Schurβs lemma
Schurβs lemma is most naturally stated in the language of modules.
Let
Very simple proof
Since
and k e r β‘ π β€ π are submodules of simple modules, they must either be trivial or equal to i m β‘ π β€ π and π respectively. If π then π β 0 and k e r β‘ π β π , hence i m β‘ π β 0 is epic and monic and thus an π -module isomorphism. π
- If
, by All elements of a finite-dimensional unital associative algebra are algebraicd i m π β‘ π < β΅ 0 - If
, by Dixmierβs lemmad i m π β‘ π < | π | - If
has a filtration like the universal enveloping algebra of a finite-dimensional Lie algebra, by Quillenβs lemmaπ΄
which also rely on the result from Division algebra with only algebraic elements over an algebraically closed field.
Schurβs lemma for unitary group representations
Schurβs lemma is a statement about linear maps which βcommuteβ with an irrep.12
Schurβs lemma, first form β’
Let
for all
Proof
Let
be an eigenvalue of π , and π΄ . Then π£ β E π ( π΄ ) for all π΄ Ξ ( π ) π£ = Ξ ( π ) π΄ π£ = π Ξ ( π ) . Therefore π β πΊ , meaning Ξ ( π ) π£ β E π ( A ) is E π ( π΄ ) -Invariant subspace. Since Ξ is irreducible and Ξ , E π ( π΄ ) β { π } . Therefore πΈ π ( π΄ ) = π . π΄ = π π
Schurβs lemma, second form β’
Let
for all
Proof
Taking the Hermitian conjugate of both sides gives
for all Ξ ( π β 1 ) π β = π β Λ Ξ ( π β 1 ) , i.e. π β πΊ . Hence Ξ ( π ) π β = π β Λ Ξ ( π ) π β π Ξ ( π ) = π β Λ Ξ ( π ) π = Ξ ( π ) π β π thus
commutes with π β π , and by the first lemma, Ξ for some π β π = π π . Then π β β for all β¨ π£ | π β π π£ β© = β¨ π£ | π π£ β© = β¨ π π£ | π π£ β© = π β π£ β 2 = β π π£ β π£ and therefore π£ β π which is real and nonnegative. Therefore either π = β π π£ β 2 β π£ β 2 whence π = 0 or π = π and π > 0 is unitary with the equivalence π = 1 β π π . Λ Ξ ( π ) = π Ξ ( π ) π β
Corollaries
Schurβs lemma in abelian categories
Let
Proof
Essentially, apply the Freyd-Mitchell theorem to the above proof for modules.
Footnotes
-
2023, Groups and representations, p. 31 β©
-
1996, Representations of finite and compact groups, Β§II.4, pp. 27β28. The proof offered here is virtually identical, but insists on using β-representations for reasons beyond me. β©