Documentation

DerivedAlgGeo.CategoryTheory.Triangulated.SemiorthogonalDecomposition.DerivedField

An exceptional object in the derived category of a field #

The abstract layer gets its real inhabitant: the residue field, as a complex concentrated in degree zero, is an exceptional object of DerivedCategory (ModuleCat k), and it assembles into a length-one exceptional collection with its semiorthogonal sequence. This discharges the design note's requirement — .claude/notes/2026-08-28-semiorthogonal-reconnaissance.md §4: "issue #852 requires those notions to be added only when a theorem consumes each property" — for this milestone, with a non-geometric consumer that leaves the projective-families lane's territory untouched.

Why this category #

Mathlib/Algebra/Homology/DerivedCategory/Linear.lean provides noncomputable instance : Linear R (DerivedCategory C) and instance (n : ℤ) : (shiftFunctor (DerivedCategory C) n).Linear R — the latter is exactly the shift-linearity hypothesis IsExceptional.shift had to assume in general, and the derived category is one of the few places in Mathlib where it is a global instance. HasDerivedCategory.{w} (ModuleCat k) is a hypothesis throughout; HasDerivedCategory.standard supplies it at max u v for C : Type u with Category.{v} C — here u + 1, since ModuleCat.{u} k : Type (u + 1) — when a caller needs a concrete choice.

Fullness is not claimed — and is expected to be false #

ObjectProperty.triangEnvelope reaches only objects assembled from the generator by shifts, binary products, retracts, and FINITELY many extension steps (triangEnvelope = ⨆ n, triangEnvelopeIter n). An object of the unbounded DerivedCategory (ModuleCat k) with infinitely many nonzero cohomology objects is not reachable that way, so IsFull for the length-one collection is expected to be FALSE here, not merely unproved. Fullness belongs to a bounded-derived-category consumer, which this milestone does not file.

Main definitions #

Main results #

References #

Tags #

exceptional object, derived category, residue field

algebraMap into the linear endomorphisms of the ring itself is bijective: evaluation at 1 inverts it. The base case the derived-category transport stands on; only commutativity enters, so a CommRing suffices.

@[reducible, inline]

The residue field, as a complex concentrated in degree zero. Deliberately an abbrev: instance search must see the singleFunctor application to find the t-structure IsLE/IsGE instances the negative-shift argument consumes.

Equations
Instances For

    The residue field in degree zero is exceptional. Endomorphisms: singleFunctor is fully faithful and k-linear, so algebraMap bijectivity transports from End(k) = k. Positive shifts: k is projective, so higher Ext vanishes, read through Ext.homEquiv. Negative shifts: the source is IsLE 0 and the shifted target IsGE (-n) for the canonical t-structure, so the map vanishes by TStructure.zero.

    The length-one exceptional collection on the residue object — the honest inhabitant of ExceptionalCollection in the tree, through ofExceptional. Its semiorthogonal sequence exists by the collection layer; fullness is not claimed, and is expected to fail — see the module docstring.

    Equations
    Instances For
      @[simp]

      The single component of the length-one collection is the triangulated envelope of the shift closure of the residue object.