Monday, 27 February 2017

calculus - Graph of continuous function from [0,1] to [0,1].




I know that the graph of continuous functions on [0,1] is a null set (has Lebesgue measure zero). My question is following and seems relate to that I have conceded. Let f:[0,1][0,1] be a continuous and strictly increasing function. Then for any nN, we can cover the graph of f by n rectangle, with side parallel x-and y-axes, such that each rectangle has area =1n2. (Rectangles can be overlap).



How to prove this? Any idea?


Answer



First, to prevent the notation from exploding, extend f to an increasing function f:[0,)[0,) which tends to infinity at infinity.



Since f is increasing,






Note: If $0\le a




Now let x0=0. Since f is increasing and continuous we can recursively choose x1,x2, so that xj+1>xj and (xj+1xj)(f(xj+1)f(xj))=1n2.



If we can show xn1 we're done; then we've covered the graph of f by n rectangles of area 1/n2. Since f([0,1])[0,1] it's enough to show that xn1 or f(xn)1 (because if xn<1 then also f(xn)<1, which is to say f(xn)1 implies xn1).



But Cauchy-Schwarz shows that 1=n1j=0(xj+1xj)1/2(f(xj+1f(xj))1/2((xj+1xj)(f(xj+1f(xj)))1/2=((xn)(f(xn)f(0))1/2;

hence either xn1 or f(xn)f(0)1.


No comments:

Post a Comment

real analysis - How to find limhrightarrow0fracsin(ha)h

How to find limh0sin(ha)h without lhopital rule? I know when I use lhopital I easy get $$ \lim_{h\rightarrow 0}...