdot\bigr)$$9 satisfying the fiberwise compatibility identity

$j$0

In that case $j$1 is unique up to translation (Brauner et al., 13 Aug 2025).

The pole condition $j$2 is geometrically essential. The measure $j$3 can place mass at the poles when $j$4 has faces orthogonal to $j$5, and this creates non-uniqueness and a disintegration ambiguity because the poles belong to every circle $j$6 in the planar decomposition (Brauner et al., 13 Aug 2025).

A broader mixed Christoffel–Minkowski problem has been solved for bodies of revolution around a common axis. For $j$7, reference bodies of revolution $j$8, and a nonnegative centered zonal Borel measure $j$9, the existence of a body of revolution $$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$0 such that

$$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$1

is characterized by five explicit conditions: a support restriction to a spherical segment, non-concentration on the equator, nonnegativity and finiteness of a transformed Radon measure, existence of endpoint limits, and an equatorial mass inequality (Brauner et al., 13 Aug 2025). The proof uses the one-variable transform

$$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$2

together with an operator $$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$3 that converts mixed area measures with arbitrary rotationally symmetric reference bodies into disk-reference measures (Brauner et al., 13 Aug 2025).

These results place mixed area measures at the center of inverse boundary-measure problems. They also show that regularity assumptions can often be replaced by explicit transform conditions on spherical measures.

5. Weighted, fractional, and translative analogues

In weighted Brunn–Minkowski theory, classical mixed area measures are replaced by weighted surface area measures and weighted mixed surface area measures associated with a measure $$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$4 having density $$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$5. The weighted surface area measure $$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$6 is defined by

$$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$7

and the first mixed measure satisfies

$$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$8

For $$S_1(K,C;\beta)=\frac{1}{(n-1)!}\left.\left(\frac{d}{dt}\right)^{n-2}\right|_{t=0^+} S_{n-1}(K+tC,\beta),$$9, the weighted mixed surface area measure $C=B^n$0 yields the second-order formula

$C=B^n$1

but $C=B^n$2 is in general only a signed measure, and $C=B^n$3 can be negative (Fradelizi et al., 2022).

A later asymptotic study interprets these constructions as weighted analogues of mixed volumes and mixed area measures for log-concave densities of the form

$C=B^n$4

In that setting the first mixed measure has the boundary-integral representation

$C=B^n$5

and, in the smooth second-order theory,

$C=B^n$6

with an explicit expression for $C=B^n$7 involving the classical mixed area measure $C=B^n$8 and a density-gradient correction term (Lafi et al., 21 Feb 2026).

A distinct extension is provided by anisotropic fractional area measures. For an origin-symmetric convex body $C=B^n$9, the anisotropic $S(K_1,\dots,K_{n-1},\cdot)$00-fractional area measure $S(K_1,\dots,K_{n-1},\cdot)$01 is defined as the first variation of the anisotropic fractional $S(K_1,\dots,K_{n-1},\cdot)$02-perimeter $S(K_1,\dots,K_{n-1},\cdot)$03. For smooth strictly convex $S(K_1,\dots,K_{n-1},\cdot)$04,

$S(K_1,\dots,K_{n-1},\cdot)$05

where $S(K_1,\dots,K_{n-1},\cdot)$06 is the moment body of $S(K_1,\dots,K_{n-1},\cdot)$07, defined by

$S(K_1,\dots,K_{n-1},\cdot)$08

Thus the anisotropic nonlocal theory converges, after normalization, to a mixed area measure involving the moment body of the anisotropy (Cai, 6 Oct 2025).

In translative integral geometry, mixed curvature measures play an analogous role for translations and intersections rather than Minkowski sums. The translative average of curvature measures of intersections of translated sets decomposes into mixed curvature measures, and in the polyhedral case the resulting formulas are described as exactly analogous to the structure of mixed area measures, with coefficients determined by face geometry and normal cones (Hug et al., 2016). This identifies mixed area measures as one branch of a larger family of local mixed geometric measures.

6. Area-measure theory, density, and modern structural role

Mixed area measures are not only examples within area-measure theory; they generate a dense part of the theory. Continuous translation invariant area measures are weak* continuous, locally determined, translation invariant measure-valued functionals

$S(K_1,\dots,K_{n-1},\cdot)$09

and the space of $S(K_1,\dots,K_{n-1},\cdot)$10-smooth area measures coincides with the space of measures obtained by integrating translation-invariant differential forms over the normal cycle (Knoerr, 29 May 2026).

Within this framework, mixed area measures admit a differential-form representation through iterated Lie derivatives. For smooth convex bodies,

$S(K_1,\dots,K_{n-1},\cdot)$11

which expresses mixed area measures directly in normal-cycle language (Knoerr, 29 May 2026).

A finite-generation theorem then states that there exist finitely many ellipsoids $S(K_1,\dots,K_{n-1},\cdot)$12 such that every $S(K_1,\dots,K_{n-1},\cdot)$13-smooth area measure in degree $S(K_1,\dots,K_{n-1},\cdot)$14 can be written as a finite linear combination

$S(K_1,\dots,K_{n-1},\cdot)$15

Accordingly, mixed area measures generate dense submodules with respect to the uniform weak* and uniform compact topologies (Knoerr, 29 May 2026).

This density statement is the area-measure analogue of the role of mixed volumes in valuation theory. It also underlies Hadwiger-type classification theorems: if $S(K_1,\dots,K_{n-1},\cdot)$16 is compact and acts transitively on $S(K_1,\dots,K_{n-1},\cdot)$17, then every continuous, translation invariant, locally determined, $S(K_1,\dots,K_{n-1},\cdot)$18-equivariant area measure is represented by a $S(K_1,\dots,K_{n-1},\cdot)$19-invariant differential form on the normal cycle, and $S(K_1,\dots,K_{n-1},\cdot)$20 is finite dimensional (Knoerr, 29 May 2026).

Mixed area measures therefore occupy two complementary positions in modern convex geometry. At the geometric level, they localize mixed volumes, govern support and inverse problems, and admit projection formulas that expose lower-dimensional structure. At the structural level, they generate dense subspaces of area measures and connect normal-cycle calculus, invariant theory, and weighted or nonlocal extensions.