I'm studying for actuarial exams, but I always pick up mathematics books because I like to challenge myself and try to learn new branches. Recently I've bought Topology by D. Kahn and am finding it difficult. Here is a problem that I think I'm am answering sufficiently but any help would be great if I am off.

If $d$ is a metric on a set $S$, show that $$d_1(x,y)=\frac{d(x,y)}{1+d(x,y)}$$ is a metric on $S$.

The conditions for being a metric are $d(X,Y)\ge{0}, d(X,Y)=0$ iff $X=Y$, $d(X,Y)=d(Y,X)$, and $d(X,Y)\le{d(X,Z)+d(Z,Y)}$. Thus, we simply go axiom by axiom.

1) Since both $d(x,y)\ge{0}$ and $1+d(x,y)\ge{0},$ it is clear that $d_1(x,y)\ge{0}$. (Is this a sufficient analysis?)
2) $d_1(x,x)=\frac{d(x,x)}{1+d(x,x)}=\frac{0}{1+0}=0$.
3) $d_1(x,y)=\frac{d(x,y)}{1+d(x,y)}=\frac{d(y,x)}{1+d(y,x)}=d_1(y,x).$
4) $d_1(x,y)=\frac{d(x,y)}{1+d(x,y)}\le{\frac{d(x,z)+d(z,y)}{1+d(x,z)+d(z,y)}}=\frac{d(x,z)}{1+d(x,z)+d(z,y)}+\frac{d(z,y)}{1+d(x,z)+d(z,y)}\lt\frac{d(x,z)}{1+d(x,z)}+\frac{d(z,y)}{1+d(z,y)}=d_1(x,z)+d_1(z,y).$

However, #4 is strictly less, not less than or equal to, according to my analysis, so where did I go wrong?

  • 1
    $\begingroup$ You meant to ask "show that... also is a metric distance ", right? $\endgroup$ – DonAntonio Apr 9 '13 at 2:31
  • 3
    $\begingroup$ If you prove that $a<b$, then you’ve certainly proved that $a\le b$. However, your calculation in (4) is incorrect: how do you justify the claim that $$\frac{d(x,y)}{1+d(x,y)}\le\frac{d(x,z)+d(z,y)}{1+d(x,z)+d(z,y)}\;?$$ The numerator on the right is at least as big as the numerator on the left, but the denominators aren’t the same. $\endgroup$ – Brian M. Scott Apr 9 '13 at 2:32
  • $\begingroup$ It looks correct: $\frac{a}{1+a} \leq \frac{b}{1+b}$ whenever $0 \leq a \leq b$. $\endgroup$ – Lord Soth Apr 9 '13 at 2:34
  • $\begingroup$ @LordSoth: Yes, but it still ought to be justified; the justification isn’t hard, but it isn’t one of the obvious inequalities that everyone recognizes right off the bat. $\endgroup$ – Brian M. Scott Apr 9 '13 at 2:35
  • 3
    $\begingroup$ There is a little slip in 2), the conditon says $d(x,y)=0$ iff (if and only if) $x=y$. You verified only one direction. $\endgroup$ – André Nicolas Apr 9 '13 at 2:43

There is someing wrong in 4), just as Brian comments. Here I offered a proof for you:

Proof: Notice that $f(x)=\frac{x} {1+x}$ is increasing on $\mathbb R^+$: to see this, let $g(x)=\frac{1}{x+1}$. It is easily to see that $g(x)$ is descreasing on $\mathbb R^+$. And note that $f(x)+g(x)=1$. Therefore, $f(x)$ is a increasing function on $\mathbb R^+$.

Since $d(x,y) \le d(x,z)+d(z,y)$, we have $d_1(x,y)=\frac{d(x,y)}{1+d(x,y)}\le{\frac{d(x,z)+d(z,y)}{1+d(x,z)+d(z,y)}}=\frac{d(x,z)}{1+d(x,z)+d(z,y)}+\frac{d(z,y)}{1+d(x,z)+d(z,y)}\lt\frac{d(x,z)}{1+d(x,z)}+\frac{d(z,y)}{1+d(z,y)}=d_1(x,z)+d_1(z,y).$

Hope this be helpful for you.

ADDed: $d(z,y)$ and $d(x,z)$ could be zero.

  • $\begingroup$ how does that show $\frac{d(x,y)}{1+d(x,y)}\le{\frac{d(x,z)+d(z,y)}{1+d(x,z)+d(y,z)}}$? It looked trivial when I did it originally, so i offered no justification? Is it because if $a<b$ then $\frac{1}{1+a}>\frac{1}{1+b}$? Which, i assume by your proof, implies that $\frac{a}{1+a}<\frac{b}{1+b}$? $\endgroup$ – Eleven-Eleven Apr 9 '13 at 3:51
  • $\begingroup$ @CHristopher: Since $f(x)$ is a increasing function, just as i pointed out above, then $f(d(x,y))\le f(d(x,z)+d(z,y))$. $\endgroup$ – Paul Apr 9 '13 at 3:55
  • $\begingroup$ @ChristopherErnst: You are welcome. $\endgroup$ – Paul Apr 9 '13 at 4:02

Yeah, you forgot that z=y. Remember, no where is it stated that each of these points are different. This will give the in equality that you are looking for.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.