# Existence of a root between 2 given points

1. ### sara_87

773
1. The problem statement, all variables and given/known data

Show that there exists one root int (0,2) of the following function:

f(x)=(1-x^2)^2-√((1-x^2)*(1-1/2*x^2))

2. Relevant equations

3. The attempt at a solution

I first found:
f(0)=0
and
f(2)=7.268

But, i don't know what to do now. I'm not sure if it has something to do with the mean value theorem or not, or if i have to differentiate or not.

Any ideas will be very much appreciated.
Thank you.

2. ### HmBe

45
Well you've found a root at x=0, f(0)=0, but this isn't in the interval you're interested in, so it's best to try another value.

With finding a root in an interval your best bet is to try and find an a and a b in the interval such that f(a)<0 and f(b)>0, then by the intermediate value theorem there will be a root.

You've already got your b, such that f(b)>0, so now you just need to find your a.

3. ### sara_87

773
thank you.
so, i can take b=0.1
f(0.1)=-0.0124

but, this show that there exists at least 1 root in the interval, but i need to show that there exists exactly one root.
how would i go from there?
doesn't have something to do with the derivative?

4. ### MathematicalPhysicist

Well if I read your function correctly (if I am not mistaken you're a PhD student, then why don't you use TeX?), it's:
$$f(x)=(1-x^2)^2-\sqrt{(1-x^2)(1-\frac{1}{2} x^2)}$$

If this is your function then we can see that at x=1 the function vanishes, and you also know that at the boundaries at x=0 it equals 0 and at x=2 you get a positive function.

Now you should prove that for x in (0,1] there isn't another zero as in f(x) is always positive in this interval (non-negative), it's just a matter of playing with inequlities.

If you prove that f had the same sign all over the interval (0,1] then now you should show that if it were negative in (0,1) then in (1,2) it's a positive.

I'll show you the second part:
for x in (1,2) we have (1-x^2)^2>(1-4)^2=9
$$\sqrt{(1-x^2)(1-1/2 x^2)} < \sqrt{1/2}$$
so you get $$f(x)>9-\sqrt{1/2}>0$$.

Hopefully, I didn't get it wrong (with me it's always a possibility).

5. ### MathematicalPhysicist

Wait a minute, you want to show there exists exactly one, or that exists such root.

Cause I can now that there exist more than one root.

6. ### sara_87

773
thank you,
but I don't understand, where did you get the 4 from in:
(1-4)^2 ?

also, what do we do if we assume that we don't know that x=1 is a root?

7. ### sara_87

773
we want to show that there exists exactly 1 root

8. ### HmBe

45
I'd follow the method I said above for finding a root, to show that there is only one root I'd just do a rough sketch of the function in the interval you're interested in, because you're right - with a trickier function (something you're more likely to use IVT on) you won't know where the root is, so the inequality thing is much harder to do.

However, without looking at your mark scheme, I wouldn't know if this is enough. A sketch usually suffices for us though.

9. ### sara_87

773
I have plotted the graph on MATLAB, and i can see that there is exactly one root (namely x=1)
but we have to use another method (not a graphical method). We were told that finding the derivative is a clue, but i don't see why.

10. ### sara_87

773
I'm not sure if Rolle's theorem comes into this content?

11. ### HmBe

45
Can you show the function is even, and thus symmetrical? Then it will have another root in (-2,0).

Then you could show using the derivative that the root at x=0 is repeated, and because the root is quartic can only have one more root...

12. ### sara_87

773
Yes, the function is even since
f(-x)=f(x)
(since x appears only as x^2).

But, I don't understand the next part:
The derivative is

f'(x)= $-4 x \left(1-x^2\right)-\frac{-x \left(1-x^2\right)-2 x \left(1-\frac{x^2}{2}\right)}{2 \sqrt{\left(1-x^2\right) \left(1-\frac{x^2}{2}\right)}}$

should i solve f'(x)=0?

13. ### HmBe

45
Well I suppose you could cheat a little, showing that f'(0)=0 and that f(0)=0 (which you've already done) shows there is a repeated root at x=0, then you don't have to solve anything.

14. ### sara_87

773
Yes, I think that's what we are supposed to do.

but, how does showing that f'(0)=0 and that f(0)=0 show that there is a repeated root at x=0?

15. ### HmBe

45
I was pretty sure the definition of a repeated root was that if there is a root at x=a, that is f(a)=0, and f'(a)=0, then the root is repeated. Thinking about it, it seems to make sense, but I couldn't find anything on the internet to verify this.

16. ### Ray Vickson

6,305
There may be a problem with this approach in this case: as written, the function f(x) does not exist (in the reals) for 1 < x < sqrt(2). So, application of the intermediate-value method needs to be done with great care.

PS: I assume the OP means
f(x) = (1-x^2)^2 - sqrt[(1-x^2)(1 - (x^2/2))], not (1-x^2)^2 - sqrt[(1-x^2)(1 - 1/(2x^2))].

RGV

17. ### HmBe

45
Oh yes, just realised I made a pretty big mistake when I plotted the function, and that it's not continuous, so applying IVT would be very tricky.

18. ### sara_87

773
Oh right, i see.
If we assume we have a different function that IS continuous in (0,2) and that the IVT does apply, then we can use that f(0)=0 and f'(0)=0
But, what do you mean by:
''because the root is quartic can only have one more root''?

19. ### MathematicalPhysicist

Yes, I got it wrong. :-D

The sqrt there just pisses me off, I don't see any good estimate here.

Ok, I think I remembered my course I learnt in ODE2.

If you prove that at the interval (0,2) you have exactly two maximum points and one minimum at x=1, you've proven that there isn't anymore roots for this equation.

Am I wrong here again?
What I am trying to say is, if you would have found a minimum point at the interval s.t f(x0)<0 then we must have crossed the x axis somewhere, and then if we found two positive max, we should cross the x axis again, thus getting at least two roots. So indeed you should analyse the function via differentiation twice and sorting the max and min in this interval.

20. ### I like Serena

6,194
I don't think so. Rolle only says something about at least one point (same thing for IVT).

I believe that what you need is the definition of a absolute monoticity, which is implied by the derivative (being not zero).