Conformal Stability Criterion Overview

• Conformal Stability Criterion is a unifying concept that characterizes stability via invariant measures (e.g., integrated Lyapunov suppression, second-variation conditions, or Weyl functional positivity) across multiple disciplines.
• It is applied in various domains such as cosmology, nonlinear control, geometric analysis, and conformal prediction to ensure attractor behavior, robust finite-horizon stability, or reduced variability under conformal transformations.
• Practical implications include consistent stability across conformally related frames in inflationary models, reliable safety in data-driven control systems, and rigorous stability criteria in black-hole perturbation theory and minimal surface analysis.

The expression conformal stability criterion has no single universal meaning across the literature. In cosmology it denotes the frame-independence of attractor character under conformal transformations, captured by the invariance of integrated Lyapunov suppression, $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$ (Nandi, 2019). In data-driven control it denotes a Lyapunov inequality robustified by a conformal-prediction error quantile, yielding finite-horizon exponential stability and safety with coverage $1-\delta$ (Hsu et al., 6 Jun 2025). In geometric analysis it denotes second-variation or spectral conditions forcing weak conformality or conformal-biharmonic stability (Kawai et al., 2012, Branding et al., 22 Apr 2026). In black-hole perturbation theory it denotes positivity of a conformally invariant Weyl functional (Haddad, 26 Aug 2025), while in conformal prediction it denotes stability conditions on adaptive selection or on calibration-sample variability of set size (Hegazy et al., 25 Jun 2025, Min et al., 2 May 2026).

1. Terminological range and recurring structure

Domain	Criterion	Consequence
Cosmology	$$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$	attractor character is frame-independent
Nonlinear control	$$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$	finite-horizon exponential stability with probability $1-\delta$
Geometric analysis	stable C-stationary or c-biharmonic second variation conditions	weak conformality or index comparison
Conformal prediction	$(\eta,\tau)$-stable selection or set-stability variance	valid selection or reduced variability
Black-hole and horizon geometry	positivity of $\mathcal F[E,B]$ or sign conditions on CKV divergence	mode stability or MOTS stability

A common pattern is that the phrase couples a conformal transformation, conformal symmetry, or conformal invariant to a field-specific notion of stability: Lyapunov decay, second variation, spectral positivity, hyperbolic Cauchy stability, or calibration-sample robustness. This suggests that the expression is best understood as a family of criteria rather than a single theorem.

2. Cosmological frame criteria

In conformally connected cosmological frames, the central question is whether a stable background solution in one frame remains stable in the conformally related frame. For single-field models, curvature and tensor perturbations are invariant under conformal transformations, but the background equations differ, so the decay rates of deviations from fixed points need not coincide. The key result is that the attractor character is frame-independent, while the decay rate per e-fold is frame-dependent because the e-fold “clock” changes under the conformal map (Nandi, 2019).

The criterion is expressed by the equality of the integrated Lyapunov exponent in the two frames. In its discrete form,

$$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$

and in its continuous form,

$$\int_{N_0}^{N}\lambda(N')\,dN' = \int_{\tilde N_0}^{\tilde N}\tilde{\lambda}(\tilde N')\,d\tilde N'.$$

The point is not equality of $\lambda$ and $1-\delta$0 themselves, but equality of the exponent governing suppression of perturbations. A decay in one frame therefore maps to a decay in the other, provided the conformal transformation is regular and monotonic in time (Nandi, 2019).

The paper develops this explicitly for power-law Brans–Dicke cosmology and its Einstein-frame image. In the power-law case, the e-fold intervals satisfy

$1-\delta$1

while more generally

$1-\delta$2

This explains why naively comparing $1-\delta$3 and $1-\delta$4 can give an apparently contradictory sign structure: the correct invariant object is $1-\delta$5, not $1-\delta$6 alone (Nandi, 2019).

In inflationary examples, the same logic applies to Starobinsky inflation and chaotic inflation. The solutions remain attractors in both Jordan and Einstein frames, but the number of e-folds needed to approach slow roll differs. The paper states that the duration of inflation in the Jordan frame is always higher than the Einstein frame, even though the physical attractor behavior is equivalent (Nandi, 2019).

3. Relativistic conformal methods and Einstein constraints

A second major use of the term appears in conformal formulations of the Einstein equations, where stability means that a conformal evolution system remains regular up to a conformal boundary and that small perturbations of conformal initial data lead to nearby physical spacetimes. For tracefree radiation fluids, the conformal Einstein equations admit a symmetric hyperbolic reduction, and small perturbations of de Sitter-like FLRW data produce solutions that exist globally towards the future and are future geodesically complete (Lübbe et al., 2011). For Einstein spaces with spatial sections of negative scalar curvature, the extended conformal Einstein field equations together with conformal Gaussian systems turn the problem of future global existence into a finite-time Cauchy problem for a symmetric hyperbolic system; small Sobolev perturbations yield a smooth space-like conformal boundary and future geodesic completeness (Minucci et al., 2021).

In the conformal constraint equations of general relativity, stability is used in an elliptic sense. For the Einstein–Lichnerowicz conformal constraint system on closed locally conformally flat manifolds, the criterion involves focusing data, coercivity, and genericity assumptions. The key focusing coefficient is

$1-\delta$7

When $1-\delta$8, $1-\delta$9, and, in dimensions $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$0, $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$1 and $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$2 have no common critical points, sequences of solutions are uniformly bounded and converge modulo conformal Killing 1-forms; explicit instability constructions show that the result is sharp when these assumptions fail (Premoselli, 2015).

Maxwell’s drift formulation of the conformal method gives a related criterion for small data. In dimensions $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$3, on a closed locally conformally flat manifold, if the volumetric drift is small and the scalar field potential is sufficiently large, then solutions of the drift-model conformal system satisfy uniform a priori bounds,

$$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$4

and the solutions are stable under perturbations of the coefficients (Vâlcu, 2020).

4. Geometric analysis, rigidity, and compatibility

In geometric analysis, the phrase often refers to second-variation criteria forcing conformality. For a smooth map $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$5, the conformality tensor

$$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$6

vanishes exactly when $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$7 is weakly conformal, and the associated functional is

$$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$8

A map is C-stationary when it is a critical point of $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N$$9, and stable when the second variation is nonnegative for every variation. The main rigidity result is that if $$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$0 is a stable C-stationary map from $$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$1 with $$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$2, or into $$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$3 with $$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$4, then $$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$5 is weakly conformal. The proof reduces the stability condition to

$$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$6

forcing $$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$7 in dimensions at least five (Kawai et al., 2012).

A related variational use appears for the identity map of compact Einstein manifolds. The conformal-bienergy functional $$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$8 leads to a fourth-order Jacobi operator, and for the identity map on an Einstein manifold one has

$$\frac{\partial V}{\partial x}\hat f(x,u) + c_3V(x) + \left\|\frac{\partial V}{\partial x}\right\|q_1 \le 0$$9

The conformal-biharmonic index coincides with the harmonic index except for the four-dimensional round sphere. In that exceptional case, $1-\delta$0, so the identity map is unstable for the classical energy but stable for the conformal-bienergy functional (Branding et al., 22 Apr 2026).

For free boundary minimal submanifolds in conformal domains, the stability criterion takes the form of a nonexistence theorem. If $1-\delta$1 is conformal to a bounded $1-\delta$2-convex Euclidean domain, has non-negative sectional curvature, and either strictly positive sectional curvature or strict $1-\delta$3-convexity of the boundary, then there are no compact stable free boundary minimal submanifolds of dimension

$1-\delta$4

Here stability is again second-variation stability, and the proof uses conformally rescaled constant vector fields as test sections (Carvalho et al., 2024).

A structurally different usage appears in the compatibility problem for conformal and projective structures. Given a conformal class $1-\delta$5 and a projective structure $1-\delta$6, local compatibility is equivalent to the tensorial conditions

$1-\delta$7

and

$1-\delta$8

These are necessary and sufficient for the existence of a metric $1-\delta$9 whose Levi-Civita connection lies in the given projective class, and the compatible metric is unique up to multiplication by a constant (Matveev et al., 2013). This is a compatibility criterion rather than a dynamical one, but the literature supplied here explicitly treats it as a conformal stability or compatibility criterion.

5. Control, conformal prediction, and computational stability

In data-driven nonlinear control, the criterion is explicitly Lyapunov-theoretic and statistical. For an unknown continuous-time nonlinear system $(\eta,\tau)$0 with learned model $(\eta,\tau)$1, model uncertainty $(\eta,\tau)$2 is quantified by a split conformal quantile $(\eta,\tau)$3 constructed from trajectory-wise nonconformity scores. A conformally robust control Lyapunov function satisfies

$(\eta,\tau)$4

and

$(\eta,\tau)$5

Under the conformal error event, this yields

$(\eta,\tau)$6

and the corresponding norm-decay estimate for $(\eta,\tau)$7. The same framework also yields a conformally robust control barrier function for safety (Hsu et al., 6 Jun 2025).

In conformal prediction, one usage concerns valid selection among multiple conformal sets. If several conformal predictors $(\eta,\tau)$8 are valid individually, selecting the smallest one can destroy coverage. The remedy is a stability condition on the selection rule. If the selection algorithm is $(\eta,\tau)$9-stable, then the selected set satisfies

$\mathcal F[E,B]$0

so coverage is preserved after adjusting the base miscoverage. The paper formulates this directly through MinSE and AdaMinSE, which optimize expected set size subject to the stability constraints (Hegazy et al., 25 Jun 2025).

A second statistical usage defines stability as a variance functional of the calibration data. For a prediction set $\mathcal F[E,B]$1, let

$\mathcal F[E,B]$2

The stability criterion is then

$\mathcal F[E,B]$3

Standard conformal prediction has

$\mathcal F[E,B]$4

while Stable Conformal Prediction via transduction attains

$\mathcal F[E,B]$5

depending on the curvature regime of the alignment objective (Min et al., 2 May 2026).

In computational geometry, stability is interpreted algorithmically. Stable Discrete Minimization of Conformal Energy minimizes

$\mathcal F[E,B]$6

for disk conformal parameterization, and the stability conditions are geometric: nonnegative discrete conformal energy, area close to the target area, boundary orientation preservation, and positive triangle Jacobians. These conditions ensure one-to-one, on-to, and folding-free parameterizations and insensitivity to poor initial values (Tan et al., 2022).

6. Black-hole perturbations and horizon geometry

In rotating black-hole perturbation theory, the conformal stability criterion is based on a conformally invariant functional built from the electric and magnetic parts of the Weyl tensor. With

$\mathcal F[E,B]$7

the central functional is

$\mathcal F[E,B]$8

The paper states the criterion as follows: a black hole spacetime is mode-stable if and only if

$\mathcal F[E,B]$9

for all perturbations satisfying the linearised field equations. The same framework also gives an isospectrality theorem for conformally related black holes (Haddad, 26 Aug 2025).

For marginally outer trapped surfaces, the criterion is instead expressed through a conformal Killing vector field in the normal bundle. If $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$0 is a normal CKV and $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$1 is a MOTS, then under the construction used in the paper one obtains

$$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$2

where $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$3 is the MOTS stability operator and

$$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$4

Under the null energy condition, a past-pointing CKV with $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$5 yields strict stability and smooth evolution to a spacelike horizon. The paper also shows that if the restriction of the divergence of the vector field to $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$6 is non-negative, $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$7 is unstable, and if negative and $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$8 is a 2-sphere, $$\lambda\,\Delta N = \tilde{\lambda}\,\Delta\tilde N,$$9 must be strictly stable (Sherif, 3 Jul 2026).

Taken together, these formulations show that the phrase conformal stability criterion marks a recurring research strategy: encode the relevant dynamics, variational structure, or uncertainty quantification in quantities that transform naturally under conformal rescaling, and then state stability in terms of an invariant product, a second variation, a spectral sign condition, or a variance bound. The specific content changes radically from one field to another, but the structural role of conformal geometry as a stability-organizing principle is common to all of them.