# Tetration Forum

Full Version: About the fake abs : f(x) = f(-x)
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
When discussing fake function theory we came across the fake sqrt.

the fake abs function is then fake_abs(x) = fake_sqrt(x^2).

Clearly when we want a real-entire function f(x) to satisfy f(x) = f(-x) then we consider f(fake_abs(x)).

But what if we have a real-entire function f(x) and we want to remove the property f(x) = f(-x).

We can do many things like : g(x) = f(x) + exp(-x). (*)

But lets consider the context of fake functions :

Find g such that
f(x) = g(fake_abs(x))

=>

g(x) = f(fake_abs^[-1](x))

g(x) = f( sqrt ( inv_fake_sqrt(x) ) )

Lets call inv_fake_sqrt(x) := fakesquare(x).

g(x) = f ( sqrt ( fakesquare(x) ) )

Now since f(x) = f(-x) we have that F(x) = f(sqrt(x)) is also an entire function.

g(x) = F( fakesquare(x) )

Now if we want g(x) to be entire then F ( fakesquare(x) ) needs to be entire.

Since fakesquare is a multivalued function ( an inverse of an entire ) its not entire.

SO when is F( fakesquare(x) ) entire ?

And how does that look like ?

Also of interest ( when its not entire ) :

fake ( F ( fakesquare(x) ) )

From (*) one then also wonders about

F( fakesquare(x) ) - f(x)

and how that looks like.

I have some ideas and guesses but no evidence or plots.

Seems like chapter 2 in fake function theory.

The analogue questions exist for exp(x) instead of x^2 ;

removing the periodic property.

regards

tommy1729
(12/11/2014, 01:22 PM)tommy1729 Wrote: [ -> ]When discussing fake function theory we came across the fake sqrt.
... fake_sqrt(x^2).

$f(x)=\sum_{n=0}^{\infty} a_n x^n\;\;\;\;\; a_n = \frac{1}{\Gamma(n+0.5)}$
for large positive numbers, $g(x)=f(x)\exp(-x) \approx \sqrt{x}$

$g(x^2) = g((-x)^2) \approx x\;\;$ this is true if |real(x)| is large enough, and |imag(x)| isn't too large
However, at the imaginary axis $g(x^2)$ grows large exponentially, and does not behave like x at all. And g(x^2) never behaves like abs(x), anywhere in the complex plane

Here are some example calculations:
$g(25)=5 + 1.5\cdot10^{-13}\;\;\;$ 5^2, small error term
$g((5+0.1i)^2) = 5+0.1i - k\cdot10^{-13}\;\;\;$also a small error term, but not abs(x^2)
$g(-25)=-866955233 \;\;\;$ (5i)^2, huge error term, nowhere near 5i
$g(25i) = 3.53768061172 + 3.52450328163i \;\;\;\;\sqrt{25i}\approx 3.535534 +3.535534i$
Im aware that there is no fake abs(z) that is a good approximation on all of C.

That is why I used the letter x.

So fake in the sense of asymptotic near the real line , or fake of the absolute value of the reals if you like.

The focus for fake functions is Always mainly on the real line anyway , as you well know.

Im aware of what you said sheldon.

You are completely correct.

But I see no problem with that.

Or did you not claim a problem ?

Are you conjecturing a growth rate perhaps ?

regards

tommy1729