Solution: To begin, we show that A is a set of surjective functions. Indeed, for f∈A and f1=f2=f, with x=−y,
f(f(y)+y)=g(0)+2y
and thus f is surjective.
Furthermore, with x=0, we obtain that if f1,f2∈A then f1∘f2∈A.
Since A is non-empty, we take f∈A. By the previous remark, we know that f(k)∈A for all k∈N. As A is finite, there exist k1>k2 such that f(k1)=f(k2). Since f(k2)∈A, f(k2) is surjective and thus with y=f(k2), we obtain
f(k1−k2)(y)=y,∀y∈R
Since f(k1−k2)∈A, A contains the identity. It is easily verified that A={id} is a possible set. We now show that A cannot contain anything other than the identity.
By setting x=0 in our condition, we obtain that g=f1∘f2 and therefore for all f1,f2∈A
f1(f2(y)−x)+2x=f1(f2(x+y))
If A contained a function h other than the identity, with f1=id,f2=h and y=0, we would have
h(x)=x+h(0) and thus h(n)(x)=x+nh(0)
Since A is finite and h(n)∈A for all n∈N, we have h(0)=0 and thus h≡id.