Bourbaki

Proposition 12

Tag 068E, permanent. Cite it as [Bourbaki, Tag 068E].

Algebra, chapter III, § 11. Cogebras, products of multilinear forms, inner products and duality

*Let M be a free A-module of dimension n; let e∈∧n(M)e \in \wedge^n(M) be an element forming a basis of ∧n(M)\wedge^n(M) and let e∗e^* be the element of ∧n(M∗)\wedge^n(M^*) such that {(−1)n(n−1)/2θ∧(e∗)}\{(-1)^{n(n-1)/2} \theta_{\wedge}(e^*)\} is the dual basis of {e}\{e\} in (∧n(M))∗(\wedge^n(M))^*. Let ϕ:∧(M∗)→∧(M)\phi : \wedge(M^*) \to \wedge(M) be the mapping z↦z∧e∗z \mapsto z \wedge e^* and ϕ′:∧(M∗)→∧(M)\phi' : \wedge(M^*) \to \wedge(M) the mapping z∗↦z∗∧ez^* \mapsto z^* \wedge e. Let ϕp\phi_p (resp. ϕp′\phi'_p) be the restriction of ϕ\phi (resp. ϕ′\phi') to ∧p(M)\wedge^p(M) (resp. ∧p(M∗)\wedge^p(M^*)). Then:

(i) The mapping ϕ\phi is a left ∧(M)\wedge(M)-module isomorphism and the mapping ϕ′\phi' is a left ∧(M∗)\wedge(M^*)-module isomorphism; moreover the mappings ϕ\phi and ϕ′\phi' are inverses of one another.

(ii) *The mapping ϕp\phi_p is an isomorphism of the A-module ∧p(M)\wedge^p(M) onto the A-module ∧n−p(M∗)\wedge^{n-p}(M^*) and the mapping ϕp′\phi'_p is an isomorphism of the A-module ∧p(M∗)\wedge^p(M^*) onto the A-module ∧n−p(M)\wedge^{n-p}(M).

(iii) *If we write B(u,v∗)=⟨u,θ∧(v∗)⟩B(u, v^*) = \langle u, \theta_{\wedge}(v^*) \rangle for u∈∧(M)u \in \wedge(M) and v∗∈∧(M∗)v^* \in \wedge(M^*) then, for u∗∈∧p(M∗)u^* \in \wedge^p(M^*) and v∗∈∧n−p(M∗)v^* \in \wedge^{n-p}(M^*),

B(ϕp′(u∗),v∗)=(−1)p(n−p)B(u∗,ϕn−p′(v∗)). B(\phi'_p(u^*), v^*) = (-1)^{p(n-p)} B(u^*, \phi'_{n-p}(v^*)).

The fact that ϕ\phi is ∧(M)\wedge(M)-linear and ϕ′\phi' is ∧(M∗)\wedge(M^*)-linear follows from the formulae (u∧v)∧e∗=u∧(v∧e∗)(u \wedge v) \wedge e^* = u \wedge (v \wedge e^*) and (u∗∧v∗)∧e=u∗∧(v∗∧e)(u^* \wedge v^*) \wedge e = u^* \wedge (v^* \wedge e) (no. 6, formula (37)), using the fact that θ∧\theta_{\wedge} is an isomorphism of ∧(M∗)\wedge(M^*) onto the opposite algebra to ∧(M)∗\wedge(M)^*. On the other hand there exists a basis (ei)1≤i≤n(e_i)_{1 \leq i \leq n} of M such that

e=e1∧e2∧⋯∧enande∗=(−1)n(n−1)/2e1∗∧e2∗∧⋯∧en∗, e = e_1 \wedge e_2 \wedge \cdots \wedge e_n \quad \text{and} \quad e^* = (-1)^{n(n-1)/2} e_1^* \wedge e_2^* \wedge \cdots \wedge e_n^*,

where (ei∗)(e_i^*) is the basis dual to (ei)(e_i). We write I={1,n}I = \{1, n\}; it follows from (76) that, for every subset J of I with p elements,

{ϕ(eJ)=(−1)n(n−1)/2+p(p−1)/2ρJ,I−JeI∗Jϕ′(eJ∗)=(−1)p(p−1)/2ρJ,I−JeI−J. \begin{cases} \phi(e_J) = (-1)^{n(n-1)/2 + p(p-1)/2} \rho_{J, I-J} e_I^* J \\ \phi'(e_J^*) = (-1)^{p(p-1)/2} \rho_{J, I-J} e_{I-J}. \end{cases}

This proves that ϕ\phi and ϕ′\phi' are bijective; moreover ρJ,I−JρI−J,J=(−1)p(n−p)\rho_{J, I-J} \rho_{I-J, J} = (-1)^{p(n-p)} (§ 7, no. 8, formula (21)); as the number

n(n−1)2+p(p−1)2+(n−p)(n−p−1)2+p(n−p)=n(n−1) \frac{n(n-1)}{2} + \frac{p(p-1)}{2} + \frac{(n-p)(n-p-1)}{2} + p(n-p) = n(n-1)

is even, it follows that ϕ\phi and ϕ′\phi' are inverses of one another. Finally, to prove (77), it suffices to take u∗=eJ∗u^* = e_J^* and v∗=eI−J∗v^* = e_{I-J}^*; the verification also follows from the definition of θ∧\theta_{\wedge}, formulae (78) and the relation ρJ,I−JρI−J,J=(−1)p(n−p)\rho_{J, I-J} \rho_{I-J, J} = (-1)^{p(n-p)} (§ 7, no. 8, formula (21)). Note that, for u∗∈∧p(M∗)u^* \in \wedge^p(M^*) and v∗∈∧n−p(M∗)v^* \in \wedge^{n-p}(M^*),

B(\phi_p'(u^*), v^*) is, to within a sign, the coefficient of u∗∧v∗u^* \wedge v^* with respect to the basis {e∗}\{e^*\} of Λn(M∗)\Lambda^n(M^*).

Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.

Cited by

Cites

Transcribed from the printed edition, 1998, Springer. Not translated.