Show that this inner product is positive definite

5 replies · 2K views · last reply
pondzo
Messages
168
Reaction score
0

Homework Statement



Typed homework problem 5 on Sturm-Liouville equations: defines vector spaces V and V0 of functions on [0,1] with continuous derivatives, where f(0) = f(1) = 0 for V0, then defines operator T and inner product g, and asks to show g is positive definite and that the integral of f(x)Th(x) equals g(f,h).


Homework Equations

The Attempt at a Solution



I was able to do the second part of part a using integration by parts. But I am having no luck for the first part, proving that the inner product is positive definite. Pointers are appreciated!
 
Is this acceptable ?

##g(f,f)=\int_0^1 f(-p'f'-pf''+qf)dx\\~~~~~~~~~~~=\int_0^1-fp'f'-fpf''+qf^2dx\\~~~~~~~~~~~=\int_0^1-ff'p'dx-\int_0^1pff''dx+\int_0^1qf^2dx\\~~~~~~~~~~~=-[fpf']_0^1+\int_0^1p(f')^2dx+\int_0^1pff''dx-\int_0^1pff''dx+\int_0^1qf^2dx\\~~~~~~~~~~~=\int_0^1p(f')^2dx+\int_0^1pf^2dx##
Which is positive definite due to the square and p,q being positive definite and positive or zero.
 
Well, you do not have to do all of the partial integrations as this was the form the inner product was given on from the beginning, and the last p should be a q, but otherwise fine.
 
Oh of course, my mistake!

Do you think you would be able to give me some advice on part b ?

Printed homework problems 4 and 5 on Riesz representation and Sturm-Liouville inner products, with parts (a) and (b) of problem 5 underlined or circled in red.
 
pondzo said:
Is this acceptable ?

##g(f,f)=\int_0^1 f(-p'f'-pf''+qf)dx\\~~~~~~~~~~~=\int_0^1-fp'f'-fpf''+qf^2dx\\~~~~~~~~~~~=\int_0^1-ff'p'dx-\int_0^1pff''dx+\int_0^1qf^2dx\\~~~~~~~~~~~=-[fpf']_0^1+\int_0^1p(f')^2dx+\int_0^1pff''dx-\int_0^1pff''dx+\int_0^1qf^2dx\\~~~~~~~~~~~=\int_0^1p(f')^2dx+\int_0^1pf^2dx##
Which is positive definite due to the square and p,q being positive definite and positive or zero.

No, in the first part of the question, you are being asked to show that for all ##f\in V_0##

(a) ## g(f,f) \ge 0 ##
(b) ## g(f,f) = 0 \Rightarrow f = 0 ##

Point (a) is easy, point (b) is more subtle and continuity plays a central role.

EDIT:
I realized I didn't help much in my post: there is a theorem that says that if ##f: [a,b] \rightarrow \mathbb{R}## is continuous, non-negative, and such that ##\int_a^b f(x) \ dx = 0 ##, then ## f = 0 ##
 
Last edited: