gpt 6 astra ultra attempt on Math-457

Unit propagation and a trace-Bockstein reduction for Greek-letter redshift

Definitions and mathematical statement

Fix a prime at which the $n$-th and $(n+1)$-st Greek-letter families in the source exist as nonzero stable homotopy classes. Let $S$ be the sphere spectrum, so $\pi_jS$ is its $j$-th stable homotopy group. Greek-letter elements are the periodic families constructed from powers of chromatic periodicity classes and connecting maps, or equivalently the specified permanent cycles in the chromatic and $BP$-Adams spectral sequences; use Section 1 of Angelini-Knoll for the indexing.

A commutative ring spectrum $R$ detects a family $\{\alpha_k^{(n)}\}$ when every member remains nonzero under the unit: \[ \pi_*S\longrightarrow\pi_*R. \] Writing $K(R)$ for the algebraic K-theory spectrum of perfect $R$-modules, Conjecture 1.1 asks that detection of the $n$-th family by $R$ imply detection of the $(n+1)$-st by $K(R)$. The source's existence qualification on the families matters, especially at small primes. Detecting a single element or merely proving a height bound is not the entire assertion.

Formulation and concept references: [S1].

Short English statement

Does taking algebraic K-theory raise by one the Greek-letter family detected by a commutative ring spectrum?

Research attempt

Scope and the distinction from qualitative redshift.

For a ring spectrum $E$, write $\eta_E:S\to E$ for its unit and $h_E:\pi_*S\to\pi_*E$ for the induced map. The target is nonvanishing of every specified existing member of the next Greek-letter family under $h_{K(R)}$. Existence of some nonzero homotopy class in the same degree, periodicity in a homology theory, and chromatic-height growth are different assertions.

The source audit on 27 September 2026 found two especially relevant primary statements. Angelini-Knoll [S1, Theorem 1.2] proves detection of the $\beta_k$ with $k\not\equiv0\pmod p$ in $K(K(\mathbb F_q)_p)$, for $p\geq5$ and the specified prime-power generator $q$ modulo $p^2$. The surrounding discussion explicitly restricts this argument to classes surviving reduction to $V(1)$. It does not establish the entire printed conjecture for arbitrary $R$. Burklund--Schlank--Yuan [E1, Theorem E] prove a redshift theorem for chromatic height of commutative ring spectra. That theorem does not state the required elementwise unit-detection assertion. Neither result is enlarged in scope below.

This attempt develops exact functorial reductions, proves a split-unit subcase covering augmented sphere algebras, and audits the coefficient reductions that occur in the known height-two example. It does not claim to reproduce the deep spectral-sequence computation of [S1].

Functoriality has a useful but asymmetric direction.

Lemma 453.1. Let $f:E\to F$ be a unital map of ring spectra. For any $\alpha\in\pi_*S$, if $h_F(\alpha)\ne0$ then $h_E(\alpha)\ne0$.

Proof. Unitality says $f\eta_E=\eta_F$, so $\pi_*f(h_E(\alpha))=h_F(\alpha)$. A zero element cannot have a nonzero image.$\square$

Thus detection in a target transfers backwards along a unital map. It does not generally transfer forwards, since $\pi_*f$ may kill the class. For example the unital map $S\to H\mathbb Q$ kills every positive-degree torsion sphere class, although the sphere detects it.

For a map of commutative ring spectra $R\to R'$, extension of scalars sends perfect $R$-modules to perfect $R'$-modules and gives $K(R)\to K(R')$. Perfectness is preserved because finite cell modules and their retracts remain so after extension of scalars. The map is compatible with the units, represented by the free rank-one modules. Consequently, proving detection in $K(R')$ suffices to prove detection in $K(R)$. This can be used element by element: the target $R'$ may depend on which class is being tested, provided an actual unital map and a valid detection theorem for that target are supplied each time.

The hypothesis of the conjecture does not itself supply such a map to a known computable detector. Reversing the implication of Lemma 453.1, or inventing a map from a mere chromatic-height comparison, would be an invalid reduction.

The trace proves preservation of already detected classes.

We use the standard natural Dennis trace [E3] and the augmentation of topological Hochschild homology for a commutative ring spectrum: \[ K(R)\stackrel{\operatorname{tr}}{\longrightarrow}\operatorname{THH}(R) \stackrel{\epsilon}{\longrightarrow}R. \tag{453.1} \] The augmentation is induced by collapsing the circle in $\operatorname{THH}(R)\simeq R\otimes S^1$, where the tensor is in commutative ring spectra. These standard trace maps are compatible with units. Equivalently, the free rank-one module maps to the class of $1\in\pi_0R$, so the composite of $453.1$ with $\eta_{K(R)}$ is $\eta_R$. Maps from $S$ to a spectrum are classified by $\pi_0$, which also verifies this equality of unit maps.

Proposition 453.2. If $R$ detects a collection of sphere classes, then $K(R)$ detects the same collection.

Proof. If $h_{K(R)}(\alpha)=0$, applying $453.1$ gives $h_R(\alpha)=0$. Taking the contrapositive proves the assertion. $\square$

This proves preservation, not a shift from the $n$-th to the $(n+1)$-st family. Indeed, the map from $R$ into $\operatorname{THH}(R)$ also preserves units. Any sphere class killed by the unit of $R$ is therefore killed by the unit of $\operatorname{THH}(R)$. A proposed argument detecting new classes only in the underlying $\operatorname{THH}(R)$ would miss this obstruction. Cyclotomic or homotopy-fixed-point refinements used in trace calculations carry additional information; they must not be silently replaced by the underlying Hochschild-homology spectrum.

A complete split-unit subcase in every allowed height.

Proposition 453.3. Suppose $R$ admits a unital augmentation $a:R\to S$. Then the unit $S\to K(R)$ has a retraction as a map of spectra. Hence $K(R)$ detects every nonzero stable sphere class, and in particular every existing Greek-letter family in the statement.

Proof. Compose the map $K(a):K(R)\to K(S)$ with the Dennis trace \[ K(S)\longrightarrow\operatorname{THH}(S)\simeq S. \] Since $\operatorname{THH}(S)$ is the sphere and both maps preserve units, the composite with $S\to K(R)$ is the identity of $S$. Therefore the induced homomorphism $\pi_*S\to\pi_*K(R)$ is split injective.$\square$

Here the input $R$ also detects every sphere class, since $a\eta_R=\operatorname{id}_S$. Thus this is a genuine class of instances satisfying the hypothesis and conclusion of the conjecture, not a vacuous implication. It includes $R=S$ itself and commutative spherical monoid algebras $\Sigma^\infty_+M$ with the augmentation induced by $M\to *$, for commutative monoids $M$. All arguments can also be made after localization at the fixed prime, replacing the sphere consistently by $S_{(p)}$.

A unit map into an arbitrary $R$ need not split, so the proposition does not extend to every ring detecting one selected family. Detection of that family asserts injectivity on its individual chosen nonzero members, not a splitting of the full unit as a spectrum map.

Finite products offer another exact propagation rule. Perfect modules over $R_1\times R_2$ split using the two central idempotents, giving \[ K(R_1\times R_2)\simeq K(R_1)\times K(R_2). \] The sphere unit maps diagonally. A class detected in either factor is detected in the product. If neither factor detects a class, its product image is zero. This uses the actual component maps, not a maximum of chromatic heights as a substitute for detecting a specified element.

Chromatic height cannot determine the unit image.

The following elementary obstruction is useful when interpreting [E1]. If all homotopy groups of a ring spectrum $E$ are torsion-free, then every positive-degree torsion sphere class maps to zero in $\pi_*E$: its image has order dividing its finite order and must vanish.

For example, the usual Lubin--Tate spectra $E_h$ are commutative ring spectra of chromatic height $h$. Their coefficient groups are even periodic with coefficient ring \[ \pi_*E_h\cong W(k)[[u_1,\ldots,u_{h-1}]][u^{\pm1}],\qquad |u|=2 \] up to the harmless choice of the sign of the periodicity degree. For perfect fields $k$ of characteristic $p$, these groups are torsion-free as abelian groups, and odd groups vanish. Thus their units kill positive-degree torsion sphere classes, despite their nonzero height. This uses the standard coefficient description of Lubin--Tate theory, also part of the setting of [E1].

It follows that a numerical height statement cannot by itself imply detection of a specified positive-degree Greek-letter class. This is not a counterexample to the conjecture: these examples do not satisfy the required family-detection hypothesis. It identifies the missing information in an attempted deduction solely from qualitative redshift. Moreover, periodicity is usually measured after smashing with a finite complex. A nonnilpotent self-map on that complex is not the same object as the torsion sphere class produced by connecting maps.

A low-height arithmetic calculation, with the unit input separated.

Let $p$ be odd and let the prime power $q$ generate $(\mathbb Z/p^2\mathbb Z)^\times$. Quillen's finite-field calculation, used in [S1], gives \[ K_{2m-1}(\mathbb F_q)\cong\mathbb Z/(q^m-1),\qquad K_{2m}(\mathbb F_q)=0\quad(m\geq1). \] At $m=(p-1)k$, the $p$-primary cyclic group has order $p^{1+\nu_p(k)}$. Here is a direct verification of the exponent.

Since $q$ has order $p(p-1)$ modulo $p^2$, write $q^{p-1}=1+pc$ with $p\nmid c$. If $B=1+p^rb$ with $r\geq1$ and $p\nmid b$, binomial expansion for odd $p$ shows \[ \nu_p(B^p-1)=r+1. \] The linear term has valuation $r+1$, every intermediate term has valuation at least $1+2r\geq r+2$, and the final term has valuation $pr\geq r+2$. Raising to a positive exponent coprime to $p$ leaves the valuation equal to $r$, again by the binomial expansion. Writing $k=p^su$ with $p\nmid u$ proves \[ \nu_p(q^{(p-1)k}-1)=1+s=1+\nu_p(k). \tag{453.2} \]

This calculation locates nonzero target groups. It does not determine the unit homomorphism from the sphere. The separate image-of-$J$ identification discussed in [S1] is the established homotopy-theoretic input showing that the appropriate $\alpha$-family classes have nonzero unit images in $K(\mathbb F_q)_p$. Without that input, the displayed group orders alone would not prove detection: a map to a nonzero group can be the zero map.

The degrees in the height-two construction.

At $p\geq5$, let $V(0)=S/p$ and choose the standard periodic map $v_1:\Sigma^{d_1}V(0)\to V(0)$ with $d_1=2p-2$. Its cofiber is $V(1)$. A $v_2$ self-map of $V(1)$ has degree $d_2=2p^2-2$ at these primes. Write $i:S\to V(1)$ for the bottom cell map and $q_\mathrm{top}:V(1)\to\Sigma^{d_1+2}S$ for the composite of the two cofiber boundary maps to the top cell. The basic $\beta$-family construction is represented by \[ \Sigma^{kd_2}S\stackrel{\Sigma^{kd_2}i}{\longrightarrow} \Sigma^{kd_2}V(1)\stackrel{v_2^k}{\longrightarrow}V(1) \stackrel{q_\mathrm{top}}{\longrightarrow}\Sigma^{d_1+2}S. \] After desuspension its degree is \[ |\beta_k|=k(2p^2-2)-2p. \tag{453.3} \] This is consistent with the indexing in [S1]. Higher families and small primes require their own existence and indexing conventions; we do not extrapolate $453.3$ as a universal formula for them.

Smashing this diagram with $K(R)$ produces the corresponding unit image after beginning with the unit. Nonzero periodic classes somewhere in $\pi_*(V(1)\wedge K(R))$ do not automatically make this composite nonzero: the bottom-cell class, the chosen periodic map and the top-cell connecting maps all matter. The connecting map can kill a nonzero intermediate class. No commutative ring structure on $V(1)$ is assumed or needed here, and $K(V(1)\wedge R)$ is not substituted for the coefficient spectrum $V(1)\wedge K(R)$.

Reduction modulo periodicity detects only in one direction.

For any spectrum $E$, the map $i\wedge1:E\to V(1)\wedge E$ has the elementary implication \[ (i\wedge1)_*(x)\ne0\quad\Longrightarrow\quad x\ne0. \tag{453.4} \] Thus a nonzero reduced image of a sphere class gives a genuine nonzero image in $E$. The converse need not hold. Already the first cofiber sequence \[ E\stackrel{p}{\longrightarrow}E\longrightarrow V(0)\wedge E \] shows, by exactness, that the kernel of $\pi_dE\to\pi_d(V(0)\wedge E)$ is $p\pi_dE$. Even a nonzero element of order $p$ can lie in this kernel: the element $p$ in $\mathbb Z/p^2$ has precisely that property.

The second cofiber sequence gives \[ \ker\left(\pi_d(V(0)\wedge E)\to\pi_d(V(1)\wedge E)\right) =\operatorname{im}\left(v_1:\pi_{d-d_1}(V(0)\wedge E) \to\pi_d(V(0)\wedge E)\right). \tag{453.5} \] Hence a class can survive the mod-$p$ step and still disappear under the next reduction because it is $v_1$-divisible. This is why the qualification in [S1] matters: its proof detects $\beta_k$ with $p\nmid k$, whereas the indicated $\beta_{pk}$ are invisible to the reduction used there. Their disappearance in that reduction does not prove that their original unit images in iterated K-theory are zero.

An exact certificate that a trace approach would need.

Fix one existing next-family class $\beta\in\pi_dS$. It is sufficient to construct a spectrum map $\tau:K(R)\to D$ and prove \[ (\tau\eta_{K(R)})_*(\beta)\ne0. \tag{453.6} \] This immediately forces $h_{K(R)}(\beta)\ne0$, regardless of whether $\tau$ is injective on other classes. A cyclotomic trace followed by a suitable coefficient or fixed-point comparison is one possible source of such a map. What must be verified is the actual composite on the sphere, not merely the presence of a class in $\pi_dD$.

If a spectral sequence computes $\pi_*D$, finding a named class on its first page is insufficient. One must identify it as the image of the specified sphere class, exclude outgoing differentials and incoming boundaries, and use a convergence statement making its nonzero associated-graded class a nonzero element of the abutment. For a strongly convergent separated filtration, the last implication is formal; the identification and survival are the substantive work. Proving $453.6$ for finitely many $k$ does not establish all-family detection without a uniform argument covering the remaining indices.

Diagnostics, outcome and remaining gap.

The exact arithmetic checks accompanying this attempt verify $453.2$ for explicit prime-power generators and a range of indices, and check the degree shifts in $453.3$. They are checks of arithmetic and indexing, not computations of stable stems or spectral-sequence differentials. The nontrivial detection theorem for iterated finite-field K-theory is attributed to [S1], with its restriction on indices retained.

The proved results are functorial backward propagation, preservation of already detected classes via the trace, and full detection whenever $R$ admits a sphere augmentation. The exact-sequence analysis explains both the usefulness and the limitations of mod-$(p,v_1)$ detection. The first unresolved step for arbitrary $R$ is constructing and proving nonzero composites of the form $453.6$ for every existing member of the next family from only the given previous-family detection hypothesis. The known height theorem and the calculations above do not provide that missing implication. The printed conjecture remains unresolved by this attempt.

Sources

• S1 G. Angelini-Knoll, Detecting beta elements in iterated algebraic K-theory, Trans. AMS. https://arxiv.org/abs/1810.10088. Formulation source. Consulted 24 September 2026: Conjecture 1.1 and the family-detection definition in its full-text HTML version.
• E1 Robert Burklund, Tomer M. Schlank and Allen Yuan, The Chromatic Nullstellensatz, arXiv:2207.09929. https://arxiv.org/abs/2207.09929; https://arxiv.org/html/2207.09929. Primary Theorem E and the chromatic-height formulation inspected 27 September 2026. Height growth is not substituted for unit detection of individual Greek-letter classes.
• E2 Gabriel Angelini-Knoll, full-text version of [S1], https://arxiv.org/html/1810.10088. Conjecture 1.1, Theorem 1.2 and the preceding mod-$(p,v_1)$ qualification inspected 27 September 2026. The condition $p\geq5$, the prime-power generator hypothesis and the restriction $p\nmid k$ on the detected basic beta elements are retained.
• E3 Andrew J. Blumberg, David Gepner and Gonçalo Tabuada, Uniqueness of the multiplicative cyclotomic trace, Advances in Mathematics 260 $2014$, 191--232. https://arxiv.org/abs/1103.3923; https://doi.org/10.1016/j.aim.2014.02.004. Primary abstract inspected 27 September 2026 for the natural multiplicative Dennis trace and multiplicative K-theory input.