Complete Reducibility
The question of when representations decompose into irreducibles is fundamental. Complete reducibility vastly simplifies the classification problem by reducing it to understanding irreducible representations.
For a finite-dimensional representation of , the following are equivalent:
- is completely reducible (isomorphic to a direct sum of irreducibles)
- Every subrepresentation has a complementary subrepresentation: for some
- is a sum (not necessarily direct) of irreducible subrepresentations
- There exists no infinite chain of proper subrepresentations (ascending or descending chain condition)
The implication (2) (1) proceeds by induction: if is not irreducible, choose a proper subrepresentation , find a complement , and decompose each piece recursively.
Unitary representations: Every finite-dimensional unitary representation (where are unitary operators) is completely reducible. Given a subrepresentation , the orthogonal complement is automatically a complementary subrepresentation.
Finite groups over : By Maschke's theorem, every representation of a finite group in characteristic 0 is completely reducible. This fails in positive characteristic dividing .
Counterexample: The two-dimensional representation of generated by has the -axis as an invariant line, but no complementary invariant lineβit is not completely reducible.
A non-zero representation is called indecomposable if it cannot be written as a non-trivial direct sum: if , then either or .
Every irreducible representation is indecomposable, but the converse fails without complete reducibility.
Every finite-dimensional representation decomposes uniquely (up to isomorphism and reordering) into a direct sum of indecomposable representations:
When complete reducibility holds, "indecomposable" coincides with "irreducible," so this becomes the unique decomposition into irreducibles (with multiplicities).
Complete reducibility dramatically simplifies representation theory: instead of studying all representations, we need only classify irreducibles and understand their multiplicities. The representation ring (or Grothendieck group) is freely generated by irreducibles, tensor products decompose via explicit formulas (Clebsch-Gordan), and character theory provides a complete set of invariants.
When complete reducibility fails (e.g., modular representations), the theory becomes significantly more complex, involving derived categories, homological methods, and the study of indecomposable representations which are not irreducible.