A reduction technique that proves statements about a complex object by filtering it into a finite sequence of subobjects whose successive quotients (graded pieces) are simpler, and then deducing the property for the whole from properties of the pieces and the glueing data between them.

Principle

Introduce a finite filtration 0 = F_0 ⊂ F_1 ⊂ ... ⊂ F_n = X such that each successive quotient F_{i+1}/F_i belongs to a class where the desired statement is known; verify statements on each graded piece and propagate them up the filtration using exactness or extension arguments.

Demonstration

To prove a property for a coherent sheaf on a scheme, choose a finite filtration by subsheaves whose quotients are supported on simpler closed subschemes (e.g., structure sheaves of integral components), verify the property on these quotients, and use long exact sequences to lift the conclusion to the original sheaf.

Misapplication

Applying devissage without ensuring the filtration exhausts the object or that the graded pieces lie in a controllable class can invalidate the reduction; assuming properties glue automatically across extensions without checking extension obstructions is a common mistake.

Consequence

Devissage allows complex structural or cohomological problems to be reduced to tractable cases, often enabling inductive proofs and computations (for example in K-theory or cohomology) by replacing a global question with finitely many local verifications.

Reversal

Instead of breaking an object into graded pieces, one may attempt to assemble an object from known pieces via successive extensions; this constructive perspective emphasizes existence of nontrivial extension classes rather than reduction of properties.

Boundary

The method requires the existence of an appropriate finite filtration whose graded pieces fall into a class amenable to analysis; it is not applicable when no finite filtration exists or when extension data between pieces cannot be controlled or understood.

Semantic Tension

There is tension between devissage and techniques that analyze objects via spectral sequences or derived-category truncations: all aim to reduce complexity but differ in whether they use explicit finite filtrations of concrete subobjects or more global homological filtrations.

Synthesis

Devissage is a finite-step filtration strategy that reduces proofs about a complex algebraic or categorical object to verifications on simpler graded constituents and controlled extension steps, thereby converting a single difficult assertion into a sequence of manageable checks.