(1) f(f(x))=2x+f(−x),∀x∈Z
Note that, if a∈Z satisfies f(a)=0, then substituting x,y=a into the condition,
4a2=f(a+f(a))2=f(a)2=0.
We see that a=0. Now substitute y=0 to get
f(x)f(f(x))=f(x)(2x+f(−x)).
So if x=0, the conclusion holds. On the other hand, when x=0, it obviously also holds.
(2) f(x+f(x))=2x,∀x∈Z
We only need to consider the case where x is not 0. Substitute y=x into the condition:
f(x+f(x))2=4x2⇒f(x+f(x))=2x or −2x.
Suppose there exists b∈Z−{0} such that f(b+f(b))=−2b. Substituting b,2b+f(b) for x,y in the condition, we get
f(b+f(2b+f(b)))f(2b+2f(b))=0.
From the injectivity of f at 0, we know:
b+f(2b+f(b))=0 or 2b+2f(b)=0,
but the latter cannot hold, otherwise we would have f(b)=−b, and thus
f(−b)=f(f(b))=2b+f(−b)⇒b=0.
Hence f(2b+f(b))=−b. Therefore
4b2=f(b+f(b))2=f(2b+f(b)+f(2b+f(b)))2=4(2b+f(b))2.
That is, f(b)=−b or f(b)=−3b. From the earlier discussion we know the only possibility is f(b)=−3b, but then we would have
f(−b)=f(2b+f(b))=−b⇒f(b)=f(f(−b))+2b=b.
So b=0, a contradiction. Combining the above, the sub-conclusion holds.
(3) f is an injective odd function and f(2x)=2f(x), ∀x∈Z
First, substituting (−x), f(x) for x,y in the original problem gives
f(−x+f(f(x)))f(f(x)+f(−x))=0.
So f(f(x))=x or f(x)=−f(−x). But even if the former holds
⇒f(−x)=f(f(x))−2x=−xf(x)=f(f(−x))+2x=x=−f(−x).
The above discussion tells us f is an odd function. Then the first sub-conclusion becomes
f(f(x))=2x−f(x).
Thus f is injective, and finally by the second sub-conclusion
f(2x)=f(f(x))+f(f(x))=2f(x).
(4) If there exists an odd number in f(Z), then f(x)=x,∀x∈Z
Suppose the integer c satisfies that f(c) is odd, then substituting y=x+c into the original problem gives
f(x+f(x+c))f(x+c+f(x))=(2x+f(c))(2x+2c−f(c)).
In particular, f(x+c+f(x)) is odd. Using f(2x)=2f(x) we know:
x+f(x)≡0(mod2).
Then using f(2x)=2f(x) again we can deduce that
v2(f(x))=v2(x),∀x∈Z,
Substituting y=2x−f(x) gives
f(x+f(2x−f(x)))f(2x+f(x))=(2x+f(2−x−f(x)))(x−f(x)+f(2x+f(x))).
Note that f(2x+f(x))=x, so
f(x+f(2x−f(x)))=2x−f(x)=f(f(x)).
By injectivity, f(2x−f(x))=f(x)−x, so
v2(f(x)−x)=v2(f(2x−f(x)))=v2(2x−f(x)).
The only possibility is f(x)=x
(5) If f(Z) contains only even numbers, then f(x)=−2x,∀x∈Z
We can consider a new function g:Z→Z satisfying 2g(x)=f(x),∀x∈Z, then the condition can be rewritten as
g(x+2g(y))g(y+2g(x))=(x+g(y−x))(y+g(x−y)).
Substituting −f(x) for y in the above, we get
g(x+2g(−g(x)))g(g(x))=(x+g(−g(x)−x))(−g(x)+g(x+g(x))).
Since g inherits most of the properties of f, we have
(2x−g(x))2≥RHS=LHS=g(g(x))2=(2x−g(x))2.
Thus g(x+g(x))=2x+g(x). But g(2x)=2g(x) tells us
v2(g(x))≥v2(x),∀x∈Z.
So
g(x)=−x,∀x∈Z⇒f(x)=−2x,∀x∈Z.
Remark: In fact, the condition f(0)=0 is not needed; it can be derived using just the original condition. But stating it directly in the problem statement can reduce some computation. Below is the proof that f(0)=0: substituting x,y=0 gives
f(f(0))2=f(0)2⇒f(f(0))=f(0) or −f(0).
If f(f(0))=−f(0), then substituting x,y=f(0) gives
f(0)2=9f(0)2⇒f(0)=0.
If f(f(0))=f(0), then likewise substituting x,y=f(0) gives
f(2f(0))2=9f(0)2⇒f(2f(0))=3f(0) or −3f(0).
(i) f(2f(0))=3f(0): substituting x=f(0),y=0 gives
f(2f(0))f(f(f(0)))=(2f(0)+f(−f(0)))f(f(0)).
If f(0)=0 then f(−f(0))=f(0). Then substituting x=−f(0),y=0:
f(0)2=f(0)(f(f(0))−2f(0))⇒f(0)=3f(0).
Thus f(0)=0 is a contradiction, so f(0)=0.
(ii) f(2f(0))=−3f(0): at this point, again substituting x=f(0),y=0 gives
f(−f(0))=−5f(0)
Then substituting x=−f(0),y=0 gives
f(f(−f(0)))f(0)=−f(0)f(−f(0)).
Therefore when f(0)=0, f(−5f(0))=5f(0), and finally taking x,y=5f(0) gives
f(0)2=81f(0)2
which contradicts f(0)=0, so in this case, f(0)=0
Combining the above, f(0)=0.