Bourbaki

Exercise 20, § 4

Tag 0354, permanent. Cite it as [Bourbaki, Tag 0354].

Spectral Theories, IV, p. 349

Si aa and bb are real numbers with a<ba < b, we denote by C2([a,b])\mathscr{C}^2([a, b]) the vector space of continuous functions on [a,b][a, b] whose restriction to]a,b[]a, b[ is of class C2C^2. We denote by μ\mu the Lebesgue measure.

a) Let q1q_1 and q2q_2 be real-valued functions in C2([a,b])\mathscr{C}^2([a, b]) such that q1⩾q2q_1\geqslant q_2. For i∈{1,2}i\in \{1,2\}, let fi∈C2([a,b])f_i\in \mathscr{C}^2([a, b]) be a solution of the equation

fi′′+qifi=0 f''_i+q_if_i= 0

Suppose that f2(a)=f2(b)=0f_2(a) =f_2(b) = 0 and that f2≠0f_2\not = 0. If the function f1f_1 does not vanish in]a,b[]a, b[, then q1=q2q_1=q_2 and there exists c∈Cc\in \mathbf{C} such that f1=cf2f_1=cf_2. (Reduce to the case where f1f_1 and f2f_2 are strictly positive on]a,b[]a, b[; consider the wronskian w=f1f2′−f1′f2w=f_1f'_2-f'_1f_2 and prove that w(a)=w(b)w(a) =w(b), then that w=0.w= 0.)

b) Let q∈C2([0,1])q\in \mathscr{C}^2([0,1]) satisfying q⩾0q\geqslant 0. We denote by S~Lq\widetilde{S}L_q the partial differential operator on L2([0,1],μ)L^2([0,1], \mu) with domain

D={f∈C2([0,1])∣f(0)=f(1)=0} D =\{f\in \mathscr{C}^2([0,1])|f(0) =f(1) = 0\}

such that

S~Lq(f)=−f′′+qf \widetilde{S}L_q(f) =-f''+qf

(“Sturm–Liouville operator”). The operator S~Lq\widetilde{S}L_q is symmetric and essentially self-adjoint. We denote by SLq_q the closure of S~Lq\widetilde{S}L_q.

c) Let λ∈\lambda \in Sp(SLq_q) be an eigenvalue of SLq_q. The proper subspace associated with λ\lambda is of dimension 1 and generated by a real-valued function belonging to D.

d) Every eigenvalue λ\lambda of SLq_q satisfies λ⩾π2\lambda \geqslant \pi^2. (First prove that λ\lambda is positive by using question a) with q1=qq_1=q and q2=0.)q_2= 0.)

e) For s>0s >0, there exists a unique function fs∈C2([0,1])f_s\in \mathscr{C}^2([0,1]) which is a solution of the linear differential equation

−fs′′+qf−s2f=0 -f''_s+qf-s^2f= 0

and satisfies fs(0)=0f_s(0) = 0 and fs′(0)=sf'_s(0) =s. Put f(x,s)=fs(x)f(x, s) =f_s(x); the mapping ff is a function of class C2C^2 on]0,1[×R+∗]0,1[\times \mathbf{R}^*_+. The eigenvalues of SLq_q are of the form λ=s2\lambda =s^2 where ss satisfies f(1,s)=0f(1, s) = 0.

f) For t∈[0,1]t\in [0,1] and s>0s >0, let N(t,s)∈N‾N(t, s)\in \overline{\mathbf{N}} be the number of x∈[0,t]x\in [0, t] such that f(x,s)=0f(x, s) = 0. One has N(t,s)∈NN(t, s)\in \mathbf{N} and the function s↦N(t,s)s\mapsto N(t, s) is increasing for every fixed tt. (Use question a).)

g) Let t∈[0,1]t\in [0,1] and s0>0s_0>0. Let m=N(t,s0)∈Nm= N(t, s_0)\in \mathbf{N}. Let

0=x1(s0)<x2(s0)<...<xm(s0) 0 =x_1(s_0)< x_2(s_0)< . . . < x_m(s_0)

be the zeros of fs0f_{s_0} in [0,t][0, t]. For s⩾0s\geqslant 0, let

0=x1(s)<...<xm(s) 0 =x_1(s)< . . . < x_m(s)

be the first mm zeros of fsf_s in [0,t][0, t]. The functions s↦xj(s)s\mapsto x_j(s) are of class C1C^1 and one has

∂1(xj(s),s)xj′(s)=−∂2f(xj(s),s) \partial_1(x_j(s), s)x'_j(s) =-\partial_2f(x_j(s), s)

h) The set of eigenvalues of SLq_q is the set of the values of an increasing sequence of strictly positive real numbers

0<λ0<λ1<⋯<λn<⋯ 0< \lambda_0< \lambda_1<\cdots < \lambda_n<\cdots

tending to +∞+\infty. Moreover, the eigenspace corresponding to λn\lambda_n is of dimension 1; if φn\varphi_n is a basis of it, one has φn∈C2([0,1])\varphi_n\in \mathscr{C}^2([0,1]) and the function φn\varphi_n vanishes at exactly nn distinct points of [0,1][0,1]. (To prove that there exist infinitely many eigenvalues of SLq_q, use question a) with q1=M−s2q_1= M-s^2, where M∈RM\in \mathbf{R} is suitably chosen, to verify that N(1,s)N(1, s) tends to +∞+\infty when s→+∞.)s\rightarrow +\infty .)

Solution

No solution has been written yet.

Cited by

Nothing.

Cites

Nothing.

Machine translation of the English, by gpt-5-6-mini, gpt-5-mini. Not checked by a person. That model is a cut down version of the one the rest of this language was written by.