Statement
An -numerical certificate for consists of an integral projective flat model and such that
for every integral vertical -dimensional closed subscheme .
Let be finite dominant between smooth integral projective -varieties. The draft asserts that a certificate for implies , and hence a strong Green form for each prescribed representative of . The source also allows a smooth integral variety over a finite larger constant field regarded as a -variety, retaining its native degree and special-fibre multiplicities.
Motivation
The composite-trace construction uses the original generators; it does not require a previously traced form to belong again to the trace module.
Draft proof route
Pass to the closure of the generic graph so that the certificate and the pulled-back model character live on a common model. Model refinements preserve the certificate by operational projection. The numerical detector makes the certificate curvature strictly zero, and generic Chow invariance supplies a structured global primitive upstairs. Apply the restricted trace to the original scalar–curvature generators along their composite finite maps, divide by the full native degree, and test against all compact actual forms. This produces the proposed global primitive downstairs. Finally apply BB-GREEN-CORRECTION.
The source below contains the expanded argument and its hypotheses.
Source and version
def:numerical-certificate; lem:refine-certificate; thm:finite-numerical (D33 Theorem 7.3)
research-sources/green-draft/abelian_numerical.tex
File and record fingerprints
Published source SHA-256
8eddaba8870644af5af176df22f4406bc3666cd5af6e61261fed3bda8f190fd8Statement / argument record SHA-256
fd162de6ec218642aff0d9cf3df6968000d1d8c01fb84ed63cfb4f15b388f1d9Public source copies omit personal attribution; each fingerprint identifies the bytes of that published copy.
References
- Andreas Mihatsch — δ-Forms on Lubin–Tate Space
2024 version; actual δ-forms and complete-intersection Green forms
Questions, corrections and suggestions
Discuss this record in the atlas’s shared feedback board. The record and edition are carried into the form; general suggestions can be sent without them.
GitHub Issues and attachments are public. A contribution starts a discussion, not an automatic change to the mathematics.