Proving the Validity of a New Metric Space

Join the discussion
Registration is free. Start your own thread to ask a follow-up.
8 replies · 3K views
Lily@pie
Messages
108
Reaction score
0
Suppose that (X,d) is a metric

Show [itex]\tilde{d}[/itex](x,y) = [itex]\frac{d(x,y)}{\sqrt{1+d(x,y)}}[/itex] is also a metric

I've proven the positivity and symmetry of it.

Left to prove something like this

Given a[itex]\leq[/itex]b+c
Show [itex]\frac{a}{\sqrt{1+a}}[/itex][itex]\leq[/itex][itex]\frac{b}{\sqrt{1+b}}[/itex]+[itex]\frac{c}{\sqrt{1+c}}[/itex]

I try to prove this
a[itex]\sqrt{1+b}[/itex][itex]\sqrt{1+c}[/itex]=b[itex]\sqrt{1+a}[/itex][itex]\sqrt{1+c}[/itex]+c[itex]\sqrt{1+a}[/itex][itex]\sqrt{1+b}[/itex]
but I'm just stuck!

cz previously I've proven this [itex]\frac{a}{1+a}[/itex][itex]\leq[/itex][itex]\frac{b}{1+b}[/itex]+[itex]\frac{c}{1+c}[/itex] before...

I guess I can't use the same method??
 
Physics news on Phys.org
Lily@pie said:
Given a[itex]\leq[/itex]b+c
Show [itex]\frac{a}{\sqrt{1+a}}[/itex][itex]\leq[/itex][itex]\frac{b}{\sqrt{1+b}}[/itex]+[itex]\frac{c}{\sqrt{1+c}}[/itex]
Can you prove this assuming a is bigger than both b and c? Can you prove it assuming a is not the biggest?
 
What do you mean? I know a≤b+c...

So I assume a is bigger than both a and c?
 
Lily@pie said:
What do you mean?
I mean that if you can prove an assertion under some condition, and then under the complementary condition, then you are done. And I have also told you what condition works well in your case.
 
This is natural to assume given the triangle inequality for the original metric.
 
Let:
a=d(x,y)
b=d(x,z)
c=d(z,y)

a[itex]\leq[/itex]b+c
This is the triangle inequality for the original metric right?

Show [itex]\frac{a}{\sqrt{1+a}}[/itex][itex]\leq[/itex][itex]\frac{b}{\sqrt{1+b}}[/itex]+[itex]\frac{c}{\sqrt{1+c}}[/itex] -- (*)

Does this means that I try to prove (*) by two cases:
1. assuming a>b and a>c
2. assuming a<b and a<c?
 
I'm not sure whether this way of proving it is okay:

I take a look at the function

f(x)=[itex]\frac{x}{\sqrt{1+x}}[/itex]

f'(x)=[itex]\frac{2+x}{2(1+x)^{\frac{3}{2}}}[/itex]
f'(x)>0 if and only if x≥-1.

Since I know that 0≤a≤b+c, f(x) is an increasing function in the interval [0,b+c]. Therefore, by choosing a=b+c, f(a)=[itex]\frac{a}{\sqrt{1+a}}[/itex]=[itex]\frac{b+c}{\sqrt{1+b+c}}[/itex] is the maximum possible value.

Hence it is enough to show that [itex]\frac{b+c}{\sqrt{1+b+c}}[/itex][itex]\leq[/itex][itex]\frac{b}{\sqrt{1+b}}[/itex]+[itex]\frac{c}{\sqrt{1+c}}[/itex]

Since 1+b+c≥1+b and 1+b+c≥1+c,
[itex]\frac{b+c}{\sqrt{1+b+c}}[/itex]=[itex]\frac{b}{\sqrt{1+b+c}}[/itex]+[itex]\frac{c}{\sqrt{1+b+c}}[/itex][itex]\leq[/itex][itex]\frac{b}{\sqrt{1+b}}[/itex]+[itex]\frac{c}{\sqrt{1+c}}[/itex]

Therefore,
[itex]\frac{a}{\sqrt{1+a}}[/itex][itex]\leq[/itex][itex]\frac{b}{\sqrt{1+b}}[/itex]+[itex]\frac{c}{\sqrt{1+c}}[/itex]

It seems a bit dodgy :bugeye:
 
Your solution is good. You use both the idea about f'>0 and the analysis of the denomintarors in one argument.