Let Ts be the topology of simple convergence on L(E). Let f∈Cb(R). For every sequence (un)n∈N in A(E) which converges to u for the Ts-resolvent topology, the sequence (f(un))n∈N converges to f(u) in L(E) endowed with the topology Ts.
Let (un) be a sequence in A(E) which converges to u for the Ts-resolvent topology. Let x∈E and let ε>0.
For every integer N∈N, let φN∈K(R) be a function with support contained in [−(N+1),N+1] such that 0⩽φN⩽1 and φN(t)=1 for every t∈[−N,N]. The functions φN converge simply to 1 as N tends to infinity, and satisfy ∣φN∣⩽1, hence prop. 6 of IV, p. 276 implies that there exists N∈N such that ∥φN(u)x−x∥⩽ε. Put fN=fφN.
For every n∈N, one has
∥f(un)x−f(u)x∥⩽∥f(un)x−fN(un)x∥+(15)
∥fN(un)x−fN(u)x∥+∥fN(u)x−f(u)x∥
For every v∈A (E), one has
∥f(v)x−fN(v)x∥=∥(f(1−φN))(v)x∥
⩽∥f(v)∥∥x−φN(v)x∥⩽∥f∥∞∥x−φN(v)x∥
(prop. 5, c) of IV, p. 275). Since φN∈C0(R), assertion a) of prop. 11 applied to the Ts-resolvent topology implies that φN(un)x converges to φN(u)x when n→+∞. Consequently, for every sufficiently large n, one has
∥x−φN(un)x∥⩽∥x−φN(u)x∥+ε∥x∥⩽(1+∥x∥)ε
Inequality (15) then becomes
∥f(un)x−f(u)x∥⩽∥f∥∞(2+∥x∥)ε+∥fN(un)x−fN(u)x∥
for every sufficiently large n. The proof is concluded by means of assertion a) of loc. cit., applied to the function fN∈C0(R) and to the Ts-resolvent topology.
Bourbaki sets the proof directly after the statement, and the corpus does not separate them, so what is above is the statement and its proof together.
Nothing.
Nothing.