Statement

Let VV and WW be modules over a . There is a canonical isomorphism of graded algebras

Λ(VW)ΛV^ΛW,\Lambda(V\oplus W)\cong \Lambda V\,\widehat\otimes\,\Lambda W,

where ^\widehat\otimes is the graded tensor product. Under this isomorphism,

(v,0)v1,(0,w)1w.(v,0)\longmapsto v\otimes1, \qquad (0,w)\longmapsto1\otimes w.

The multiplication on the right uses the

(ab)(ab)=(1)deg(b)deg(a)(aa)(bb)(a\otimes b)(a'\otimes b') =(-1)^{\deg(b)\deg(a')}(a\wedge a')\otimes(b\wedge b')

for homogeneous elements.

Degree-by-degree form

Taking the homogeneous part of degree kk gives a canonical decomposition

Λk(VW)p+q=kΛpVΛqW.\Lambda^k(V\oplus W) \cong \bigoplus_{p+q=k}\Lambda^pV\otimes\Lambda^qW.

Explicitly, the summand indexed by (p,q)(p,q) maps

(v1vp)(w1wq)(v_1\wedge\cdots\wedge v_p)\otimes (w_1\wedge\cdots\wedge w_q)

to the wedge of the corresponding vectors in VWV\oplus W, with all VV-vectors written before the WW-vectors. The graded sign rule makes this prescription independent of how products are regrouped.

Equivariance

If a group or acts on both VV and WW, the direct sum carries the diagonal action and the displayed isomorphisms are equivariant. In terms of ,

Λk(VW)p+q=kΛpVΛqW\Lambda^k(V\oplus W) \cong \bigoplus_{p+q=k}\Lambda^pV\otimes\Lambda^qW

is therefore an isomorphism of representations.

Setting k=dimV+dimWk=\dim V+\dim W recovers the determinant identity

det(VW)det(V)det(W).\det(V\oplus W)\cong\det(V)\otimes\det(W).
Why the tensor product must be graded

Using the ordinary ungraded tensor-product multiplication would make elements from VV commute with elements from WW. In Λ(VW)\Lambda(V\oplus W) they anticommute in degree one, so the Koszul sign is essential.

References
  1. Nicolas Bourbaki, Algebra I: Chapters 1–3, Springer, 1989, Chapter III. Publisher record.
  2. Christian Kassel, Quantum Groups, Springer, 1995, Chapter XI, §1. Publisher record.