Transfer of mass--Hom bounds #
Let G : C ⥤ D detect a preimage stability condition and let L ⊣ G be a
linear adjunction. A mass--Hom bound for a test object A : D transfers to
L.obj A: the adjunction identifies Hom(LA, F) with Hom(A, GF), while
stabilityMass_preimage identifies the two masses.
Consequently, a bound on a target test class T transfers to the essential
image T.map L and then, by thick closure, to every source test class contained
in (T.map L).triangEnvelope. The generation comparison is deliberately an
explicit hypothesis: the abstract theorem neither supplies a geometric
generator nor assumes that L preserves one.
This is the categorical core of arXiv:2607.28411v1, Lemma 7.4. Its finite and faithfully-flat geometric realizations are separate consumers.
An objectwise mass--Hom bound transfers across a linear left adjoint.
No conservativity hypothesis is needed here. The supplied PreimageData
already says that the preimage slicing is a genuine slicing, and the proof uses
only the adjunction on Hom spaces and the exact mass identity.
A mass--Hom bound transfers to the essential image of the test class under the linear left adjoint.
Transfer a mass--Hom bound to a source test class generated by the left-adjoint images of the bounded target tests.
The containment T' ≤ (T.map L).triangEnvelope is the exact generation input.
Geometric applications may instantiate T and T' by perfect complexes, but
that comparison is not asserted by this categorical theorem.