a)
Both
x→alim[f(x)+g(x)]andx→alim[f(x)g(x)]can exist.
For arbitrary a, let
f(x)={0,1,x≤a,otherwise,g(x)={1,0,x≤a,otherwise.Hence, we have that f(x)+g(x)=1 and f(x)g(x)=0. Both resulting functions are constant and thus have limits defined at x=a.
b)
Yes, limx→ag(x) must exist.
Let
lf=x→alimf(x),L=x→alim[f(x)+g(x)],lg=L−lf.Choose any ϵ>0. By definition, there exist δ1 and δ2 such that
0<∣x−a∣<δ10<∣x−a∣<δ2⟹∣f(x)−lf∣<2ϵ,⟹∣f(x)+g(x)−(lf+lg)∣<2ϵ.Let δ=min(δ1,δ2). We thus have, for 0<∣x−a∣<δ,
∣g(x)−lg∣=∣g(x)−lg+f(x)−f(x)+lf−lf∣=∣f(x)+g(x)−(lf+lg)−(f(x)−lf)∣≤∣f(x)+g(x)−(lf+lg)∣+∣f(x)−lf∣<2ϵ+2ϵ=ϵ.c)
x→alim[f(x)+g(x)]cannot exist if limx→af(x) exists and limx→ag(x) does not exist. This can be proven with a proof by contradiction along the lines of the proof in b). We show an alternative proof that follows from the statement proven in b). We define the following propositions:
PQR:x→alimf(x) exists,:x→alimg(x) exists,:x→alim[f(x)+g(x)] exists.We want to prove that
P∧¬Q⟹¬R.This is equivalent to
¬(P∧¬Q)∨¬R≡¬P∨Q∨¬R≡¬P∨¬R∨Q≡¬(P∧R)∨Q≡P∧R⟹Q.and because the last statement was proven in b), so the original implication is also true.
d)
x→alimg(x)does not necessarily exist.
As a counterexample, let f(x)=0 and g(x)=x1. Notice that
x→0limf(x)g(x)=0,but limx→0g(x) does not exist.