a) We know that ∥cf∥ must exist since {∣cf(x)∣:x∈[0,1]} is bounded. For contradiction, suppose that ∣c∣∥f∥=∥cf∥ for some c=0.
If ∣c∣∥f∥<∥cf∥, there must exist x∈[0,1] such that
∥cf∥−(∥cf∥−∣c∣∥f∥)<∣cf(x)∣≤∥cf∥.Since ∥cf∥ is the least upper bound of {∣cf(x)∣:x∈[0,1]}, this means that
∣c∣∥f∥<∣c∣∣f(x)∣⟹∥f∥<∣f(x)∣,which cannot be because ∥f∥ is an upper bound of ∣f(y)∣ for y∈[0,1].
Likewise, if ∣c∣∥f∥>∥cf∥, then ∥f∥>∣c∣1∥cf∥, and there must exist x∈[0,1] such that
∥f∥−(∥f∥−∣c∣1∥cf∥)<∣f(x)∣≤∥f∥.This implies
∥cf∥<∣cf(x)∣≤∣c∣∥f∥,which invalidates ∥cf∥ as an upper bound. Hence, it ensues that ∥cf∥=∣c∣∥f∥ for all c. We left out the proof that ∥cf∥=∣c∣∥f∥ for c=0 because it is bafflingly trivial.
b) Since ∥f+g∥ is a least upper bound, for any ε>0 there exists x∈[0,1] such that
∥f+g∥−ε<∣f(x)+g(x)∣≤∥f+g∥.This implies that
∥f+g∥<ε+∣f(x)+g(x)∣≤ε+∣f(x)∣+∣g(x)∣≤ε+∥f∥+∥g∥.Hence,
∥f+g∥≤∥f∥+∥g∥.c) This inequality follows from b) by defining s=h−g and p=g−f; we then have
∥s+p∥≤∥s∥+∥p∥.