Compact Reverse Polar Formulation

• Compact reverse polar formulation is a framework that replaces global, implicit structures with explicit, low-dimensional representations across diverse mathematical and physical contexts.
• It provides closed-form inverses for retractions on the Stiefel manifold, achieving second-order accuracy and computational efficiency compared to traditional methods.
• The formulation also underpins semigroup characterizations in convex geometry, strengthened compactness in reverse mathematics, and variable separation in compact-star oscillations.

Searching arXiv for the papers on arXiv and closely related context. Search query: "An new polar factor retraction on the Stiefel manifold with closed-form inverse" Compact reverse polar formulation denotes a family of technically distinct constructions in which compactness, polarity, or a reverse map is encoded through a reduced representation. In the geometry of the compact Stiefel manifold, it refers to a new polar-light retraction with a closed-form inverse (Jensen et al., 23 Feb 2026). In convex geometry of compact group actions, it refers to a reverse characterization of polar representations through a semigroup property for convex hulls of orbits (Gichev, 2009). In reverse mathematics, it appears as strengthened formulations of compactness for countable second-countable spaces, designed to recover classical behavior in weak subsystems of second-order arithmetic (Dorais, 2011). In the theory of polar oscillations of compact stars, a related compact reverse-polar idea is the $H\!-\!S$ formulation, which separates matter and spacetime dynamics into two nearly independent sectors (Zhang et al., 2012).

1. Terminological scope and recurring pattern

The expression is used for different objects in different research areas. The following summary organizes the main meanings.

Context	Formulation	Central feature
Stiefel geometry	Polar-light retraction	Second-order under the Euclidean metric and closed-form inverse
Compact group actions	Semigroup property $\SP$ for orbit hulls	Characterizes polar representations with a Coxeter condition
Reverse mathematics	Effective and $\Sigma^0_n$-compactness	Repairs failures of naive compactness in weak subsystems
Compact-star perturbations	$H\!-\!S$ formulation	Separates matter and spacetime sectors

In the Stiefel setting, the reverse aspect is literal: the retraction admits an explicit inverse map. In the convex-geometric setting, the reverse aspect is characterizational: a semigroup property of orbit hulls forces polarity. In reverse mathematics, the reverse aspect is logical: one asks which strengthened compactness notions restore classical theorems in weak axiomatic systems. In compact-star perturbation theory, the reverse-polar idea is a reorganization of variables so that the dominant physical sector is isolated first.

This suggests a common methodological pattern: globally coupled or implicit structures are replaced by reduced formulations whose inverse, section, or effective coding is explicit. The similarity is structural rather than formal, since the four settings concern unrelated mathematical and physical objects.

2. Closed-form reverse maps on the compact Stiefel manifold

The most literal compact reverse polar formulation appears in the study of retractions on the compact Stiefel manifold

$$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$

A standard base point is

$E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$

and a tangent vector at $\hat U\in St(n,p)$ is represented, using an orthogonal completion $\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$, as

$\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$

The construction is developed for the usual one-parameter family of Stiefel metrics from Hüper–Markina–Leite, with second-order accuracy obtained in the Euclidean case (Jensen et al., 23 Feb 2026).

The motivation is a comparison with standard retractions. The Riemannian exponential map is geometrically exact but expensive, and its inverse is typically iterative. The classical polar factor retraction

$$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$

is second-order under the Euclidean metric, but its inverse is not closed form and usually requires solving a matrix equation of Lyapunov, Sylvester, or Riccati type. The QR retraction is cheaper but only first-order. Cayley retractions are efficient in many situations, but a simple closed-form inverse is not available. Quasi-geodesic retractions do have a closed-form inverse, but only at first order. The stated goal is therefore to retain the favorable geometry of the polar factor retraction while obtaining a simple explicit inverse.

At the base point $\SP$0, the local chart is

$\SP$1

for a Stiefel point split as

$\SP$2

Its inverse is the polar-light retraction

$\SP$3

At an arbitrary base point $\SP$4, the retraction is

$\SP$5

An equivalent efficient form is

$\SP$6

The reverse map is explicit. At $\SP$7, one has $\SP$8. For arbitrary $\SP$9, the paper gives

$\Sigma^0_n$0

and an $\Sigma^0_n$1 SVD form: if

$\Sigma^0_n$2

then

$\Sigma^0_n$3

The factor $\Sigma^0_n$4 is noted to be the Procrustes rotation closest to $\Sigma^0_n$5 relative to $\Sigma^0_n$6.

The theoretical guarantees are local invertibility, the retraction property for the full metric family, and second-order accuracy under the Euclidean metric. The Taylor expansion

$\Sigma^0_n$7

matches the Euclidean Stiefel exponential up to second order. Computationally, inverse evaluation is emphasized as cheap: only $\Sigma^0_n$8 matrix functions or decompositions are needed, the large $\Sigma^0_n$9 operations are mainly matrix-matrix products, and no iterative Lyapunov or Sylvester solve is required. Forward evaluation is slightly more expensive than the classical polar factor retraction because of the extra term $H\!-\!S$0, but the inverse is faster and numerically straightforward. Experiments reported in the paper show that polar-light curves are consistently closer to Stiefel geodesics than the classical polar factor retraction, that the improvement is more noticeable for larger $H\!-\!S$1, and that for $H\!-\!S$2 the two retractions coincide.

3. Reverse characterization of polar representations by orbit-hull semigroups

A second meaning arises in the convex geometry of compact linear group actions. Let $H\!-\!S$3 be a compact linear group acting on a finite-dimensional real Euclidean space $H\!-\!S$4. Writing $H\!-\!S$5 for the convex hull of a set $H\!-\!S$6, and $H\!-\!S$7 for Minkowski addition, the semigroup property $H\!-\!S$8 is that the family

$H\!-\!S$9

is a semigroup under Minkowski addition. Equivalently, the Minkowski sum of two orbit hulls is again an orbit hull. The paper first notes the inclusion

$$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$0

and studies precisely when the orbit-hull family is closed under addition (Gichev, 2009).

A compact linear group $$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$1 is called polar if there exists a linear subspace $$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$2, called a Cartan subspace, such that every orbit meets $$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$3 and, for each $$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$4, the tangent space $$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$5 is orthogonal to $$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$6. The associated Weyl group is

$$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$7

or equivalently

$$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$8

For polar groups, orbit geometry reduces to the action of a finite group on the section $$St(n,p)=\{U\in\mathbb{R}^{n\times p}\mid U^T U = I_p\}.$$9.

The central theorem states that a compact linear group satisfies $E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$0 if and only if it is polar and its Weyl group is a Coxeter group. The theorem specializes to two sharper statements. If $E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$1 is finite, then

$E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$2

If $E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$3 is connected, then

$E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$4

In the general compact, possibly disconnected, case, the additional Coxeter condition on $E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$5 is necessary.

A key structural step is Proposition 1: if $E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$6 is polar, then the map

$E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$7

is a semigroup isomorphism between the family $E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$8 of all $E=\begin{bmatrix}I_p\0\end{bmatrix}\in St(n,p),$9-invariant convex sets in $\hat U\in St(n,p)$0 and the family $\hat U\in St(n,p)$1 of all $\hat U\in St(n,p)$2-invariant convex sets in $\hat U\in St(n,p)$3. In particular,

$\hat U\in St(n,p)$4

and every $\hat U\in St(n,p)$5-invariant convex set is determined by its slice in the Cartan subspace. For $\hat U\in St(n,p)$6,

$\hat U\in St(n,p)$7

The paper also formulates the orbit geometry in terms of support functions and peak sets: $\hat U\in St(n,p)$8 For compact sets $\hat U\in St(n,p)$9,

$\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$0

The cone

$\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$1

satisfies

$\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$2

In the finite-group case, these cones become Dirichlet–Voronoi cells of the orbit, and the Coxeter conclusion follows from the fact that adjacent walls must be reflection hyperplanes.

Here the reverse aspect is not an explicit inverse map but a converse theorem: a convex-geometric semigroup property forces the representation to be polar, with Coxeter control on the Weyl group. This makes the formulation “reverse polar” in a characterizational sense.

4. Strengthened compactness in reverse mathematics

A third usage appears in the reverse mathematics of countable second-countable spaces. The basic object is a triple $\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$3, where $\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$4 is the set of points, $\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$5 is a countable indexed basis, and

$\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$6

is a refinement function witnessing effective intersection refinement. The basis axioms are

$\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$7

and

$\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$8

Because arbitrary unions of basic opens may not exist as sets in weak subsystems, the framework uses open codes: an enumerable subset $\widehat Q=[\hat U\ \hat U_\perp]\in O(n)$9 represents the open set $\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$0 (Dorais, 2011).

Naive compactness is defined by

$\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$1

for every enumerable $\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$2. The paper shows that this ordinary notion behaves unexpectedly over $\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$3: compactness may fail to imply finiteness in discrete spaces, may fail to imply sequential compactness, and products of compact spaces may fail to be compact.

To repair these pathologies, the paper introduces stronger formulations. A space is effectively compact if it is compact and its basis has a finite cover relation, namely a set $\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$4 coding exactly the finite subfamilies that cover $\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$5. For $\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$6, a basis is $\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$7-compact if whenever $\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$8 is a $\xi=\widehat Q\begin{bmatrix}A\B\end{bmatrix}=\hat U A+\hat U_\perp B,\qquad A\in \mathrm{skew}(p),\; B\in\mathbb{R}^{(n-p)\times p}.$9 formula and

$$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$0

there exists a finite set $$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$1 such that

$$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$2

The paper notes explicitly that $$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$3-compactness is equivalent to ordinary compactness, whereas for $$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$4, $$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$5-compactness is much more robust and invariant under ordinary homeomorphisms.

Several reverse-mathematical consequences are established. If $$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$6 is an effectively continuous surjection and $$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$7 is compact, then $$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$8 is compact; if $$R_{\hat U}^{\mathrm{PF}}(\xi)=(\hat U+\xi)\,(I_p+\xi^T\xi)^{-1/2}$$9 is effectively compact, then $\SP$00 is effectively compact. Every effectively closed subspace of a compact space is compact, and of an effectively compact space is effectively compact. Every sequentially compact countable second-countable space is compact, but the converse fails in $\SP$01. By contrast,

$\SP$02

The strengthened compactness notions interact with separation axioms. The paper defines effective Hausdorffness by the existence of uniform functions $\SP$03 separating distinct points by disjoint basic neighborhoods, and proves

$\SP$04

It also proves that every effectively compact subspace of an effectively Hausdorff space is effectively closed; every compact effectively Hausdorff space is effectively regular; every effectively compact effectively Hausdorff space is effectively normal; and, under $\SP$05, every $\SP$06-compact effectively Hausdorff space is effectively normal.

Product results further clarify the distinction between naive and strengthened compactness. The product of sequentially compact spaces is sequentially compact. If $\SP$07 is compact and $\SP$08 effectively compact, then $\SP$09 is compact. If both are effectively compact, then $\SP$10 is effectively compact. If both are $\SP$11-compact for $\SP$12, then $\SP$13 is $\SP$14-compact. Over $\SP$15, the classical landscape is recovered: every compact space is effectively compact, every discrete space is effectively discrete, every Hausdorff space is effectively Hausdorff, and the usual implications hold. Over $\SP$16, the paper gives explicit counterexamples showing that plain compactness is too weak to play the classical role.

In this setting, the compact reverse formulation is a program rather than a single formula: naive compactness is replaced by effective or $\SP$17-strengthened compactness so that the theory behaves correctly inside weak subsystems.

5. Separation of spacetime and matter in polar oscillations of compact stars

In relativistic stellar perturbation theory, the relevant construction is the $\SP$18 formulation for polar oscillations of compact stars. The background is a static, spherically symmetric perfect-fluid star in geometric units $\SP$19. In the Regge–Wheeler gauge, polar perturbations are written as

$\SP$20

In the AAKS formalism, the variables are

$\SP$21

Here $\SP$22 measures the fluid pressure perturbation, $\SP$23 measures the relativistic spacetime curvature perturbation, and $\SP$24 is related to the Newtonian-like part of the metric perturbation; in the weak-field limit,

$\SP$25

so $\SP$26 carries mainly Newtonian gravitational-potential information (Zhang et al., 2012).

The exact polar problem in the frequency domain involves $\SP$27, $\SP$28, and $\SP$29, governed by three coupled second-order equations plus a Hamiltonian constraint. By eliminating $\SP$30, the paper obtains a two-variable system in $\SP$31 and $\SP$32. Inside the star,

$\SP$33

$\SP$34

Here

$\SP$35

$\SP$36 is the sound speed, $\SP$37, the $\SP$38-terms are the main potentials, the $\SP$39 are correction terms from eliminating $\SP$40, and the $\SP$41 are the spacetime-matter coupling terms. Outside the star, $\SP$42, so the system reduces to a single second-order equation for $\SP$43.

The approximation scheme is based on the observation that the coupling terms $\SP$44 are weak for the relevant mode classes. In fluid modes, matter oscillates strongly but spacetime perturbations are weak. In $\SP$45-modes, spacetime oscillates strongly but fluid motion is negligible. Setting

$\SP$46

therefore yields two independent equations: $\SP$47

$\SP$48

For fluid modes, especially $\SP$49-modes, the paper further neglects $\SP$50 and $\SP$51, obtaining the Cowling-type equation

$\SP$52

This is rewritten as a Sturm–Liouville problem

$\SP$53

with

$\SP$54

$\SP$55

The eigenfunctions satisfy orthogonality and completeness relations,

$\SP$56

$\SP$57

At the center,

$\SP$58

and at the surface, for a polytropic EOS $\SP$59, the bounded solution behaves like

$\SP$60

The paper reports that the lowest eigenvalue approximates the $\SP$61-mode with about $\SP$62 error, while higher eigenvalues approximate $\SP$63-modes very accurately, usually better than $\SP$64.

For $\SP$65-modes, the inverse-Cowling approximation drops the $\SP$66 terms in the spacetime equation. Defining

$\SP$67

the equation becomes

$\SP$68

with

$\SP$69

For quasi-normal modes, the boundary condition at infinity is

$\SP$70

The paper reports that this approximation gives extremely accurate polar $\SP$71-mode frequencies, often near-perfect for the real part and highly accurate for the damping rate, including even low-frequency $\SP$72-modes.

The formulation is presented as more physically transparent than the fully coupled polar system. It is not a new physical theory but a reorganization of perturbation variables so that matter-dominant and spacetime-dominant dynamics are isolated first.

6. Comparative interpretation and recurrent misconceptions

These four uses are mathematically distinct. One concerns a retraction on $\SP$73 with a closed-form inverse. One concerns compact group actions on Euclidean spaces and the convex hulls of their orbits. One concerns logical strength and effective coding in second-order arithmetic. One concerns quasi-normal modes of relativistic compact stars. Treating the phrase as a single formalism across all areas would therefore be misleading.

At the same time, the recurrence of the phrase is not arbitrary. In the Stiefel case, the reverse map is explicit and computationally cheap, while second-order accuracy under the Euclidean metric is preserved (Jensen et al., 23 Feb 2026). In the convex-geometric case, a semigroup property of orbit hulls forces polarity and, in the disconnected case, a Coxeter Weyl group (Gichev, 2009). In reverse mathematics, ordinary compactness is too weak over $\SP$74, so effective or $\SP$75-strengthened compactness is introduced to recover stable behavior (Dorais, 2011). In compact-star oscillations, the $\SP$76 reduction separates a matter variable from a spacetime variable, making mode identification and computation substantially clearer (Zhang et al., 2012).

Several misconceptions are directly addressed by the underlying works. In the Stiefel setting, the polar-light retraction is not the exponential map; it only matches the Euclidean Stiefel exponential up to second order. In the convex-geometric setting, polarity alone does not characterize $\SP$77 for arbitrary compact groups; the Coxeter condition on the Weyl group is essential in the general disconnected case. In reverse mathematics, plain compactness does not automatically carry the classical consequences familiar from ordinary topology. In compact-star perturbation theory, the $\SP$78 approach does not replace the exact coupled system; it reorganizes it and then exploits weak coupling to derive accurate approximations.

A plausible unifying interpretation is that compact reverse polar formulations arise when a theory replaces an implicit global object by a representation whose decisive structure is accessible in low dimension, on a section, through an effective code, or in a dominant physical sector. The details are discipline-specific, but the shared strategy is the extraction of a tractable reverse description from a compact or polar framework.