Isometric Gauge Transformations in Geometry

• Isometric gauge transformations are mappings that preserve or reverse specific gauge metrics, playing a central role in cone geometry and pure-connection gravity.
• They provide a unified framework linking metric isometries in Hilbert/Thompson metrics with gauge covariance in chiral and integrable connection formulations.
• Their analysis underpins classification of symmetric cones and informs the recovery of Killing symmetries, with implications for integrable hierarchies and immersion theory.

Across current literatures, “isometric gauge transformations” does not denote a single universally fixed construction. The phrase covers, at minimum, three distinct regimes: gauge-preserving and gauge-reversing maps that are genuine isometries of Hilbert and Thompson geometries on cones; infinitesimal spacetime isometries detected through gauge covariance of a chiral connection in four-dimensional pure-connection geometry; and gauge transformations that preserve an underlying metric, conformal, or Galilei structure while altering auxiliary data, as in integrable surface theory, low-regularity isometric immersions, and twistless torsional Newton–Cartan geometry (Walsh, 2013, Frodden et al., 2020, Li, 2024, Blanckenburg et al., 2024).

1. Conceptual scope

The most literal use of the term occurs in cone geometry. There, the basic “gauge” is the order-theoretic quantity

$M(x/y):=\inf\{\lambda>0:\ x\le_C \lambda y\},$

defined on a proper open convex cone $C$. A map is gauge-preserving if it preserves $M(x/y)$, and gauge-reversing if it sends $M(x/y)$ to $M(y/x)$; both are automatically isometries for the Hilbert and Thompson metrics. In this setting, “gauge” and “isometry” are tied directly to metric geometry (Walsh, 2013).

A different meaning arises in four-dimensional pure-connection gravity. There the metric is reconstructed from the curvature of an $\mathrm{SO}(3)$ connection, and infinitesimal isometries are characterized by the requirement that the Lie derivative of the connection along a vector field be a gauge transformation. Isometry is therefore read off at the level of the connection rather than from a Killing equation for a primary metric field (Frodden et al., 2020).

By contrast, several gauge-theoretic literatures use “gauge transformation” in a non-metric sense. Finite BRST-antiBRST transformations act on the enlarged field space of quantized gauge theory and relate gauge-fixing functionals, but are explicitly not spacetime or manifold isometries (Moshin et al., 2014). This suggests that any encyclopedia treatment must separate genuine metric isometries induced by gauge data from broader gauge-equivalence constructions.

2. Pure-connection geometry and the recovery of Killing symmetries

In the pure-connection formalism, the basic field is an $\mathrm{SO}(3)$ connection $A=(A^i)$ on a rank-3 oriented Euclidean vector bundle $E\to M$, with curvature

$$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$

Given a nowhere-vanishing 4-form $C$0, the curvature defines a symmetric matrix $C$1 through

$C$2

A connection is definite when $C$3 is definite at every point. In that case, the curvature determines a conformal metric via the Urbantke-type formula

$C$4

with $C$5, and one can choose a preferred representative by fixing the conformal factor. Relative to a basis $C$6 of self-dual 2-forms, the curvature takes the form

$C$7

The geometry is thus encoded algebraically in the connection curvature rather than postulated as an independent metric field (Frodden et al., 2020).

The central isometry criterion is

$C$8

with $C$9 a vector field and $M(x/y)$0 an $M(x/y)$1-valued gauge parameter. Writing

$M(x/y)$2

this becomes

$M(x/y)$3

Because the metric is a gauge-invariant algebraic functional of $M(x/y)$4, this condition implies $M(x/y)$5: infinitesimal isometries are precisely those infinitesimal diffeomorphisms whose action on the connection is gauge-trivial (Frodden et al., 2020).

A further step eliminates $M(x/y)$6 and yields a first-order differential equation for the shifted gauge parameter alone:

$M(x/y)$7

Once $M(x/y)$8 is found, the Killing vector is recovered algebraically as

$M(x/y)$9

The paper works this out for $M(x/y)$0, $M(x/y)$1, and a spherically symmetric connection, recovering the full isometry algebras directly from the gauge-parameter equation (Frodden et al., 2020).

The formalism also yields a no-symmetry theorem. For a negative definite connection on a compact manifold,

$M(x/y)$2

and the sign definiteness forces $M(x/y)$3 and $M(x/y)$4. This is the pure-connection analogue of the classical statement that compact Riemannian manifolds with negative Ricci curvature admit no nonzero Killing fields (Frodden et al., 2020).

3. Gauge-preserving and gauge-reversing isometries on cones

Let $M(x/y)$5 be a proper open convex cone in a finite-dimensional real vector space $M(x/y)$6, with order

$M(x/y)$7

The order gauge

$M(x/y)$8

generates two geometries:

$M(x/y)$9

The first is Hilbert’s metric on projective space $M(y/x)$0; the second is Thompson’s metric on $M(y/x)$1 itself (Walsh, 2013).

A map $M(y/x)$2 is gauge-preserving when

$M(y/x)$3

and gauge-reversing when

$M(y/x)$4

Both are isometries for $M(y/x)$5 and $M(y/x)$6. Gauge-preserving bijections between finite-dimensional cones are rigid: any such bijection is the restriction of a linear isomorphism. The genuinely new phenomena therefore arise from gauge-reversing maps (Walsh, 2013).

The central theorem is exact: a proper open convex cone admits a gauge-reversing map if and only if it is symmetric, meaning both homogeneous and self-dual. On symmetric cones the relevant inversion is Vinberg’s $M(y/x)$7-map,

$M(y/x)$8

which is anti-homogeneous and antitone. In Euclidean Jordan algebra language it coincides, up to scalar, with the Jordan inverse; in the positive-definite matrix cone it is ordinary matrix inversion $M(y/x)$9, and in the positive orthant it is coordinatewise inversion $\mathrm{SO}(3)$0 (Walsh, 2013).

This theorem controls the isometry groups of both Hilbert and Thompson geometries. For Hilbert geometry, every isometry arises as the projective action of either a gauge-preserving or a gauge-reversing map. Hence the full isometry group exceeds the collineation group only for projective spaces of symmetric non-Lorentzian cones; in that case the collineation group is a normal subgroup of index two. For Lorentz cones, by contrast, the projective action of the $\mathrm{SO}(3)$1-map is already a collineation, so no extra Hilbert isometries appear (Walsh, 2013).

Thompson geometry is more flexible. Every surjective Thompson isometry splits, after decomposing the cone as a direct product, into a gauge-preserving part on one factor and a gauge-reversing part on another. The product formula

$\mathrm{SO}(3)$2

explains why Thompson geometry behaves as an $\mathrm{SO}(3)$3-product and why preserving and reversing behaviors can coexist on different factors (Walsh, 2013).

4. Flat connections, integrable hierarchies, and structure-preserving gauges

In integrable geometry, the natural “isometric” analogue is not usually a metric isometry of an immersion, but a gauge transformation preserving a zero-curvature representation. For $\mathrm{SO}(3)$4-surfaces, with Gauss map $\mathrm{SO}(3)$5, the harmonic-map decomposition

$\mathrm{SO}(3)$6

gives the associated family

$\mathrm{SO}(3)$7

and $\mathrm{SO}(3)$8 is harmonic exactly when $\mathrm{SO}(3)$9 is flat for all $\mathrm{SO}(3)$0. Trivialising gauges $\mathrm{SO}(3)$1 generate the spectral deformation and the Sym formula, while Bäcklund transforms arise from rational dressing gauges built from $\mathrm{SO}(3)$2-parallel null lines. For isothermic surfaces the analogous flat family is

$\mathrm{SO}(3)$3

with $\mathrm{SO}(3)$4 closed; $\mathrm{SO}(3)$5-transforms come from trivialising gauges, and Darboux transforms from $\mathrm{SO}(3)$6-parallel null line subbundles. These gauge actions are orthogonal on the ambient bundle, but the resulting surface transforms generally preserve the relevant integrable structure rather than the first fundamental form itself (Burstall, 2015).

The same pattern persists in zero-curvature formulations of integrable PDEs. In the $\mathrm{SO}(3)$7 framework, the Drinfeld–Sokolov gauge

$\mathrm{SO}(3)$8

is preserved by residual gauge transformations generated by

$\mathrm{SO}(3)$9

which induce infinitesimal symmetries

$A=(A^i)$0

After a further reduction $A=(A^i)$1, the parameter $A=(A^i)$2 must satisfy a linear compatibility equation, and the residual gauge transformations generate the higher-symmetry hierarchies of KdV and Harry Dym. In this sense, the closest analogue of an “isometric gauge transformation” is the stabilizer of the chosen gauge slice, not an arbitrary element of the full gauge group (0705.3530).

For generalized $A=(A^i)$3-KdV hierarchies, Miura and Bäcklund transformations are likewise realized as gauge transformations of zero-curvature operators. The mKdV and KdV Lax operators are related by

$A=(A^i)$4

and the Miura gauge $A=(A^i)$5 is not unique. Instead, there are $A=(A^i)$6 branches, organized by the identity and the kernel generators $A=(A^i)$7, $A=(A^i)$8, associated with the exponents of $A=(A^i)$9. Bäcklund transformations satisfy

$E\to M$0

and the KdV Bäcklund gauge is obtained by composition,

$E\to M$1

The zero-curvature representation is therefore the universal object preserved across all flows of the hierarchy (Ferreira et al., 2021).

5. Gauge transforms in isometric immersion theory and TTNC geometry

In low-regularity immersion theory, gauge transformations appear inside the PDE analysis of the isometric immersion problem

$E\to M$2

The second fundamental form and normal connection are assembled into an antisymmetric matrix-valued connection $E\to M$3-form $E\to M$4, and the Gauss–Codazzi–Ricci system is the flatness condition

$E\to M$5

A gauge transformation is a change of local orthonormal frame,

$E\to M$6

under which flatness is invariant. The analytically decisive choice is the Coulomb–Uhlenbeck gauge,

$E\to M$7

which converts the weak flatness system into an elliptic equation amenable to compensated compactness and Hardy–BMO estimates. In the Morrey-scale theorem stated there, if $E\to M$8 and

$E\to M$9

is a weak solution of Gauss–Codazzi–Ricci, then there exists an isometric immersion

$$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$0

with extrinsic geometry $$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$1, unique modulo rigid motions. Here the gauge transformation serves the isometric immersion problem, but it is not itself a metric isometry of the ambient space (Li, 2024).

Twistless torsional Newton–Cartan geometry provides a different “metric-preserving modulo gauge” example. The underlying Galilei structure consists of a clock form $$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$2, a degenerate spatial metric $$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$3, and a Bargmann-form representative $$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$4. Type I gauge transformations act by

$$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$5

while type II gauge transformations act by

$$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$6

Both preserve the Galilei metric structure $$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$7 while changing the Bargmann representative. The distinguished locally Galilei-invariant potential

$$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$8

can always be gauged away locally. In type I, the gauge-fixing equation

$$F^i = dA^i + \frac12 \epsilon^{ijk} A^j \wedge A^k .$$9

becomes a Hamilton–Jacobi equation in coordinates with $C$00; in type II, an appropriate spacelike subleading diffeomorphism performs the same local gauge fixing. A plausible implication is that this is best regarded as an isometric notion only modulo gauge: the Galilei metric data are fixed, while auxiliary Bargmann data move within their equivalence class (Blanckenburg et al., 2024).

6. Gauge equivalence, covariance, and non-metric uses

Several adjacent literatures sharply delimit what should not be called an isometric gauge transformation. In BRST-antiBRST quantization, the finite transformations

$C$01

act on the enlarged field space of gauge fields, ghosts, antighosts, and auxiliary fields. For constant Grassmann parameters they form a 2-parametric Abelian supergroup with unit Jacobian; for field-dependent $C$02, the Jacobian implements a finite change of gauge-fixing functional. These transformations are explicitly gauge-related changes of variables in the path integral, not spacetime or manifold isometries (Moshin et al., 2014).

In Jacobi geometry, gauge transformations are defined by a $C$03-form $C$04 through the Dirac–Jacobi modification $C$05. When admissibility holds, the transformed Jacobi structure satisfies

$C$06

Such transformations preserve the characteristic distribution, induce isomorphic Lie algebroids, preserve the distinguished $C$07-cocycle and Lichnerowicz–Jacobi cohomology, and lift to contact groupoids, but no Riemannian metric-preservation condition is involved (Das, 2018).

In nonlinear gauge-coupled quantum fluids, the issue is form-invariance of the hydrodynamic equations and Galilean covariance. External $C$08 gauge functions preserve the canonical field equations, but density-dependent gauge functionals do not; the transformed phase equation acquires extra terms, and nonlinear gauge potentials cannot generally be gauged away. The symmetry notion is covariance of equations of motion, not metric isometry (Buggy et al., 2020).

A quantum-information variant appears in lattice gauge simulation. Local gauge transformations

$C$09

are unitary, hence norm-preserving on the full Hilbert space, and the ancilla-assisted filter

$C$10

acts exactly as the identity on the physical gauge-invariant subspace while suppressing unphysical amplitudes by destructive interference. This is an isometric effect on the physical sector, but not a geometric spacetime isometry (Ball, 2024).

Taken together, these cases indicate that “isometric gauge transformation” should be reserved for settings in which either the gauge map is itself a metric isometry, as in cone geometry, or a geometric isometry is characterized by gauge covariance, as in the pure-connection formalism. In many other areas, gauge transformations preserve gauge classes, flatness conditions, foliations, or equations of motion rather than any literal metric structure.