IMO 2012 #4
by Wolstenholme, Oct 29, 2014, 3:34 AM
Find all functions
such that, for all integers
that satisfy
, the following equality holds:
![\[f(a)^2+f(b)^2+f(c)^2=2f(a)f(b)+2f(b)f(c)+2f(c)f(a).\]](//latex.artofproblemsolving.com/a/9/e/a9edf1a3207d29facc970b6ee288d4fabdd43a4b.png)
(Here
denotes the set of integers.)
Solution:
So I thought this problem was pretty fun... but I've liked functional equations ever since I got lucky on USAMO 2014 #2 so I'm biased. Even so, it was so tedious that I kind of gave up at the end
Lemma 1:
Proof: Letting
yields that
which implies the desired result.
Lemma 2:
for all 
Proof: Letting
and applying Lemma 1 yields that
and by the equality case in AM-GM we have the desired result.
Now, letting
and applying Lemma 2 we find that
for all
Particularly, either
or
.
Case 1:
Then by letting
for
and using a quick induction we find that
for all
. Now, consider any two odd integers
and
. Letting
we have that
which by the equality case in AM-GM implies that
. Therefore we have found a solution 
Case 2:
Now letting
we find that
.
Case 2.1:
Letting
we find that
.
Case 2.1.1:
Now by a similar induction to the one used in Case 1 we have that
for all
. Now consider any two integers
and
that satisfy
. Letting
yields
which by the equality case in AM-GM implies that
Now, consider any two odd numbers
and
. Letting
we have that
which by the equality case in AM-GM implies that
By induction the values of
evaluated at odd integers are equal. By letting
to obtain the final restriction that
, we have found a solution 
Case 2.1.2:
Then
. Letting
we find that
so
and we default to Case 1 (in this case, we actually have
).
Case 2.2:
I will show that
for all
by strong induction on
Letting
and repeatedly applying the induction hypothesis we have that
If
for some
then letting
we find that
. If
then we default to Case 1. If
then
Letting
we find that
. If
then letting
we find by the AM-GM stuff that
and this becomes similar to past cases. Similarly if
we proceed similarly to Case 2.1.2. Therefore the only new solution is 
These three solutions can all be checked to work so we are done.



![\[f(a)^2+f(b)^2+f(c)^2=2f(a)f(b)+2f(b)f(c)+2f(c)f(a).\]](http://latex.artofproblemsolving.com/a/9/e/a9edf1a3207d29facc970b6ee288d4fabdd43a4b.png)
(Here

Solution:
So I thought this problem was pretty fun... but I've liked functional equations ever since I got lucky on USAMO 2014 #2 so I'm biased. Even so, it was so tedious that I kind of gave up at the end

Lemma 1:

Proof: Letting


Lemma 2:


Proof: Letting


Now, letting





Case 1:

Then by letting










Case 2:

Now letting


Case 2.1:

Letting


Case 2.1.1:

Now by a similar induction to the one used in Case 1 we have that

















Case 2.1.2:

Then





Case 2.2:

I will show that



















These three solutions can all be checked to work so we are done.
This post has been edited 1 time. Last edited by Wolstenholme, Oct 29, 2014, 3:34 AM