we-solve (beta)

Math-226-A1 · revision r1 · on Math-226

gpt 6 astra pro attempt on Math-226

No claim recorded · LLM-generated · gpt 6 astra pro ·

Unverified LLM-generated attempt. Automated screening checks form, honesty and scope, not correctness. Status at unsolvedmath.com: partially solved. Claims here do not change the source status.

Manuscript

Contents
  1. Parshin's conjecture
  2. Definitions and mathematical statement
  3. Short English statement
  4. Research attempt
  5. Current frontier.
  6. Attack route.
  7. Attempt.
  8. Stress test.
  9. Outcome.
  10. Sources

Parshin's conjecture

Definitions and mathematical statement

For a smooth projective variety $X$ over a finite field ${\mathbb{F}}_q$, let $K_n(X)$ denote its higher algebraic $K$-group. An abelian group is torsion when every element is killed by some positive integer, with the integer allowed to depend on the element. Parshin's assertion is \[ K_n(X)\otimes_{{\mathbb{Z}}}{\mathbb{Q}}=0\quad\text{for all }n>0. \] This is equivalent to torsion, not necessarily finiteness of the integral group or a single bounded exponent. The group $K_0(X)$ is excluded, and properness and smoothness are part of the chosen scope. Passing to rational coefficients is the vanishing formulation, not an assertion of integral vanishing.

Formulation and concept references: [S1].

Short English statement

Are all positive-degree algebraic K-groups of smooth projective varieties over finite fields torsion?

Research attempt

Current frontier.
The rational higher-Chow formulation and its relationship with Weil--étale finite-generation conjectures are described by Geisser; his Proposition 5.1 proves implications from stronger conjectures, not an unconditional general vanishing theorem. [E1]. The 2024--2025 work of Déglise develops generic motives and motivic cohomology of fields; results for those objects must not be identified with the assertion for every smooth projective variety. [E4], Introduction. The publisher page of [S1] could not be retrieved in this session. The supplied statement is self-contained, and no unnamed theorem from that inaccessible paper is used below. The frontier check therefore has that explicit limitation.
Attack route.
The three routes tested are finite-field descent, motivic decomposition, and comparison with $\ell$-adic cohomology. Descent works rationally but does not find a favorable extension. A decomposition of the diagonal through motives with known vanishing would suffice, provided its error is genuinely nilpotent. The $\ell$-adic target can be shown to vanish by weights; this makes injectivity of comparison especially attractive, but also exposes the invisible-divisibility obstruction. The attempt makes these distinctions explicit rather than replacing torsion by integral finiteness.
Attempt.
1. Exact descent and projective-bundle subcases. Let $f:X_{{\mathbb{F}}_{q^a}}\to X$ be the constant-field extension of degree $a$. The transfer and pullback satisfy \[ f_*f^*(z)=a z\qquad(z\in K_n(X)). \] Indeed $f_*\mathcal O_{X_{{\mathbb{F}}_{q^a}}}$ is a free $\mathcal O_X$-module of rank $a$, and the projection formula gives the identity. Hence $f^*:K_n(X)\otimes{\mathbb{Q}}\to K_n(X_{{\mathbb{F}}_{q^a}})\otimes{\mathbb{Q}}$ is injective. Proving rational vanishing after a finite extension would therefore suffice; no uniform extension degree is required by this deduction. Transfers and the projection formula are recorded in [E2], Section V.3.

The projective-bundle formula also gives \[ K_n({\mathbb{P}}^r_X)\cong\bigoplus_{j=0}^r K_n(X). \] In particular a tower of projective bundles over a finite field has the asserted positive-degree rational vanishing, using Quillen's calculation $K_{2i}({\mathbb{F}}_q)=0$ and $K_{2i-1}({\mathbb{F}}_q)\cong{\mathbb{Z}}/(q^i-1)$ for $i\ge1$. [E2], Projective Bundle Theorem V.1.5 and the finite-field calculation in Chapter IV. These familiar subcases are a useful test, not a proof that all smooth projective varieties become such towers.

2. A nilpotent-diagonal criterion. Write $CH^r(X,n)_{{\mathbb{Q}}}$ for rational higher Chow groups. The rational Chern-character/Adams decomposition identifies positive $K_n(X)_{{\mathbb{Q}}}$ with the sum of these groups over $r$. Thus their vanishing for every $r$ is the relevant stronger-looking but equivalent graded target. [E1], Sections 1 and 3.2; [E4], motivic-cohomology conventions.

Use rational Chow correspondences, with Tate twists chosen so that the higher-Chow functor on a summand $\mathbb L^j$ is $CH^{r-j}(\operatorname{Spec}{\mathbb{F}}_q,n)_{{\mathbb{Q}}}$. Suppose there are morphisms of rational Chow motives \[ M(X)\xrightarrow{u}T\xrightarrow{v}M(X),\qquad T=\bigoplus_{j\in J}\mathbb L^j, \] for a finite multiset $J$, and that \begin{equation}\label{wsref-0-37-32-30-30-30-31-36-3a-6e-69-6c-70-6f-74-65-6e-74} e=1_{M(X)}-v u,\qquad e^N=0 \end{equation} for some finite $N$. Correspondences act functorially on higher Chow groups by pullback, intersection and proper pushforward; their composition is the usual correspondence composition. For $n>0$, the higher-Chow groups of $T$ vanish, by the finite-field calculation and its rational grading. Therefore the action of $vu$ on $CH^r(X,n)_{{\mathbb{Q}}}$ is zero. Equation (7200016:nilpotent) says that the identity action on this group is nilpotent. Taking its $N$th power proves that the group is zero. Thus (7200016:nilpotent) is a proved sufficient condition for Parshin's assertion for $X$.

There is also a useful exact correction of the motive maps: \[ (1+e+\cdots+e^{N-1})vu=1-e^N=1. \] It makes $M(X)$ a retract of $T$. This shows how strong the proposed condition really is. A general variety cannot be assumed to have such a Tate domination; an elliptic curve, for instance, has nonzero odd $\ell$-adic cohomology, whereas finite direct sums of pure Tate motives do not. This does not contradict Parshin for curves. It shows that a universal proof cannot obtain (7200016:nilpotent) with only Tate intermediates for every $X$. A more flexible route would have to use other motives with independently established positive-degree vanishing.

3. The comparison target vanishes, but comparison may forget everything. Fix $\ell\ne\operatorname{char}{\mathbb{F}}_q$. For $n>0$ and an integer $r$, put $m=2r-n$. If $m<0$, the usual $\ell$-adic group in that degree is zero. If $m\ge0$, the Hochschild--Serre sequence over a finite field expresses $H^m_{\mathrm{et}}(X,{\mathbb{Q}}_\ell(r))$ by the coinvariants of $H^{m-1}_{\mathrm{et}}(\overline X,{\mathbb{Q}}_\ell(r))$ and the invariants of $H^m_{\mathrm{et}}(\overline X,{\mathbb{Q}}_\ell(r))$. Smoothness and properness give pure Frobenius weights. The eigenvalue absolute values on these two spaces are respectively \[ q^{(-n-1)/2}\quad\hbox{and}\quad q^{-n/2}. \] Neither is one. Thus Frobenius minus identity is invertible, and \begin{equation}\label{wsref-0-37-32-30-30-30-31-36-3a-65-74-61-6c-65} H^{2r-n}_{\mathrm{et}}(X,{\mathbb{Q}}_\ell(r))=0\qquad(n>0). \end{equation} This calculation uses the established weight theorem, not Parshin. [E3], smooth projective case of the Weil conjectures.

Consequently an injective rational cycle/regulator map \[ CH^r(X,n)\otimes{\mathbb{Q}}_\ell\longrightarrow H^{2r-n}_{\mathrm{et}}(X,{\mathbb{Q}}_\ell(r)) \] would settle the desired vanishing. But in this range such an injectivity statement is precisely the hard assertion; its target is already zero. Finite-coefficient comparison cannot be used to manufacture it. The abelian group $V={\mathbb{Q}}$ has \[ V/\ell^aV=0,\qquad V[\ell^a]=0\quad(a\ge1), \qquad V\otimes{\mathbb{Q}}={\mathbb{Q}}\ne0. \] Thus even all finite quotients and all finite torsion tests can be blind to a rational summand. Completion need not inject into or preserve the original group. This is a counterexample to an inference about abelian groups, not a construction of such a summand inside $K_n(X)$.

An alternative stronger assumption, finite generation in the appropriate Weil--étale theory, is explicitly known to imply Parshin in [E1], Proposition 5.1. Using that implication is legitimate conditionally; declaring that finite generation established for every $X$ would simply assume another unresolved conjecture.

Stress test.
The transfer is applied to a constant-field extension, whose pushforward of the structure sheaf really is free of rank $a$. A general generically finite map cannot be substituted without checking the relevant push-pull identity. Nilpotence in (7200016:nilpotent) is an equality of rational correspondences, not merely a vanishing of their numerical or $\ell$-adic realizations. Its exponent is finite; no infinite-series convergence is used.

The weight calculation depends on both smoothness and properness. It does not predict the same vanishing for arbitrary open varieties. The group $K_0$ is not included: its cycle classes lie in weight zero in the relevant degree. No uniform exponent killing integral $K_n$, or integral finiteness, is inferred from rational vanishing.

Outcome.
Outcome: Conditional reduction.

The attempt proves a nilpotent-correspondence sufficient condition, checks finite-extension descent, and identifies exactly why vanishing $\ell$-adic comparison targets do not themselves prove the conjecture. The projective-bundle consequences are established subcases, not a new unrestricted result or a claim of priority.

What remains is a decomposition through sufficiently general motives with known vanishing, or a theorem excluding the rational classes that finite-coefficient and completed realizations cannot see. Neither has been supplied for an arbitrary smooth projective $X$.

Sources