Let us recall (TA, I, p. 43, def. 2) that this means that for every open subset V of U and every open covering (Vi)i∈I of V, the following conditions are satisfied:
(i) The mapping (rViV)i∈I:D′(V)→∏i∈ID′(Vi) is injective;
(ii) For every family (fi)∈∏i∈ID′(Vi) such that
r(Vi∩Vi′)Vi(fi)=r(Vi∩Vi′)Vi′(fi′)
for every pair (i,i′)∈I×I, there exists a distribution f∈D′(V) such that for every i∈I, one has rViV(f)=fi.
Let V be an open set of U and V=(Vi)i∈I an open covering of V. Let (Wj)j∈J be a locally finite open covering of V, finer than V and formed of relatively compact subsets (lemma 2 of IV, p. 201). Let us fix a partition of unity (φj)j∈J subordinate to the covering (Wj)j∈J such that the support of φj is contained in Wj for every j∈J (lemma 1 of IV, p. 196).
Let φ∈D(V). Since the set of j∈J such that Wj meets the support of φ is finite (TG, I, §9, no1, p. 59), one has
φ=j∈J∑φφj
where the sum consists of only a finite number of non-zero terms.
Let us prove (i). Suppose that f∈D′(V) satisfies rViV(f)=0 for every i∈I. This means that ⟨f,φ⟩=0 for every test function φ whose support is contained in one of the open sets Vi. This is fortiori true if the support of φ is contained in one of the open sets Wj. But then, for every φ∈D(V), we have
⟨f,φ⟩=⟨f,j∈J∑φφj⟩=j∈J∑⟨f,φφj⟩=0
hence f=0.
Let us prove (ii). Let (fi)i∈I be a family such that fi∈D′(Vi) for every i∈I and rVi∩Vi′,Vi(fi)=rVi∩Vi′,Vi′(fi′) for all i and i′ in I. Let ι:J→I be a mapping such that Wj⊂Vι(j) for every j∈J. For every j∈J, put fj=rWj,Vι(j)(fι(j)). We then have
where the sum is finite since only a finite number of terms can be non-zero.
The mapping λ is a linear form on D(V). Let us prove that it is a distribution. Let B be a bounded subset of D(V) and let K be a compact subset of V such that B⊂CK∞(V). By TG, I, §9, no1, p. 59, there exists a finite subset J′ of J such that
λ(φ)=j∈J′∑⟨fj,φφj∣Wj⟩
for every φ∈B. Since fj is a distribution for every j∈J′ and D(V) is a topological algebra, we deduce that λ is bounded on B, whence the result (lemma 4).
Let j∈J and let φ∈D(V) whose support is contained in Wj. We then have
by (1), since φφj′ has support contained in Wj∩Wj′. Consequently
⟨λ,φ⟩=⟨fj,j′∈J∑φφj′∣Wj⟩=⟨fj,φ∣Wj⟩
whence rWjV(λ)=fj for every j∈J.
Let i∈I. Let us show finally that the restriction of λ to Vi coincides with fi. By condition (i), applied to the covering of Vi by the open sets Wj, it is enough to verify that for every j∈J, the restriction of λ to Vi∩Wj coincides with that of fi. By what precedes, it is a question of verifying that the restriction of fi to Vi∩Wj is that of fj. Now one has