Helffer-Nourrigat Cone in Hypoelliptic Analysis

• The Helffer-Nourrigat cone is a conic set defined via anisotropic blow-up in filtered manifolds that replaces the standard cotangent bundle for principal symbol analysis.
• It encapsulates operator-valued principal symbol theory, linking the injectivity of symbols on graded nilpotent Lie groups to maximal hypoellipticity.
• Its singular stratification and desingularization play pivotal roles in analyzing sub-Laplacians on Carnot manifolds and in refining microlocal hypoellipticity.

Searching arXiv for the cited papers to ground the article and citations. arxiv_search(query="Helffer Nourrigat cone maximal hypoellipticity principal symbol", max_results=10) arxiv_search(query="(Wang, 2024, Androulidakis et al., 2022, Goffeng et al., 2024, Cren, 12 Dec 2025, Mohsen, 29 Jan 2026, Mohsen, 20 Jan 2026, Louis, 1 Sep 2025)", max_results=10) The Helffer-Nourrigat cone is a conic object attached to Hörmander-type systems of vector fields, filtered manifolds, graded nilpotent Lie groups, Carnot manifolds, and singular foliations. In its now standard sub-Riemannian formulation, it is the closed subset of covectors, coadjoint orbits, or irreducible unitary representations that remains visible under the anisotropic blow-up determined by a filtration; in that sense it replaces the cotangent bundle $T^*M$ as the phase space relevant for hypoelliptic regularity and principal symbol theory (Androulidakis et al., 2022, Cren, 12 Dec 2025). Closely related literature also uses Helffer-Nourrigat cone language for conic parameter sets governing hypoelliptic sub-Laplacians on graded Lie algebras, while harmonic-analysis work on cone restriction invokes Helffer and Nourrigat in a different historical lineage associated with the geometric cone $\mathcal C \subset \mathbb R^n$ of Fourier restriction (Goffeng et al., 2024, Wang, 2024).

1. Definitions and scope

Let $M$ carry a Lie filtration $F^1 \subset F^2 \subset \dots \subset TM$ generated by vector fields satisfying Hörmander’s bracket-generating condition. At each $x\in M$, the associated graded Lie algebra is

$gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$

and the osculating group is $G_x := \exp(gr_x(F))$. The Helffer-Nourrigat cone arises from the limiting behavior of cotangent directions under the dilations dictated by the filtration, and only the corresponding representations of $G_x$ enter the principal symbol calculus (Androulidakis et al., 2022, Cren, 12 Dec 2025).

A pointwise definition used in the pseudodifferential-calculus literature is

$$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$

The global cone is then described as

$$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$

This construction identifies the relevant “directions at infinity” for hypoellipticity, not merely the commutative cotangent directions but also those created by higher commutators (Androulidakis et al., 2022, Cren, 12 Dec 2025).

A geometrically equivalent description emphasizes tangent cones. If $\mathcal C \subset \mathbb R^n$0 is the nilpotent Lie algebra generated by the model vector fields at $\mathcal C \subset \mathbb R^n$1, and $\mathcal C \subset \mathbb R^n$2 is the set of limit subalgebras arising from tangent cones, then

$\mathcal C \subset \mathbb R^n$3

This formulation makes explicit that the cone is the union of annihilators of all possible tangent Lie subalgebras, encoding the directions relevant for the symbol test for maximal hypoellipticity (Debord, 18 Apr 2025).

Context	Representative definition	Analytic role
Filtered/sub-Riemannian calculus	$\mathcal C \subset \mathbb R^n$4, $\mathcal C \subset \mathbb R^n$5, $\mathcal C \subset \mathbb R^n$6	Phase space for operator-valued principal symbols
Carnot sub-Laplacians	$\mathcal C \subset \mathbb R^n$7	Parameter cone for hypoellipticity and index questions
Cone restriction theory	$\mathcal C \subset \mathbb R^n$8	Geometric cone for Fourier restriction estimates

2. Operator-valued principal symbols and maximal hypoellipticity

For vector fields $\mathcal C \subset \mathbb R^n$9 on $M$0, possibly with weights $M$1, every differential operator $M$2 may be written as a noncommutative polynomial $M$3. The generalized principal symbol is defined at $M$4 and an irreducible representation $M$5 of the osculating group by

$M$6

where $M$7 is the top weighted-homogeneous part of $M$8 with coefficients frozen at $M$9. A key point is that this representation-theoretic symbol is well defined on the Helffer-Nourrigat cone $F^1 \subset F^2 \subset \dots \subset TM$0 (Androulidakis et al., 2022).

The central theorem proved in the modern calculus is the equivalence between invertibility of this symbol on the Helffer-Nourrigat cone and maximal hypoellipticity. Concretely, for a differential operator of Hörmander order $F^1 \subset F^2 \subset \dots \subset TM$1, the following are equivalent: for every $F^1 \subset F^2 \subset \dots \subset TM$2 and every nontrivial $F^1 \subset F^2 \subset \dots \subset TM$3, the symbol $F^1 \subset F^2 \subset \dots \subset TM$4 is injective on $F^1 \subset F^2 \subset \dots \subset TM$5; for all or some $F^1 \subset F^2 \subset \dots \subset TM$6, $F^1 \subset F^2 \subset \dots \subset TM$7 implies $F^1 \subset F^2 \subset \dots \subset TM$8; and, when $F^1 \subset F^2 \subset \dots \subset TM$9 is compact, $x\in M$0 is left-invertible modulo compact operators. This gives an affirmative resolution of the conjecture due to Helffer and Nourrigat (Androulidakis et al., 2022).

The same equivalence appears in a more abstract $x\in M$1-calculus. In the graded nilpotent group setting, one considers spaces

$x\in M$2

with $x\in M$3 the group dilations. For $x\in M$4, injectivity of $x\in M$5 and of the adjoint symbol in every nontrivial irreducible unitary representation is equivalent to topological isomorphism on Sobolev and distribution spaces and to the existence of a two-sided parametrix modulo smoothing. This packages the original Helffer-Nourrigat theorem and the Rockland criterion into a Type I $x\in M$6-algebra framework (Mohsen, 20 Jan 2026).

3. Microlocal refinement

The microlocal version introduces a cone depending on both a base point and a classical covector. For $x\in M$7, the Helffer-Nourrigat cone is

$x\in M$8

Here $x\in M$9 is the osculating graded nilpotent Lie algebra at $gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$0. This microlocal cone records all directions accessible by rescaling a fixed cotangent direction through the filtered geometry (Mohsen, 29 Jan 2026).

In the bi-graded calculus $gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$1, which contains both the classical pseudodifferential calculus and a calculus adapted to the sub-Riemannian structure, the principal symbol becomes a family of operators

$gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$2

indexed by irreducible unitary representations $gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$3 associated with $gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$4. The open set on which microlocal maximal hypoellipticity holds is

$gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$5

On $gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$6, the paper proves

$gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$7

together with a microlocal parametrix on every closed cone $gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$8. This establishes the microlocal version of the Helffer-Nourrigat conjecture (Mohsen, 29 Jan 2026).

A plausible implication is that the Helffer-Nourrigat cone should be viewed not merely as a global replacement for $gr_x(F) := \bigoplus_{j \geq 1} F^j_x / F^{j-1}_x,$9, but as the precise microlocal defect set where hypoellipticity can fail. That formulation is consistent with both the global theorem and its microlocal refinement (Androulidakis et al., 2022, Mohsen, 29 Jan 2026).

4. Topology, stratification, and $G_x := \exp(gr_x(F))$0-algebraic structure

The topology of the Helffer-Nourrigat cone is typically singular. Even for a fixed nilpotent group, the coadjoint-orbit quotient is usually not Hausdorff, and across a filtered manifold the dimensions of the osculating groups may vary. For that reason the cone is often too singular to support a direct geometric treatment comparable to the smooth cotangent bundle (Cren, 12 Dec 2025).

A desingularization is provided by extending the Pedersen-Pukanszky stratification of the unitary dual. One obtains a filtration by open sets

$G_x := \exp(gr_x(F))$1

with strata $G_x := \exp(gr_x(F))$2 that are locally compact Hausdorff. At regular points, these strata restrict to the Pedersen strata of the osculating group $G_x := \exp(gr_x(F))$3. The dilation action preserves the stratification (Cren, 12 Dec 2025).

This stratification controls the structure of the symbol algebra and the order-zero calculus. The relevant short exact sequence is

$G_x := \exp(gr_x(F))$4

and the spectrum of the principal symbol algebra satisfies

$G_x := \exp(gr_x(F))$5

Moreover, the associated ideals admit explicit subquotients

$G_x := \exp(gr_x(F))$6

where $G_x := \exp(gr_x(F))$7 is the compact-operator algebra on a separable Hilbert space, infinite-dimensional for $G_x := \exp(gr_x(F))$8 and $G_x := \exp(gr_x(F))$9 for $G_x$0. In this sense both $G_x$1 and $G_x$2 are solvable with explicit subquotients (Cren, 12 Dec 2025).

5. Sub-Laplacians on Carnot manifolds and cone invariance

In the Carnot-manifold literature the Helffer-Nourrigat cone appears as a conic set of coefficients for hypoelliptic sub-Laplacians. For

$G_x$3

with $G_x$4 spanning $G_x$5 and $G_x$6 spanning $G_x$7, one studies

$G_x$8

This set is described as a conic, star-shaped set in $G_x$9 (Goffeng et al., 2024).

The basic cone-invariance condition is property $$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$0: $$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$1 If this property holds at each point of a Carnot manifold, then any H-elliptic sub-Laplacian $$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$2 on a rank $$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$3 bundle has

$$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$4

The proof proceeds by contracting $$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$5 through H-elliptic operators to a self-adjoint operator, using the star-shapedness of the cone (Goffeng et al., 2024).

Several explicit descriptions of $$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$6 are available. In the scalar case $$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$7, if $$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$8,

$$T^*_pF=\{\xi\in \operatorname{gr}(F)^*_p : (p, \xi, 0) \in \overline{T^*M \times \mathbb{R}_+^\times} \subseteq a^*\}.$$9

For step $$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$0 with $$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$1,

$$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$2

These formulas show that, in many higher-step settings, the cone is genuinely dilation invariant; by contrast, on contact or polycontact manifolds the cone is not star-shaped in the same way, and nontrivial index theory can occur (Goffeng et al., 2024).

6. Singular foliations, Nash blowups, and symplectic leaves

For a singular foliation $$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$3 presented by an anchored bundle $$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$4, the Helffer-Nourrigat cone is defined by

$$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$5

This realizes the cone as the closure of the regular cotangent images under the dual anchor. The Nash blowup produces a Nash algebroid $$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$6, and there is an injective vector bundle morphism

$$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$7

whose image is exactly $$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$8. In this formulation the cone is independent of the chosen anchored bundle (Louis, 1 Sep 2025).

When the relevant algebroids carry their canonical linear Poisson structures, the Helffer-Nourrigat cone is a union of symplectic leaves. This is proved by showing that Hamiltonian flows preserve the images $$HN(F) := \bigsqcup_{x \in M} \frac{T^*_xF}{Ad^*(G_x)} \subset \bigsqcup_{x \in M} \widehat{G_x}, \qquad HN_0(F)=HN(F)\setminus\text{trivial rep}.$$9 and hence preserve their closure. The result gives a Poisson-geometric interpretation of the cone that is compatible with the Nash-blowup description (Louis, 1 Sep 2025).

The same framework yields a notion of longitudinal ellipticity for differential operators on singular foliations. A longitudinal differential operator $\mathcal C \subset \mathbb R^n$00 of degree $\mathcal C \subset \mathbb R^n$01 is longitudinally elliptic precisely when its symbol is strictly positive on $\mathcal C \subset \mathbb R^n$02 away from the zero section, equivalently on $\mathcal C \subset \mathbb R^n$03. Thus, as in the filtered-manifold calculus, the correct symbolic domain is not all of $\mathcal C \subset \mathbb R^n$04, but the Helffer-Nourrigat cone (Louis, 1 Sep 2025).

7. Relation to cone restriction theory

The phrase “Helffer-Nourrigat cone” also appears in discussions of the cone restriction problem, but here the underlying cone is the geometric Fourier-restriction cone

$\mathcal C \subset \mathbb R^n$05

The corresponding conjecture asks for sharp $\mathcal C \subset \mathbb R^n$06 bounds for the cone extension operator

$\mathcal C \subset \mathbb R^n$07

In this setting Helffer and Nourrigat enter as part of the historical development of the cone restriction conjecture, rather than through the representation-theoretic phase space described above (Wang, 2024).

Recent progress revisits the polynomial-partitioning method of Ou and Wang. By restructuring their induction into a recursive algorithm and using the nested polynomial Wolff axioms, Wang obtained for $\mathcal C \subset \mathbb R^n$08

$\mathcal C \subset \mathbb R^n$09

whenever

$\mathcal C \subset \mathbb R^n$10

The paper emphasizes that this improves the previously best high-dimensional range and that the nested polynomial Wolff axioms are decisive for handling the cone geometry (Wang, 2024).

This usage should therefore be distinguished from the sub-Riemannian and filtered-calculus Helffer-Nourrigat cone. In the latter, the cone is the noncommutative phase space governing principal symbols and maximal hypoellipticity; in the former, the cone is the hypersurface $\mathcal C \subset \mathbb R^n$11 on which one studies Fourier restriction.