Bourbaki

Proposition 8

Tag 08D7, permanent. Cite it as [Bourbaki, Tag 08D7].

Functions of a Real Variable, chapter IV, § 2. LINEAR DIFFERENTIAL EQUATIONS

The set of integrals of the homogeneous linear equation (37), defined on JJ, is a vector space of dimension nn over the field C\mathbf{C}.

Given any nn integrals uju_j (1≤j≤n1 \leq j \leq n) of equation (37) one calls the Wronskian of this system of integrals the determinant (with respect to the canonical basis of Cn\mathbf{C}^n) of the corresponding nn integrals uj\mathbf{u}_j of the equation dx/dt=A(t).xd\mathbf{x}/dt = A(t).\mathbf{x}, that is to say, the function

W(t)=∣u1(t)u2(t)…un(t)Du1(t)Du2(t)…Dun(t)⋮⋮⋱⋮Dn−1u1(t)Dn−1u2(t)…Dn−1un(t)∣. W(t) = \begin{vmatrix} u_1(t) & u_2(t) & \ldots & u_n(t) \\ Du_1(t) & Du_2(t) & \ldots & Du_n(t) \\ \vdots & \vdots & \ddots & \vdots \\ D^{n-1}u_1(t) & D^{n-1}u_2(t) & \ldots & D^{n-1}u_n(t) \end{vmatrix}.

For the nn integrals uju_j to be linearly independent it is necessary and sufficient that W(t)≠0W(t) \neq 0 on JJ; moreover, it suffices for this that W(t0)≠0W(t_0) \neq 0 at only one t0t_0 of JJ (IV, p. 184, prop. 4); further, one has (IV, p. 184, prop. 5)

W(t)=W(t0)exp⁡(∫t0ta1(s) ds). W(t) = W(t_0) \exp \left( \int_{t_0}^t a_1(s)\, ds \right).
(38)

We identify the resolvent C(t,t0)C(t, t_0) of equation (35) with its matrix with respect to the canonical basis of Cn\mathbf{C}^n; the columns vj(t,t0)\mathbf{v}_j(t, t_0) (1≤j≤n1 \leq j \leq n) of this matrix are then nn linearly independent integrals

vj(t,t0)=(vj(t,t0),Dvj(t,t0),…,Dn−1vj(t,t0)) \mathbf{v}_j(t, t_0) = (v_j(t, t_0), Dv_j(t, t_0), \ldots, D^{n-1}v_j(t, t_0))

of the homogeneous equation dx/dt=A(t).xd\mathbf{x}/dt = A(t).\mathbf{x} which correspond to nn linearly independent integrals vj(t,t0)v_j(t, t_0) of equation (37) such that

Dk−1vj(t0,t0)=δjk D^{k-1}v_j(t_0, t_0) = \delta_{jk}

(Kronecker delta) for 1≤j≤n,1≤k≤n1 \leq j \leq n, 1 \leq k \leq n (on agreeing to put D0vj=vjD^0 v_j = v_j). It results in particular that the method of variation of parameters (IV, p. 182) applied to equation (35) here gives as a particular integral of (33), equal to 0 together with its first n−1n-1 derivatives at the point t0t_0, the function

w(t)=∫t0tvn(t,s)b(s) ds. w(t) = \int_{t_0}^t v_n(t, s)b(s)\, ds.
(39)

In the particular case of the equation Dnx=b(t)D^n x = b(t) the formula (39) again gives the formula expressing the nthn^{th} primitive of the regulated function b(t)b(t) which vanishes with its first n−1n - 1 derivatives at the point t0t_0, namely

w(t)=∫t0tb(s)(t−s)n−1(n−1)! ds w(t) = \int_{t_0}^t b(s) \frac{(t-s)^{n-1}}{(n-1)!} \, ds
(II, p. 62, formula (19)): the integral of Dnx=0D^n x = 0 which vanishes together with its first n−2n - 2 derivatives at the point t0t_0, and whose n−1n - 1th derivative is equal to 1 there, is actually the polynomial (t−t0)n−1/(n−1)!(t-t_0)^{n-1}/(n-1)!.

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 volume. Not translated.