Skip to content

Harish-Chandra's regularity theorem

An invariant eigendistribution on a real semisimple Lie algebra, or in the corresponding real-group formulation, has a locally integrable representative analytic on the regular locus.

Version
v1 · 2026-10-31 · History
Domain-specific #
14507
Domain group
Formal Sciences
Origin domain
Mathematics
Subdomains
Representation Theory, Harmonic Analysis, Lie Theory → Mathematics

Core Idea

Harish-Chandra's regularity theorem says that, under the relevant invariance and operator eigenconditions, a distribution on a real semisimple Lie algebra has a locally integrable representative analytic on its regular set. A representation-theory survey states the corresponding scalar-center result for real groups. The algebra and group have different carriers and operator actions; the conclusion does not assert analyticity at every singular point.[1][2]

How would you explain it like I'm…

 

No faithful explanation at this level. All three generators judged any five-year-old story reduces to 'messy things are secretly smooth', erasing the distribution-to-function upgrade and falsely implying analyticity everywhere rather than only on the regular locus.

Fuzzy Objects That Are Secretly Functions

In math, a function is a rule that gives a number at each point. Mathematicians also use fuzzier objects that can only be studied by averaging over regions. Harish-Chandra's regularity theorem says that if one of these fuzzy objects is very symmetric and behaves simply under certain special operations, it is really an ordinary function after all. Even better, that function is perfectly smooth at most points — the 'regular' ones — though maybe not at some special points.

Regularity of Invariant Eigendistributions

Distributions are generalized functions: they're defined by how they act when averaged against smooth test functions, and they can be much rougher than ordinary functions. Harish-Chandra's regularity theorem says that, under strong symmetry and operator conditions, such a distribution is actually given by an ordinary function. In its original version on a real semisimple Lie algebra, a distribution that is unchanged by the adjoint action and is an eigendistribution of all invariant constant-coefficient differential operators is represented by a locally integrable function. That function is analytic (extremely smooth) on the set of regular points, but the theorem doesn't claim it's analytic at singular points. A similar version holds on the real group for conjugation-invariant distributions where the center acts by scalars.

 

Harish-Chandra's regularity theorem converts certain distributions into functions under strong symmetry and differential-operator hypotheses. In the original real semisimple Lie algebra formulation, a distribution that is invariant under the adjoint action and is an eigendistribution of every invariant constant-coefficient differential operator is represented by a locally integrable function. That function is analytic on the regular set. The theorem does not assert analyticity at the singular points. A corresponding real-group formulation, as stated in a representation-theory survey, says that a conjugation-invariant distribution on the group on which the relevant center acts by scalars has a locally integrable representative that is analytic on the regular elements. The two formulations differ in the underlying space and in which operators act. The survey attributes the group result to Harish-Chandra; the original general-group announcement and proof were identified bibliographically but their exact statements were not checked here.

How It Works

On the algebra, adjoint invariance is paired with an eigencondition for invariant constant-coefficient differential operators. The original paper also states a broader, separately qualified reductive-algebra version using a finite-codimensional operator ideal. On the group, the inspected survey requires conjugation invariance and scalar action of the center of the complexified enveloping algebra. Under their respective hypotheses, the distribution is represented by a locally integrable function and is analytic on the regular locus. The original 1963 announcement and 1965 full group paper are bibliographically identified, but their exact theorem text remains for library verification.[1][2][3][4]

Examples

A derived, elementary algebra instance takes the constant-density distribution on the traceless real 2-by-2 matrices: the adjoint action preserves its Lebesgue measure, and integrating a constant-coefficient operator against compactly supported tests leaves only its constant term. Its representative is the analytic constant function 1. This checks the hypotheses but does not illustrate the theorem's difficult control of singularities.[1]

In a restricted original construction, Harish-Chandra builds group-side eigendistributions for acceptable connected semisimple real groups with compact Cartan subgroup and a selected regular character. The constructed distribution is locally summable and analytic on regular group elements. The additional weighted bound and Cartan formula belong to this construction, not to every invariant eigendistribution.[5]

Boundaries

Adjoint invariance alone is insufficient: delta at zero is invariant but not represented by a locally integrable function and does not meet the full eigenoperator condition. This is a derived near miss, not an example quoted from Harish-Chandra. The theorem's real-algebra, surveyed real-group and restricted construction hypotheses must stay separate; p-adic analogues require separate sources. Neither original general-group paper was directly inspected, so exact original hypotheses, proof and any broader operator or character claims remain for later library checking.[1][3][4]

Neighborhood in Abstraction Space

Harish-Chandra's regularity theorem sits in a moderately populated region (45th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Lie Groups & Representation Theory (18 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-31

References

[1] Harish-Chandra, “Invariant eigendistributions on a semisimple Lie algebra,” Publications mathématiques de l'IHÉS 27 (1965), 5–54, introduction at printed p. 5 and Theorem 2 at printed p. 11. This original paper was directly inspected; some mathematical glyphs in text extraction are noisy. The elementary constant-density and delta examples above are derived checks, not examples attributed to the paper. registry ↩a ↩b ↩c ↩d

[2] Zijian Yao, Representation Theory of Real Groups, §2.2, Theorem 2.5 (PDF p. 7). This directly inspected scholarly survey states the scalar-center group formulation and attributes regularity to Harish-Chandra. It is not the original general-group theorem paper. registry ↩a ↩b

[3] Harish-Chandra, “Invariant eigendistributions on semisimple Lie groups,” Bulletin of the American Mathematical Society 69 (1963), 117–123. Bibliographic identity was checked, but this announcement’s theorem text was not directly inspected after reasonable access attempts. Library follow-up: verify its exact group hypotheses, quantifiers, regular-set description and any claims attributed specifically to the announcement. registry ↩a ↩b

[4] Harish-Chandra, “Invariant eigendistributions on a semisimple Lie group,” Transactions of the American Mathematical Society 119 (1965), 457–508. Bibliographic identity was checked, but this full group paper’s theorem text was not directly inspected after reasonable access attempts. Library follow-up: verify its exact hypotheses, proof, operator relaxation and any character formula before attributing those claims to this paper. registry ↩a ↩b

[5] Harish-Chandra, “Discrete series for semisimple Lie groups I: Construction of invariant eigendistributions,” Acta Mathematica 113 (1965), 241–318, introduction at printed p. 242, §19/Theorem 3 at pp. 288–289, regular-set corollary at p. 298 and Lemmas 54–55 at pp. 300–301. This original paper was directly inspected for the restricted construction; OCR formula glyphs were not independently collated against page images. registry ↩