Let g\mathfrak g be a Lie algebra and let ρ:g→gl(V)\rho:\mathfrak g\to \mathfrak{gl}(V) be a on a vector space VV, possibly infinite-dimensional. A linear subspace W⊆VW\subseteq V is a subrepresentation (or g\mathfrak g-submodule) if it is invariant under the action:

ρ(X)(W)⊆Wfor all X∈g.\rho(X)(W)\subseteq W \quad \text{for all } X\in\mathfrak g.

In this case, restricting ρ(X)\rho(X) to WW defines a representation ρ∣W:g→gl(W)\rho|_W:\mathfrak g\to \mathfrak{gl}(W).

Quotients and irreducibility

If W⊆VW\subseteq V is a subrepresentation, then the quotient space V/WV/W inherits a natural g\mathfrak g-action via

X⋅(v+W)=(X⋅v)+W,X\cdot (v+W) = (X\cdot v)+W,

well-defined precisely because WW is invariant. A representation is irreducible (see ) if its only subrepresentations are {0}\{0\} and VV.

Why this matters

Subrepresentations are the “building blocks” for decomposing representations. For a finite-dimensional representation of a semisimple Lie algebra over a field of characteristic zero, says every subrepresentation has an invariant complement. This conclusion does not extend to arbitrary infinite-dimensional modules.