If $1<x, y< 1$ then $-1<x*y<1$ and $0<1+x*y <2$, but then I can't prove that $-1<x\circ y =(x+y)/(1+x*y)<1$. Help please.
-
I am not sure what you are not able to prove. The thing you have written is the definition, so there is nothing to prove there. – Tobias Kildetoft Nov 07 '16 at 11:18
-
I can't prove that $-1< (x + y)/(1+x*y)<1$ – Bona Nov 07 '16 at 12:55
2 Answers
Knowing formula: $tanh(a + b) =\dfrac{tanh(a) + tanh(b)}{1 + tanh(a) . tanh(b)}$.
written under the form
$$a+b=tanh^{-1} \left(\dfrac{tanh(a) + tanh(b)}{1 + tanh(a) . tanh(b)}\right)$$
and setting $A=tanh(a)$ and $B=tanh(b)$, you get:
$$tanh^{-1}(A)+tanh^{-1}(B)=tanh^{-1}(A\circ B)$$
Thus you have proved that $((-1,1),\circ)$ is a group law by a mere transport of the additive law on $(\mathbb{R},+)$ by bijective function $tanh^{-1}$.
Have a look at (https://en.wikipedia.org/wiki/Formal_group) which provides different "transport" examples, "$\circ$" operation being one of them.

- 81,803
-
See also section $6$ in Sam Northshield's paper on Associativity of the Secant Method. – Bill Dubuque Nov 10 '16 at 04:18
-
-
You're quite welcome. And +1 for presenting a conceptual viewpoint. Btw, in English we often say transport the (group) structure. – Bill Dubuque Nov 10 '16 at 04:35
-
@Bill Dubuque Thanks. Yes, I think that it is even one of our main roles as "seniors", to bring, wherever we can, this conceptual viewpoint. Besides, I have made the correction transfer->tranport in the text ; I appreciate your remark; being French speaking, it is sometimes tricky because although a good part of the abstract vacabulary is very similar in both languages, there are quite subtile differences... – Jean Marie Nov 10 '16 at 05:12
For $|x|<1,|y|<1$, to prove $$ \left|\frac{x+y}{1+xy}\right|<1\tag1 $$ Just prove $\:\:|x+y|<|1+xy|$. Square both side, we have $$ x^2+y^2+2|xy|<1+2|xy|+x^2y^2 $$ Rearrage it we get $$ (1-x^2)(1-y^2)>0 $$ This is true for $|x|<1,|y|<1$. So we conclude that $(1)$ be true.
To prove $(G, \circ)$ is group, note that $0$ is identity and $-x$ is inverse. I assume that you know how to prove for associativity and rest.

- 16,805