Bourbaki

Remark

Tag 05WY, permanent. Cite it as [Bourbaki, Tag 05WY].

Algebra, chapter II, § 10. Matrices

(1) We now start with a left AA-module EE with two bases (ei),(ei′)(e_i), (e'_i), each with nn elements; if we write ei′=∑j=1najieie'_i = \sum_{j=1}^n a_{ji} e_i, P=(aji)P = (a_{ji}) is also called the matrix of passing from (ei)(e_i) to (ei′)(e'_i); it is also the matrix of the automorphism of EE such that u(ei)=ei′u(e_i) = e'_i, with respect to the basis (ei)(e_i) and also the matrix of lEl_E with respect to the bases (ei′)(e'_i) and (ei)(e_i) in that order. The above results then hold with only the following modifications: formulae (33 D) to (36 D) are respectively replaced by

(33 G)tM′(u)=tP.tM(u).tQ−1(34 G)tM′(u)=tP.tM(u).tP−1(35 G)tM(u)=tM′(u).tP. \begin{align*} (33\text{ G}) &\quad {}^t M'(u) = {}^t P . {}^t M(u) . {}^t Q^{-1} \\ (34\text{ G}) &\quad {}^t M'(u) = {}^t P . {}^t M(u) . {}^t P^{-1} \\ (35\text{ G}) &\quad {}^t M(u) = {}^t M'(u) . {}^t P. \end{align*}

LINEAR ALGEBRA

(36 G)

xi=∑j=1nxj′aij(1≤i≤n). x_i = \sum_{j=1}^n x_j' a_{ij} \quad (1 \leq i \leq n).

(2) Under the hypotheses of Proposition 4, consider an element x∗∈E∗x^* \in \mathbf{E}^*; as the matrix of passage from (ei∗)(e_i^*) to (ei′∗)(e_{i'}^*) is tP−1tP^{-1} (Proposition 5), for the matrices M(x∗)M(x^*) and M′(x∗)M'(x^*) of x∗x^* with respect to these two bases respectively,

tM(x∗)=tM′(x∗).P−1 tM(x^*) = tM'(x^*) . P^{-1}

or also

(37 D)

tM′(x∗)=tM(x∗).P tM'(x^*) = tM(x^*) . P

which is equivalent to the system of equations

(38 D)

xi′∗=∑j=1nxj∗aji(1≤i≤n) x_{i'}^* = \sum_{j=1}^n x_j^* a_{ji} \quad (1 \leq i \leq n)

for the elements (xi∗)(x_i^*) and (xi′∗)(x_{i'}^*) of the matrices M(x∗)M(x^*) and M′(x∗)M'(x^*). The corresponding formulae for a left A-module E are

(37 G)

M′(x∗)=tP.M(x∗) M'(x^*) = tP . M(x^*)

(38 G)

xi′∗=∑j=1najixj∗(1≤i≤n). x_{i'}^* = \sum_{j=1}^n a_{ji} x_j^* \quad (1 \leq i \leq n).

(3) Let A, B be two rings, σ:A→B\sigma : A \to B a homomorphism of A into B, E a right (resp. left) A-module, (ei)(e_i), (ei′)(e_i') two bases with nn elements of E, F a right (resp. left) B-module, (fj)(f_j), (fj′)(f_j') two bases with mm elements of F and PP (resp. QQ) the matrix of passage from (ei)(e_i) to (ei′)(e_i') (resp. from (fj)(f_j) to (fj′)(f_j')).

For every semi-linear mapping u:E→Fu : E \to F, relative to σ\sigma, let M(u)M(u) be the matrix of uu with respect to (ei)(e_i) and (fj)(f_j) and M′(u)M'(u) its matrix with respect to (ei′)(e_i') and (fj′)(f_j'). Then

(39 D)

M′(u)=Q−1M(u)σ(P) M'(u) = Q^{-1} M(u) \sigma(P)
(resp. (39 G)
tM′(u)=σ(tP).tM(u).tQ−1. tM'(u) = \sigma(tP) . tM(u) . tQ^{-1}.

The proof is the same as that for (33 D) and (33 G), this time using formulae (27 D) and (27 G) (no. 6).

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

Cited by

Nothing.

Cites

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