Biproducts and Biproduct Decomposition

Core Idea

In an additive category, when a product and coproduct of the same objects exist and coincide, they form a biproduct—simultaneously a product and coproduct with canonical projections and injections. Biproducts enable matrix-like representations of morphisms and are central to the theory of finitely-generated modules and linear algebra.

How It's Best Learned

Study biproducts in the category of abelian groups and modules, and verify that they coincide with direct sums. Practice decomposing objects via idempotents and representing morphisms as matrices with respect to biproduct decompositions.

Common Misconceptions

Not every category with products and coproducts has biproducts; the coincidence of product and coproduct is non-trivial. Some students confuse the existence condition with the existence of direct summands.

Explainer

From products and coproducts, you know these are dual constructions with opposite universal properties. A product A × B comes with projections π₁: A × B → A and π₂: A × B → B: any object X with maps to both A and B factors uniquely through A × B. A coproduct A ⊔ B comes with injections ι₁: A → A ⊔ B and ι₂: B → A ⊔ B: any object X receiving maps from both A and B factors uniquely from A ⊔ B. In most categories these are different objects — in Set, the product is the Cartesian product and the coproduct is the disjoint union.

In an additive category — where every hom-set Hom(X, Y) is an abelian group, composition distributes over addition, and a zero object exists — the product and coproduct of any two objects are naturally isomorphic. This coincidence is called a biproduct, written A ⊕ B. The biproduct carries *both* the projection maps π₁, π₂ and the injection maps ι₁, ι₂, satisfying a specific set of identities: π₁ι₁ = id_A, π₂ι₂ = id_B (each injection followed by its own projection is the identity), π₁ι₂ = 0, π₂ι₁ = 0 (cross terms vanish), and ι₁π₁ + ι₂π₂ = id_{A⊕B} (the identity decomposes as a sum of "projection-then-injection" maps). This last identity requires the additive structure — you cannot form that sum without addition on morphisms.

The reason these identities force product and coproduct to coincide is that the zero morphisms (zero elements of each hom-set) allow you to construct each universal property from the other. Given the projections, you construct the coproduct injection ι₁ = (id_A, 0): A → A ⊕ B by combining the identity on A with a zero map to B's factor; the coproduct universal property then follows. Conversely, from the injections you construct the product projections. The additive structure provides the glue.

The payoff is matrix calculus for morphisms. Any morphism f: A ⊕ B → C ⊕ D can be written as a 2×2 matrix [[f₁₁, f₁₂], [f₂₁, f₂₂]] where fᵢⱼ = πᵢ ∘ f ∘ ιⱼ. Composition of morphisms becomes matrix multiplication — the formula for matrix products is exactly the categorical composition law expressed through the biproduct decomposition. This is not a coincidence: Vec_k (vector spaces over a field k) is an additive category, the biproduct is the direct sum of vector spaces, and linear maps are matrices. Biproducts are therefore the categorical explanation for why linear algebra has matrix multiplication at all. Moving to module categories, biproducts correspond to direct sums of modules, and the matrix representation of module homomorphisms follows the same pattern. This connection makes biproducts foundational to the theory of abelian categories, where they are always present and central to structural decomposition theorems.