20071224, 01:16  #1 
Nov 2005
B5_{16} Posts 
rational powers of negative one
Let's say I have y=f(x)=(1)^x as a function.
If x=...,1,0,1,... then y alternates between opposite signs. If x=...,1.5,0.5,2.5,... then y is the square root of 1 which is of course i. f(5.3)=f(1.3) * f(2)^2=f(1.3) (1)^1.3=(1) * 10th root of (1)^3 = (1)* 10th root of (1) = i I think that there could be rational numbers with more than one possible value since A) Those are just the primary roots in my example. B) Other ways to rewrite the fraction which might give different signs? I see that this is close to how e^xi = cis(x) = cos(x)+ i sin(x) works for some values. My reasoning for this statement is that the angle (aka argument) of y is the same as the value of x for right angles. Too bad this property doesn't hold for all values of x. (After scaling) My goal was to try to make a derivative of (1)^x for nonintegers. The closest I've gotten is a function that looks like this: (x ln 1) * (1)^x = xi pi * (1)^x Is this remotely correct? If so, there's a follow up topic I'd like to discuss involving derivatives. I need this derivative for another. BTW Rational powers meaning negative one taken to a rational power, not that the result is rational or even real. Last fiddled with by nibble4bits on 20071224 at 01:17 Reason: Forgot to mention scaling 
20071224, 15:51  #2 
(loop (#_fork))
Feb 2006
Cambridge, England
6,323 Posts 
The obvious function equal to (1)^x at integers is exp(pi * I * x), which does have the derivative you want.
1 has ten tenth roots, amongst which is I; the cube of any tenth root of 1 is another tenth root of 1, but that doesn't mean that they're either equal or equal to I. 
20071224, 23:04  #3 
Dec 2003
Hopefully Near M48
2·3·293 Posts 
Actually, what you have is a "multivalued function" (which is not technically a function). To actually get a function, you need to choose a branch cut.
where can be any fixed integer. Last fiddled with by jinydu on 20071224 at 23:05 
20071226, 05:21  #4 
Nov 2005
181 Posts 
jinydu: I think I mentioned that  the comment about primary roots. It means there's nonprimary roots such as 1 at the angles 2k*Pi/10 where k is an integer. Of course for k>9 or k<0, there are 'mirrors' that are equivilent. This is because k mod 10 is always in the set {0,1,2,3,4,5,6,7,8,9}.
I was just wondering if there's something that makes the derivative useless for my purposes. It seems OK for integers but it would be nice to find out why it doesn't work if it is wrong. For one thing, finding the derivative of a discontinous function... 
20071226, 09:47  #5  
Dec 2003
Hopefully Near M48
2·3·293 Posts 
Quote:
is definitely a continuous (in fact infinitely differentiable) function of x. So is for any fixed integer n. Last fiddled with by jinydu on 20071226 at 09:49 

20080108, 04:58  #6  
Nov 2005
181 Posts 
Quote:
Last fiddled with by nibble4bits on 20080108 at 04:58 

Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
a rational sequence  MattcAnderson  MattcAnderson  2  20161214 16:58 
rational integers  wildrabbitt  Math  17  20150615 08:47 
Rational if and only if Decimal Expansion Repeats  jinydu  Lounge  4  20081001 07:45 
need help with C++ code (nonnegative integers)  ixfd64  Programming  11  20080320 01:52 
Rational solution.  mfgoode  Homework Help  9  20070819 07:19 