Sunday, 31 December 2017

functional equations - Proof that solutions to Cauchy F.E. over mathbbQ are linear



I would like to prove that the solutions to Cauchy's functional equation over Q are linear, that is, all solutions f:QQ have the property f(x)=cx for some cQ.



The proof is to be done in a very specific order: First, we must prove that it works for all xN, then for all xZ. Third we must show it works for all 1x where xZ. Finally, we must show that it works for all pq where p,qZ, q0.



I am having trouble with the third part, where I have to show it works for 1x.
Here is what I have done up to this point:




Let f be a solution to the cauchy functional equation.
We know that f(0)=c0=0, as f(x)=f(0+x)=f(0)+f(x) for some number x.
Suppose xN and x0.
Then,
f(x)=f(1)+f(1)++f(1)x times=xf(1)=xc
If xZ and x0, then we have two cases: x>0 or x<0.
For x>0, the proof with xN suffices.

If x<0, then we have
f(x)=f(1)+f(1)++f(1)x times=xf(1)=xc



I don't know how to proceed with this method to show that f(1x)=cx.


Answer



The right way to proceed lies in the realization that your argument for xN can be extended slightly so say that if xN, then f(xk)=xf(k). Why? For the same reason you pointed out. If xN, then

f(xk)=f(k+k++k)=f(k)++f(k)=xf(k)This result is also what provides the last part of the question.



With this in hand, you're in fine shape! This is because you know f(x1x)=f(1)=c=xf(1x).


No comments:

Post a Comment

real analysis - How to find limhrightarrow0fracsin(ha)h

How to find lim without lhopital rule? I know when I use lhopital I easy get $$ \lim_{h\rightarrow 0}...