Patching submodules across two principal localizations #
For a finite module over a Noetherian ring, submodules of its localizations away from r and
s that agree in the common localization away from r * s come from one finitely generated
submodule. The witness is the intersection of their inverse images in the original module.
The overlap hypothesis compares the localizations at r * s of those two inverse images. This
is a purely algebraic two-chart statement. It constructs no sheaf or coherent subobject on a
union of basic opens, and proves no quasi-compact-open descent theorem.
Two compatible submodules on the principal localizations of a finite module have a single
finitely generated lift. Compatibility is equality in the canonical common localization away
from r * s, after pulling both submodules back to M. The witness is the intersection of
these two inverse images. No geometric sheaf-gluing statement is asserted.