Finite arithmetic in every degree of the local sphere

← All digests

Family 313 · Topology
Importance 5/5 · Provisional editorial rating
Sources consulted 2026-10-22

1. The question

Localization can make a mathematical object easier to study by focusing on one kind of information. It can also turn a finite object into something with infinitely many layers. The sphere spectrum illustrates both effects: it is the basic unit of stable homotopy theory, yet its chromatic localizations have homotopy in arbitrarily negative as well as positive degrees.

Family 313 asks whether each individual degree nevertheless contains only finitely much arithmetic data. At a prime $p$ and height $n$, is every homotopy group of the local sphere finitely generated over the $p$-adic integers?

M1 claims an affirmative answer for every prime, every positive height, and every integer degree. It does not calculate all these groups. Instead, it asserts that each consists of finitely many $p$-adic directions together with finite torsion.

2. Background and essential definitions

The stable homotopy group of spheres in degree $t$ is obtained by repeatedly suspending maps between spheres until suspension no longer changes the answer. A spectrum organizes this stable information. The sphere spectrum $S$ is the unit for the smash product of spectra.

For a fixed prime $p$, Morava $K$-theory $K(n)$ detects chromatic height $n$. One can think of heights as different periodic layers in stable homotopy theory. The localization $L_{K(n)}X$ retains precisely the information in $X$ seen by $K(n)$. Localization need not preserve connectivity, so negative-degree homotopy matters here.

The ring $\mathbb Z_p$ consists of compatible systems of residues modulo $p,p^2,p^3,\ldots$. Its fraction field is $\mathbb Q_p$. A finitely generated $\mathbb Z_p$-module has the form

$$\mathbb Z_p^{\,r}\oplus\bigoplus_{i=1}^{s}\mathbb Z/p^{a_i},$$

for finite integers $r,s\geq0$ and positive $a_i$. “Finitely generated” therefore does not mean a finite set: even one copy of $\mathbb Z_p$ is infinite. The theorem instead bounds the number of independent $p$-adic directions and makes the torsion subgroup finite in each degree.

A finite $p$-local spectrum is built from finitely many cells, allowing retracts, with primes other than $p$ inverted. The mod-$p$ Moore spectrum $S/p$ is the cofiber of multiplication by $p$ on the sphere. It plays the role of reduction modulo $p$, although derived constructions retain information about kernels as well as quotients.

The calculation uses Morava $E$-theory, whose coefficient ring contains formal deformation parameters. Its stabilizer group acts on those coefficients. Continuous group cohomology measures the obstructions to descent through that action; “continuous” keeps track of the $p$-adic and deformation-parameter topologies.

3. Development history and previous partial results

Hovey and Hovey–Strickland explicitly asked the degreewise finite-generation question. At height one, classical calculations describe the local sphere using an Adams operation on completed topological $K$-theory. At height two and primes at least five, Hovey–Strickland’s analysis of Shimomura’s computations gives the required finiteness for finite torsion inputs, including the Moore spectrum.

Devinatz studied related homotopy fixed-point spectra for smaller stabilizer subgroups. His results used an “essentially finite rank” condition involving periodicity and torsion. Those are important partial approaches, but their hypotheses and finiteness notion differ from the all-height statement for the full local sphere.

A separate advance determined the rational answer at every prime and height. Barthel, Schlank, Stapleton, and Weinstein identify it with an exterior algebra on $n$ generators in degrees $1-2i$, for $1\leq i\leq n$. This determines potential free ranks, but rationalization cannot control integral torsion.

For example, the compact module

$$\prod_{j\geq1}\mathbb F_p$$

has exponent $p$ and zero rationalization, yet is not finitely generated over $\mathbb Z_p$. Even a uniform bound on torsion exponent would therefore leave the principal finiteness question open.

The descent machinery was also largely available: Devinatz–Hopkins relate the local sphere to Morava homotopy fixed points, and existing descendability results provide a horizontal vanishing line on a finite page of the relevant spectral sequence. M1 identifies finite coefficient cohomology as the remaining input and develops a geometric argument to supply it. Its introduction gives the precise primary references for these earlier results.

4. What the new manuscripts claim

M1, dated 24 September 2026, states that

$$\pi_t L_{K(n)}S_p^{\wedge} \quad\text{is finitely generated over }\mathbb Z_p$$

for every prime $p$, every $n\geq1$, and every integer $t$. Here $S_p^{\wedge}$ denotes the completed sphere.

Equivalently, the same conclusion holds for

$$\pi_tL_{K(n)}F$$

for every finite $p$-local spectrum $F$. Finite cofiber sequences, suspension, and retracts transfer the sphere conclusion to these inputs. The Noetherian property of $\mathbb Z_p$ ensures that kernels, cokernels, and extensions preserve finite generation.

The algebraic theorem driving the argument says that the continuous stabilizer cohomology of the even Morava coefficients modulo $p$ is finite in each cohomological and internal degree. It also treats a carefully specified larger class of coefficients arising from whole finite division-point schemes and associated constructions. It is not asserted for every imaginable semilinear coefficient module.

Combined with the earlier rational calculation, the result determines the free ranks. They are encoded by

$$\mathbb Q\otimes_{\mathbb Z}\pi_*L_{K(n)}S \cong\Lambda_{\mathbb Q_p}(\zeta_1,\ldots,\zeta_n), \qquad |\zeta_i|=1-2i.$$

An exterior algebra has basis given by products of distinct generators, including the empty product in degree zero. Thus the rank in a degree counts subsets of these degrees with the required sum. The new finiteness assertion concerns the integral groups left after restoring torsion.

5. Why the advance matters

Importance: 5/5, provisionally. If correct, this settles a basic structural problem about the arithmetic size of the local sphere, uniformly across primes and heights.

Degreewise finite generation is a foundation for further computation. It says that each answer can be described by a finite rank and finitely many finite cyclic summands, even though locating those summands may remain difficult. It also separates controlled integral information from the much weaker statement that the rational answer is finite-dimensional.

The method would connect a broad topological finiteness problem with geometry of formal groups and $p$-adic period spaces. The significance lies both in the uniform conclusion and in the claimed passage from geometric support calculations back to integral homotopy.

6. The main idea

Reduce integral finiteness to finite coefficients

The first target is finiteness of the local Moore spectrum’s homotopy groups. Morava descent organizes these groups using continuous stabilizer cohomology. Finite entries on its second page, together with the established finite-page horizontal bound, leave finitely many contributions to each resulting homotopy degree.

One must distinguish that later-page bound from finite cohomological dimension of the full stabilizer. At small primes, the stabilizer can contain $p$-torsion. The paper retains this possibility rather than imposing a large-prime hypothesis.

After treating $S/p$, finite power cofibers and derived $p$-completeness recover the integral group as a compact $\mathbb Z_p$-module. Compact Nakayama then turns finiteness modulo $p$ into finite generation.

Induct through the deformation strata

The coefficient argument moves downward through formal-group heights. At a step, one deformation parameter is inverted, separating an open stratum from a deeper boundary stratum. The induction hypothesis controls the finite boundary strips. The remaining issue is cohomology on the open part.

A useful intermediate goal is countability. The original coefficient cohomology is compact and Hausdorff; in this setting, proving it countable forces it to be finite. But countability must hold at a sufficiently deep fixed finite marking level, not merely after taking an unlimited union of levels.

Markings choose compatible division-point data. A completed tower of such choices is interpreted using independent tuples of extensions of vector bundles on the Fargues–Fontaine curve. Compact-support calculations on that geometric space yield a finite stable image for a transfer operator.

Return from the completed tower

The transfer is a Cartier-type operation, related to Frobenius in characteristic $p$. Repeated application reduces pole orders up to a controlled loss. The manuscript chooses one normalization and one sufficiently large lattice that work through all subsequent iterates.

A Laurent-series model and residue pairing connect the geometric compact-support calculation with the transpose of this transfer tower. This controls the intersection of all late transfer images. It does not yet prove that the whole open-stratum cohomology is countable.

For that final step, the paper uses finite translation representations and a Koszul extension associated with their distribution algebra. If uncountable cohomology persisted in a highest offending degree, the resulting operation would have countable cokernel but would factor through every sufficiently late transfer image. The finite stable-image calculation would then force its image to be countable too, giving a contradiction.

Keeping a common finite marking level and uniform pole bounds is central to the argument as presented. Once countability descends to the original coefficients, compactness supplies finiteness and closes the height induction. The descent and completion steps then deliver the stated sphere theorem.

7. Scope, limitations, and open questions

The theorem is degreewise. It provides no uniform bound on ranks, generator counts, or torsion orders as the prime, height, or degree varies. It also does not compute the finite torsion summands.

The input extension is to localizations of finite spectra. Arbitrary dualizable or invertible local spectra cannot replace those inputs: the cited height-two literature already contains invertible objects whose homotopy is not of finite type. Thus this result has a different scope from the construction-class classification in family 311.

No chromatic splitting conjecture is assumed. Finite generation alone does not describe the maps joining different chromatic layers or imply a wedge decomposition of the local sphere. Those are separate questions, some addressed by the manuscripts in family 318.

The geometric coefficient class is also specific. Its closure properties are designed to survive the height induction and finite marking constructions. Extending the same finiteness assertion to a much larger class of continuous modules would require additional results.

8. Sources and further reading

M1 is the source for the theorem and the coefficient, transfer, and descent strategy. The original problem appears in Hovey–Strickland’s Morava K-theories and localisation and Hovey’s problem list. Devinatz’s earlier finiteness approach provides historical context. Li’s Invertible spectra of finite type explains why arbitrary invertible local objects require a separate analysis.

Source manuscripts

• M1. OpenAI, Finite generation for the K(n)-local sphere (September 24, 2026).

Sources consulted on 22 October 2026 at repository revision 1581d4c. Download the manuscript citations.