20100409, 17:17  #1 
Apr 2010
17 Posts 
extremely difficult problem with a function.
Let f:R>R is a function which its second derivative f'' is continuous function.
Prove that if lim(x>infinity) (f(x)+ f '(x) + f ''(x))=L then lim(x>infinity) f(x)=L. L is in R*. Note: this exercise is not like a usually exercise for home, is like a challenge exercise and is very hard. If anyone can think anything that would help me just write it down.I will appreciate any kind of help. if you want you can ask me and i will post my thoughts about this exercise(how we can try to solve it). please write back if anything is not clear or if you want to ask about something. thank you. 
20100409, 18:45  #2  
"Bob Silverman"
Nov 2003
North of Boston
16522_{8} Posts 
Quote:
Start with the following. Suppose lim x>oo f(x) = L. What does this say about lim x>oo f'(x)??? can lim x>oo f'(x) = M, M !=0 and still have lim x>oo f(x) = L?? i.e. if the slope isn't 0 as x>oo what happens to f(x)? Similarly if f' >0 as x>oo, what can we say about f''?? i.e. start by looking at the converse of the problem. Then consider the following. if lim x>oo f + f' > (say) K, what can we say about f and f' individually? BTW, I recall seeing this exercize in my undergrad days. It may be in Rudin, IIRC. 

20100409, 19:06  #3  
"Bob Silverman"
Nov 2003
North of Boston
2·3^{3}·139 Posts 
Quote:
Further hint (I omit the lim x>oo in my notation) if (f + f') > L, and f > M, then f' > LM. but if f' > LM, what happens to f? If f is not bounded, then f must go to oo and f' must go to oo. Can this happen???? [clearly if both go to oo, then their sum does as well, and if one is bounded and the other is not then their sum is again unbounded]. 

20100409, 20:47  #4  
"Tapio Rajala"
Feb 2010
Finland
13B_{16} Posts 
Quote:
The reason I replied is because I do not see how Silverman's hints might help in solving the problem. I might just be to sleepy to notice it. Here a small observation those hints alarmed me to point out. One can not a priori assume the limits , and to exist. Otherwise one could prove the claim in a same way  a claim which is trivially false. 

20100409, 21:12  #5  
Apr 2010
17 Posts 
Quote:
Thanks for your hints..you are just very near my thoughts..actually I am trying to prove the following: if a function g has continuous derivative g' and lim(x>oo)g(x) is in R then lim(x>oo)g'(x)=0. I am quite sure that this statement is true( i hope so)..i couldn't find a counter example so at this moment i am trying to prove it..if i manage or anyone else from this forum manage to prove that lemma(it shouldn't be very difficult i believe) then our problem would take easier form, because if we assume that lim f(x) exists and is in R then from the lemma lim f'(x)=0 and hence lim f''(x)=0 so is true. So it would remain to prove the problem by contradiction assuming that lim f(x) doesn't exist. I wait for your replies or any other thoughts..thanks 

20100409, 21:23  #6 
"Tapio Rajala"
Feb 2010
Finland
3^{2}×5×7 Posts 

20100409, 21:44  #7  
Apr 2010
17 Posts 
Quote:
I will come back with something new.. 

20100410, 13:24  #8 
Apr 2010
17 Posts 
I am just trying to figure out. if the condition from the problem holds with L in R, then it has to be limf '(x)=limf ''(x)=0??..can anybody find an example such that limf '(x)=  limf ''(x)<>0 or
such that lim f '(x) and lim f ''(x) don't exist but does exist (lim f '(x) + lim f''(x)) and that is 0?? if nobody will find such an example i will begin trying to prove that limf '(x)=limf ''(x)=0..i wait for reply Last fiddled with by kakos22 on 20100410 at 13:28 
20100410, 13:58  #9 
"Bob Silverman"
Nov 2003
North of Boston
2·3^{3}·139 Posts 

20100410, 14:02  #10 
Apr 2010
17 Posts 

20100410, 17:18  #11 
Apr 2010
17_{10} Posts 
I am just trying to prove the following lemma:
Let function g has continuous derivative g' and lim x>oo g(x) exists and is in R. Then lim x>oo g '(x) doesn't exist or exists and is 0. I think that this time the lemma is true.If that so then the beginning problem will be easier. if anybody could found a counter example or something to help me with that i will appreciate it. 
Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Extremely lucky assignments  apocalypse  GPU to 72  6  20150407 04:41 
Extremely basic questions  schwerlin  Information & Answers  6  20150120 19:26 
Extremely nonsmooth Factor of M61^1201  kosta  Factoring  10  20130329 14:23 
Calculating a difficult sum  CRGreathouse  Math  3  20090825 14:11 
extremely large numbers  TimSorbet  Programming  10  20080731 17:04 