Rubin's Generalized Minkowski–Funk Transforms

• The paper introduces Rubin’s framework that unifies classical Minkowski–Funk transforms with λ-cosine transforms and higher-rank generalizations on Stiefel and Grassmann manifolds.
• It leverages analytic continuation, Fourier and zeta analysis to derive explicit inversion formulas and invariant differential operators in integral geometry.
• The approach provides insights into injectivity, kernel structures, and Radon-type transforms, with practical applications in tomography, harmonic analysis, and PDEs.

Rubin's generalized Minkowski–Funk transforms form a rigorous extension of the classical Minkowski–Funk and Funk–Radon transforms, encompassing a family of integral transforms on the sphere and their higher-rank analogues on Stiefel and Grassmann manifolds. Rubin’s framework unifies $\lambda$-cosine transforms, higher-dimensional integration over matrix domains, and an analytic approach to inversion and regularity, connecting spherical harmonic analysis, Radon transforms, and invariant differential operators. These transforms are widely studied due to their centrality in integral geometry, harmonic analysis, and applications to PDEs and tomography.

1. Classical and Generalized Minkowski–Funk Transforms

The classical Minkowski–Funk transform, or Funk–Radon transform, maps $f\in C(S^{n-1})$ to

$$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$

where $d\sigma_u$ is the invariant measure on the great subsphere orthogonal to $u$. The generalized Minkowski–Funk transforms introduced by Rubin encompass one-parameter families of integral operators (“$\lambda$-cosine transforms”) defined by

$$(C_\lambda f)(u) = \int_{S^{n-1}} |u \cdot v|^\lambda f(v) d\sigma(v),$$

meromorphically continued in $\lambda$ with poles at $\lambda = -1, -2, \ldots$ (Rubin, 2020). For $\lambda = 0$, this recovers the classical Minkowski–Funk transform up to a constant, and for general $f\in C(S^{n-1})$0 it intertwines representations of the rotation group $f\in C(S^{n-1})$1, echoing the principal series intertwiners and analytic continuation studied by Helgason.

Rubin’s generalizations further comprise higher-rank analogues defined on Stiefel $f\in C(S^{n-1})$2 (orthonormal $f\in C(S^{n-1})$3-frames) and Grassmann $f\in C(S^{n-1})$4 (subspaces), via

$f\in C(S^{n-1})$5

valid for $f\in C(S^{n-1})$6, with analytic continuation and residues yielding a spectrum of “intermediate” Funk–cosine transforms integrating over lower-rank matrix loci (Rubin, 2020).

2. Analytic Continuation and Inversion via Fourier and Zeta Analysis

Rubin’s construction leverages homogeneous extension of $f\in C(S^{n-1})$7 from the Stiefel manifold to $f\in C(S^{n-1})$8, exploiting the polar decomposition $f\in C(S^{n-1})$9 ($$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$0, $$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$1) to rewrite the $$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$2-cosine integrals as zeta-type integrals in $$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$3. This approach yields:

• Meromorphic continuation of $$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$4 with poles at $$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$5, generalizing the pole structure for spheres ($$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$6).
• The normalized kernels $$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$7 are entire in $$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$8.
• Inversion formulas are derived explicitly: if $$(F f)(u) = \int_{v \in S^{n-1} : u \cdot v = 0} f(v) d\sigma_u(v),$$9 is an intermediate transform at special value $d\sigma_u$0, and $d\sigma_u$1 is even,

$d\sigma_u$2

where $d\sigma_u$3 is an $d\sigma_u$4-invariant differential operator (the “Cayley–Laplace” determinant operator $d\sigma_u$5 on $d\sigma_u$6) (Rubin, 2020).

In the classical case ($d\sigma_u$7), this reduces to the action of the spherical Beltrami–Laplace operator stepping down $d\sigma_u$8 by two, as previously developed by Helgason, and all the familiar inversion formulas for the Funk transform are recovered (Rubin, 2020).

3. Invariant Differential Operators and Lowering Procedure

A central feature is the construction of explicit invariant differential operators that “lower” the $d\sigma_u$9-order:

$u$0

with $u$1 restricted to the Stiefel manifold, and $u$2 an explicit ratio of Siegel gamma factors (Rubin, 2020, Rubin, 2020). By repeated application, any $u$3 is reduced by an even integer, allowing inversion of the transforms at critical (pole) values.

In the rank-one case ($u$4), $u$5 coincides with the Beltrami–Laplace polynomial $u$6:

$u$7

which underpins Helgason’s inversion for the classical Minkowski–Funk transform. In full generality, this paradigm extends to the Cayley–Laplace determinant as the higher-rank invariant.

4. Non-Central, Shifted, and Dimension-Interpolated Funk Transforms

Rubin’s theory systematically extends the Minkowski–Funk paradigm to non-central and shifted variants, e.g., transforms integrating over sphere sections by planes not passing through the origin, or through arbitrary exterior/interior points (Agranovsky et al., 2019, Agranovsky, 2019, Rubin, 2018):

• The “shifted” Funk transform $u$8 integrates over sections $u$9, where $\lambda$0 passes through $\lambda$1.
• The related “parallel-slice” transform integrates over sections by planes parallel to a fixed direction and is explicitly intertwined with classical Radon–John transforms on the unit ball via Möbius automorphisms and precise Jacobian weightings (Agranovsky et al., 2019).
• An explicit relationship between the shifted Funk and parallel-slice transforms is given by

$\lambda$2

with $\lambda$3 an explicit multiplicative weight and $\lambda$4 a bijection of plane-families (Agranovsky et al., 2019).

The inversion formula for the shifted Funk transform is thus constructed via inversion of the Radon–John transform, together with coordinate-changes and weights. Transform composition and dimension-interpolation correspond to integration over Stiefel families and allow construction of $\lambda$5-plane transforms from $\lambda$6-plane transforms (Agranovsky et al., 2019).

5. Injectivity, Kernels, and Reflection Symmetry

The injectivity and kernel structure of Rubin’s generalized transforms are governed by symmetry and reflection principles:

• The classical Funk transform’s kernel consists of the space of odd functions: $\lambda$7 (Agranovsky, 2019).
• The shifted and parallel-slice transforms have kernel structures determined by weighted “oddness” under involutive symmetries associated with the center $\lambda$8 or direction, e.g., $\lambda$9 (Agranovsky et al., 2019).
• For paired transforms (with multiple centers), injectivity is characterized dynamically: injectivity holds if the composition of reflection-induced involutions does not have finite orbits (i.e., V-map $$(C_\lambda f)(u) = \int_{S^{n-1}} |u \cdot v|^\lambda f(v) d\sigma(v),$$0 for all $$(C_\lambda f)(u) = \int_{S^{n-1}} |u \cdot v|^\lambda f(v) d\sigma(v),$$1) (Agranovsky, 2019).
• The group-theoretic structure extends to arbitrary finite collections of centers, with injectivity translation into the (non)existence of nontrivial periodic orbits in the reflection-group they generate (Agranovsky, 2019).

6. Analytic Families: Cosine and Semyanistyi Transforms, Spectral Structure

Analytic families of transforms, such as the $$(C_\lambda f)(u) = \int_{S^{n-1}} |u \cdot v|^\lambda f(v) d\sigma(v),$$2-cosine and Semyanistyi fractional Radon transforms, interpolate between classical and generalized Minkowski–Funk transforms:

• The $$(C_\lambda f)(u) = \int_{S^{n-1}} |u \cdot v|^\lambda f(v) d\sigma(v),$$3-cosine transform

$$(C_\lambda f)(u) = \int_{S^{n-1}} |u \cdot v|^\lambda f(v) d\sigma(v),$$4

yields the Funk transform as $$(C_\lambda f)(u) = \int_{S^{n-1}} |u \cdot v|^\lambda f(v) d\sigma(v),$$5 (Rubin, 2018).

• The “shifted” cosine transform

$$(C_\lambda f)(u) = \int_{S^{n-1}} |u \cdot v|^\lambda f(v) d\sigma(v),$$6

interpolates between first-kind and second-kind transforms, with inversion and analytic continuation via modified stereographic projections (Rubin, 2018).

• On Sobolev spaces, the action of Rubin's generalized Minkowski–Funk transforms is dictated by their spectral multipliers, with explicit harmonic expansions and asymptotic bounds (Han et al., 14 Jan 2026). For irrational sphere cap angles and non-critical indices, small denominator phenomena obstruct endpoint regularity of inversion (almost everywhere in parameter space), as proven in the small denominator problem (Han et al., 14 Jan 2026).

7. Summary of Inversion, Regularity, and Applications

Rubin’s generalized Minkowski–Funk transforms underpin a comprehensive analytic framework:

• Explicit inversion via Fourier analysis and differential operators holds on suitable Sobolev and distributional spaces (Rubin, 2020, Rubin, 2020).
• The machinery extends seamlessly to higher-rank Stiefel and Grassmann domains, with Cayley–Laplace-style differential operators supplanting lower-dimensional Laplacians.
• The transforms are intimate with Radon–John transforms (via parallel slice and shifted variants) and with spectral analysis of spherical harmonics.
• Generic regularity theory exposes Diophantine small-divisor obstructions at critical smoothing exponents, establishing limits to endpoint Sobolev-mapping for the inverses (Han et al., 14 Jan 2026).

Rubin’s program thus unifies and extends core tools of integral geometry and analysis, enabling fine control of transform domains, inversion, kernel structure, and mapping properties necessary for applications in tomography, harmonic analysis, and the study of partial differential equations.