Strong Gap Condition

• Strong gap condition is a set of quantitative hypotheses that enforce sufficient separation in various mathematical domains, enabling explicit decoupling or rigidity.
• It appears in contexts like adiabatic spectral theory and parabolic PDEs, where it upgrades qualitative isolation to specific decay rates and higher integrability.
• The condition is also pivotal in combinatorial embeddings, isovariant topology, and stochastic optimization, establishing well-partial-orderings and connectivity thresholds.

“Strong gap condition” does not designate a single invariant definition across the literature. In current arXiv usage, the expression names several non-equivalent hypotheses that impose a quantitative separation: a uniform isolation radius for spectral subsets in adiabatic theory, a restriction on the exponent gap $q-p$ in parabolic double-phase regularity, an “outer-gap” constraint on embeddings of labelled trees or sequences, a codimension/connectivity range in isovariant topology, and closely related gap notions in stochastic saddle-point optimization, hyperbolic spectral theory, optimal control, and fourth-order geometric PDEs (Schmid, 2013, Sen et al., 9 Jun 2026, Freund, 2020, Bassily et al., 2023, Kelmer et al., 20 Mar 2026). This variety is substantive rather than terminological: the shared pattern is separation strong enough to force quantitative decoupling, higher integrability, high connectivity, or exclusion of unwanted spectral or combinatorial configurations.

1. Terminological range and recurrent structure

Across the cited works, “strong gap” is attached to different mathematical objects. In J. Schmid’s adiabatic theorem, it is a uniform spectral gap: $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$ In Sen–Siltakoski’s parabolic double-phase theory, it is the exponent restriction

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$

In Anton Freund’s reconstruction of Friedman’s tree embeddings, the strong gap condition is exactly $(\mathrm{G1})+(\mathrm{G2})$, constraining labels on nodes inserted between embedded images. In semifree isovariant Poincaré spaces, the Strong Gap Condition is the pair

$d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$

By contrast, in differentially private stochastic saddle-point problems the central object is the strong (primal–dual) gap, a performance criterion rather than a side condition (Schmid, 2013, Sen et al., 9 Jun 2026, Freund, 2020, Kirstein et al., 27 Oct 2025, Bassily et al., 2023).

Domain	Strong-gap formulation	Consequence emphasized
Adiabatic operator theory	$$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$	$O(\varepsilon)$ adiabatic decoupling
Parabolic double phase	Bounds on $q-p$ such as $q\le p+\frac{qκ}{q-2γ}$ or $q\le p+\alpha$	Local higher integrability of $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$0
Tree/sequence embeddings	Outer-gap constraints $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$1 or $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$2	WPO theorems and maximal order types
Isovariant topology	$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$3, $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$4	$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$5-connectivity of $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$6
Hyperbolic spectral theory	Exclusion of complementary series near $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$7	Exponential decay and mixing

This suggests a common usage: the adjective “strong” marks a regime in which a weaker asymptotic or existential statement becomes quantitative. In the adiabatic setting it upgrades convergence to $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$8; in double-phase flows it yields reverse Hölder estimates and $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$9-gain; in topology it upgrades nonemptiness to $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$0-connectivity; and in spectral theory it upgrades mere isolation of the base eigenvalue to an explicit spectral hole (Schmid, 2013, Sen et al., 9 Jun 2026, Kirstein et al., 27 Oct 2025, Kelmer et al., 20 Mar 2026).

2. Uniform spectral separation in adiabatic operator theory

For closed operators $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$1 on a complex Banach space $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$2, with $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$3 and compact spectral subsets $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$4, Schmid defines a uniform (or strong) spectral gap by the existence of a $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$5-independent $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$6 such that

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$7

Equivalently, $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$8 is isolated in $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$9 with a $(\mathrm{G1})+(\mathrm{G2})$0-independent isolation radius $(\mathrm{G1})+(\mathrm{G2})$1 (Schmid, 2013).

In the time-independent-domain theorem, the hypotheses are a common dense domain $(\mathrm{G1})+(\mathrm{G2})$2, closedness of each $(\mathrm{G1})+(\mathrm{G2})$3, generation of a strongly continuous evolution $(\mathrm{G1})+(\mathrm{G2})$4 for $(\mathrm{G1})+(\mathrm{G2})$5, and $(\mathrm{G1})+(\mathrm{G2})$6-stability in the sense that $(\mathrm{G1})+(\mathrm{G2})$7 for all $(\mathrm{G1})+(\mathrm{G2})$8 and all $(\mathrm{G1})+(\mathrm{G2})$9. One further assumes that each $d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$0 is compact and isolated, that $d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$1 is continuous in the Hausdorff sense, and that the Riesz projection

$d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$2

is well-defined with $d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$3 (Schmid, 2013).

The comparison evolution $d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$4 is defined as the unique solution of

$d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$5

The theorem yields

$d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$6

uniformly in $d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$7, and in particular

$d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$8

Under higher regularity, with $d^G + 3 \le d^e,\qquad k \le d^e - 2d^G - 3.$9 and $$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$0 of class $$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$1, one constructs nested projections $$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$2 and an evolution $$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$3 satisfying $$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$4 (Schmid, 2013).

The proof mechanism is organized around three ingredients. First, the uniform gap makes the contour formula for $$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$5 operator-norm convergent and transfers $$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$6-regularity from $$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$7 to $$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$8. Second, one solves a commutator equation using

$$\mathrm{dist}(\sigma(t),\sigma(A(t))\setminus\sigma(t))\ge\delta$$9

with $O(\varepsilon)$0. Third, a Duhamel/Grönwall argument absorbs the commutator remainder into the $O(\varepsilon)$1 estimate (Schmid, 2013).

A recurrent misconception is that adiabatic theorems of this type require isolated eigenvalues. Schmid’s strong-gap theorem only requires the spectral subsets $O(\varepsilon)$2 to be compact; “in particular, they need not consist of eigenvalues.” The strong gap is therefore a statement about spectral isolation rather than point spectrum alone. Another important distinction is between the uniform and non-uniform gap cases: Theorem 3.2 allows finitely many points where the gap closes and still gives

$O(\varepsilon)$3

but no explicit $O(\varepsilon)$4-rate is claimed (Schmid, 2013).

3. Gap bounds in parabolic double-phase regularity

In parabolic double-phase equations, the term “strong gap condition” refers to restrictions on the exponent difference between the $O(\varepsilon)$5-phase and the $O(\varepsilon)$6-phase. Sen–Siltakoski consider

$O(\varepsilon)$7

with $O(\varepsilon)$8, $O(\varepsilon)$9, and $q-p$0, where

$q-p$1

and

$q-p$2

For $q-p$3, $q-p$4, the main result is obtained under

$q-p$5

The paper describes this as the first purely parabolic “strong gap” in double-phase regularity theory (Sen et al., 9 Jun 2026).

The coefficient class $q-p$6 is used to encode the vanishing rate of $q-p$7 near its zero set; in particular, near any zero of $q-p$8,

$q-p$9

The proof uses a time mollification with spatial shift, the slanted Steklov average: $q\le p+\frac{qκ}{q-2γ}$0 and similarly $q\le p+\frac{qκ}{q-2γ}$1. These averages satisfy

$q\le p+\frac{qκ}{q-2γ}$2

The spatial shift $q\le p+\frac{qκ}{q-2γ}$3 is chosen so that the additional term in the time derivative cancels against the slant of the test function, allowing time mollification without stronger time regularity of $q\le p+\frac{qκ}{q-2γ}$4 (Sen et al., 9 Jun 2026).

The derivation of $q\le p+\frac{qκ}{q-2γ}$5 is based on intrinsic cylinders, a Caccioppoli inequality, and a balance between the Hölder control of $q\le p+\frac{qκ}{q-2γ}$6 and the vanishing profile $q\le p+\frac{qκ}{q-2γ}$7. In the $q\le p+\frac{qκ}{q-2γ}$8-phase, the estimates force

$q\le p+\frac{qκ}{q-2γ}$9

The paper emphasizes that this bound is strictly stronger than the elliptic Lavrentiev gap $q\le p+\alpha$0 whenever $q\le p+\alpha$1, because

$q\le p+\alpha$2

It also notes that as $q\le p+\alpha$3, the right-hand side blows up, reflecting that near $q\le p+\alpha$4 the equation is uniformly parabolic (Sen et al., 9 Jun 2026).

Kim–Oh study the same prototype flow under $q\le p+\alpha$5, $q\le p+\alpha$6, and formulate two related gap bounds. For bounded solutions $q\le p+\alpha$7, Theorem 1.1 assumes the “strong gap”

$q\le p+\alpha$8

and proves that for every parabolic cylinder $q\le p+\alpha$9 there exist $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$00 and $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$01 such that

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$02

hence $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$03. For $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$04, $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$05, Theorem 1.2 assumes

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$06

which interpolates between the earlier $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$07-based parabolic bound and the bounded case: $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$08 The proof excludes a third intrinsic regime precisely by boundedness or by the $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$09 assumption combined with the stated gap inequality (Kim et al., 17 Nov 2025).

A persistent theme in this literature is that the parabolic gap is not merely an algebraic compatibility condition. It is the threshold at which the intrinsic Calderón–Zygmund or reverse Hölder machinery closes. Sen–Siltakoski formulate this in terms of the slanted Steklov framework and minimal time regularity, while Kim–Oh formulate it in terms of intrinsic stopping-time cylinders and the exclusion of an “impossible” regime (Sen et al., 9 Jun 2026, Kim et al., 17 Nov 2025).

4. Outer-gap embeddings in combinatorics and proof theory

In finite labelled trees, Freund defines an embedding of $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$10-trees

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$11

as an injective order-preserving map $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$12 preserving meets and labels, together with the gap conditions. The strong gap condition is:

• $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$13 whenever $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$14 is an immediate successor of $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$15 and

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$16

then

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$17

• $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$18 whenever $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$19, then

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$20

The paper states succinctly: Strong gap = $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$21 (Freund, 2020).

Freund’s main structural claim is that this strong gap order is reconstructed from iterated applications of the uniform Kruskal theorem. Starting from $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$22, one forms the multiset dilator $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$23, takes the Kruskal derivative $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$24 of $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$25, and then sets

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$26

Proposition 6.3 and Proposition 5.3 identify

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$27

with order exactly the strong gap condition. Under $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$28-comprehension, each normal WPO-dilator admits a WPO-derivative, and therefore $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$29 is a well-partial-order for every fixed $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$30 (Freund, 2020).

For sequences over a well-order $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$31, Uftring’s exposition distinguishes the weak gap order $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$32 from the strong gap condition $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$33, also called the outer-gap condition. If

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$34

then $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$35 requires a strictly increasing embedding $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$36 such that $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$37 for all $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$38, the weak-gap inequalities hold in every internal gap $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$39, and additionally

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$40

Equivalently, $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$41 admits a recursion with concatenation $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$42 and an outer-gap clause: if $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$43 and

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$44

then

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$45

Over $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$46, the theorem states that the following are equivalent:

1. arithmetical transfinite recursion $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$47;
2. for every well-order $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$48, the strong gap order $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$49 is a well-partial-order (Uftring, 29 Jul 2025).

The same paper computes the maximal order type: $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$50 while the weak and symmetric variants satisfy

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$51

The strict inequality between weak and strong gap orders is therefore not only combinatorial but proof-theoretic: the strong gap ordering has the larger maximal order type $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$52 (Uftring, 29 Jul 2025).

These two lines of work clarify a common misconception. The “gap” is not a secondary decoration on ordinary embeddings; it changes the induced partial order so substantially that both the WPO strength and the associated ordinal analysis jump. Freund’s reconstruction explains the tree version as a canonical output of uniform Kruskal iteration, while Uftring quantifies the corresponding sequence order by explicit Veblen-hierarchy functions (Freund, 2020, Uftring, 29 Jul 2025).

5. Codimension and connectivity in semifree isovariant topology

For a semifree $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$53-Poincaré space $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$54, with

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$55

the paper “Semifree Isovariant Poincaré Spaces and the Gap Condition” defines the Strong Gap Condition as the pair of inequalities

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$56

where $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$57 measures the desired connectivity. The first inequality is the codimension $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$58 hypothesis; the second is usually called the (strong) gap hypothesis (Kirstein et al., 27 Oct 2025).

An isovariant structure on $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$59 is a homotopy pushout square in $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$60-spaces

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$61

such that $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$62 is an equivariant spherical fibration, $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$63 and $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$64 carry free $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$65-actions, and the underlying pair $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$66 is a non-equivariant Poincaré pair. The space of such structures is denoted $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$67 (Kirstein et al., 27 Oct 2025).

The main result, Theorem A, states: let $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$68 be a semifree $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$69-Poincaré space and $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$70 a periodic finite group. If integers $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$71 satisfy the two gap inequalities above, then

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$72

is $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$73-connected. In particular, for $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$74 it is nonempty. The proof has two steps. The first is destabilisation, where the stable normal bundle

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$75

is lifted to an honest equivariant spherical fibration of fibrewise dimension $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$76. The second is the construction of the complement $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$77 via an equivariant complement problem, reduced on fixed points to Klein’s non-equivariant Poincaré embedding theorem. The inequalities $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$78 and $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$79 are used in both steps (Kirstein et al., 27 Oct 2025).

The paper explicitly contrasts this with a weak gap condition that only assumes

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$80

That weaker range suffices for some non-equivariant Poincaré arguments or to build isovariant maps, but does not yield quantitative connectivity estimates of the moduli space of structures. The strong gap is therefore the range in which one can prove that $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$81 is highly connected rather than merely nonempty (Kirstein et al., 27 Oct 2025).

Applications include a Browder–Straus-type theorem for closed semifree smooth $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$82-manifolds $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$83 with

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$84

where equivariant homotopy equivalences lift to isovariant ones, and an aspherical-manifold application in which the theorem produces an isovariant structure on $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$85 when

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$86

In the semifree setting one also shows that $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$87 must have periodic cohomology, so that free generalised $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$88-homotopy representations exist in all large dimensions (Kirstein et al., 27 Oct 2025).

6. Strong gap as objective or no-gap criterion in optimization and control

In stochastic saddle-point optimization, the relevant notion is the strong (primal–dual) gap

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$89

as opposed to the weak gap

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$90

By Jensen’s inequality,

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$91

but the paper exhibits a simple one-dimensional bilinear example with weakgap $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$92 and gap $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$93. This shows that controlling weak gap does not guarantee small strong gap (Bassily et al., 2023).

For convex–concave $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$94-Lipschitz stochastic saddle-point problems, the paper proves an $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$95-DP algorithm achieving

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$96

which is nearly optimal up to logarithmic factors. The construction uses a recursive-regularization framework over $$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$97 stages with regularized subproblems

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$98

and yields gradient complexity

$$\exists\,\delta>0:\quad \forall t\in[0,1],\quad \mathrm{dist}\bigl(\sigma(t),\;\sigma\bigl(A(t)\bigr)\setminus\sigma(t)\bigr)\;\ge\;\delta.$$99

in the non-smooth case and $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$00 in the smooth case (Bassily et al., 2023).

In state-constrained optimal control, the operative notion is the relaxation gap

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$01

The paper “Sufficient conditions for the absence of relaxation gaps in state-constrained optimal control” gives three sufficient no-gap regimes. The classical Vinter condition assumes convexity of $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$02 and convexity of $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$03, yielding

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$04

Section 3 replaces convexity by the Filippov–Ważewski conditions FW1–FW2 on

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$05

again proving

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$06

Section 4 replaces these by an inward-pointing condition H1–H2 and obtains the same conclusion. Finally, with compact inner approximations

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$07

the paper derives the explicit upper bound

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$08

Here the role played by a “strong gap condition” is the complete elimination of the difference between original and relaxed values (Augier et al., 17 Mar 2025).

These two contexts illustrate two different uses of “gap.” In saddle-point optimization, the strong gap is the quantity minimized by the algorithm. In optimal control, the central issue is whether a relaxation gap exists at all, and the technical conditions are sufficient criteria for its vanishing (Bassily et al., 2023, Augier et al., 17 Mar 2025).

7. Spectral and geometric gap phenomena

For geometrically finite hyperbolic manifolds, Kelmer–Khalil–Sarkar define a strong spectral gap for $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$09, where $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$10 and $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$11 is Zariski-dense and geometrically finite with critical exponent $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$12. The definition has two parts:

1. no non-trivial quasi-complementary series $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$13 with $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$14 is contained at $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$15;
2. there is $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$16 such that for every self-dual $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$17-type $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$18 and every

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$19

the representation $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$20 is not weakly contained in $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$21.

The main theorem states that under $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$22, $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$23 has a strong spectral gap. Equivalently, for the orthogonal complement $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$24 of all complementary-series summands at $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$25, there exists $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$26 such that

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$27

This yields explicit decay of matrix coefficients and exponential mixing of the frame flow with rate exactly $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$28 (Kelmer et al., 20 Mar 2026).

The proof compares two meromorphic continuations of the Laplace transform

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$29

On the representation-theoretic side, possible poles come from complementary or quasi-complementary summands $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$30; on the dynamical side, exponential mixing shows that there is only one simple pole at $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$31. The absence of extra poles in the overlap strip forces the relevant residues to vanish and excludes the unwanted summands (Kelmer et al., 20 Mar 2026).

A different type of gap phenomenon appears in fourth-order geometric PDEs on immersed surfaces with boundary. Wheeler studies immersions $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$32 satisfying

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$33

under either umbilic boundary conditions

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$34

or flat boundary conditions

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$35

In the umbilic-boundary case, if

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$36

then there is a universal $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$37 such that

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$38

In the almost-flat case, if

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$39

then small $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$40 again forces $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$41. Under flat boundary, the conclusion strengthens to planarity, and one may allow $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$42 (Wheeler, 2013).

The analytic structure is an absorption argument: one multiplies the PDE by $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$43, integrates by parts, controls lower-order curvature terms using a boundary Michael–Simon Sobolev inequality and multiplicative Sobolev estimates, and uses the smallness of $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$44 or $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$45 to force

$2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$46

The paper emphasizes that the threshold $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$47 is universal, depending only on $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$48, not on the topology or boundary of $2 \le p \le q \le p + \frac{qκ}{q - 2γ}.$49 (Wheeler, 2013).

Taken together, these spectral and geometric works show that “strong gap” can refer either to exclusion of a whole band of non-tempered representations below a critical exponent or to a universal smallness threshold that rigidifies a high-order curvature equation. In both cases, the strength of the gap lies in turning qualitative rigidity into quantitative decay or exact classification (Kelmer et al., 20 Mar 2026, Wheeler, 2013).