solve and nth roots of ( - 1 )



Edwin Woollett wrote:
> On  May 19, 2009, on thread "ratsimp bad,  rectform good",
>   Edwin Woollett wrote:
>     " Maxima's solve function can return a solution in terms of z = 
>         (-1)^(1/n), which  is any (in general complex) number z   
>         such that z^n = -1. "
>
> On May 20, 2009, on the same thread, Richard Fateman wrote:
>   "  Actually, I think the expression (-1)^(1/6) in Maxima is not "any" 
>          number z such that z^6=-1.  It is a particular one.
>         Just as 3^(1/2) is  only the positive value.  This is a notational 
>         issue, and Maxima is not, generally, able to deal with
>          multiple values."
>
> Let's look at the 4th roots of ( - 1 ) as an example.
>
>  solve ( a^4 + 1) returns a list of four answers, each
>   with a  factor  ( - 1 )^(1/4).  Since the latter factor
>  can be four possible complex numbers, one might
>  think that maxima is returning more than four  answers.
>   

You might think this, but of course you would be wrong, since a 
polynomial of degree 4 has exactly 4 roots.
There is a fundamental challenge to the correctness of the result of 
solve if you separate the 4 answers and treat them as though each one 
can be further bifurcated into 4 separate results regardless of the 
other 3 answers that come from solve.

> For some reason, solve offers the user the choice of
> replacing  ( - 1 )^(1/4)  (using ratsubst, say) throughout 
> the list with a single choice and that will result in a list of
> four complex numbers.
>   
That seems like an OK thing to do, to me.
> For any user choice, you get the same list of four
> complex numbers, just in a different (cyclic) order.
>   
Exactly why it is OK to do it.

 I am not sure I understand what you are doing in the rest of the message. 

It is not adequate to determine that you have 4 distinct roots of a 
quartic by taking each of the roots, inserting it into the quartic, and 
simplifying to zero.
All that tells you is that you have 4 expressions, each of which might 
be equivalent to some root, but maybe not the 4 distinct ones.

  How could this be?  It is well known that 1/sqrt(2) and 2/sqrt(2) are 
equal, but not identical.  If you posed them as the two roots of the 
equation x^2-2=0,  what would your exploration say?

Let me try to be clear:  If you use a single expression that includes 
(-1)^(1/4),  you are asking for trouble.  If solve produces an 
expression that includes (-1)^(1/4), it probably also includes 3 more 
expressions with that form.  That's sort of OK, if you keep the 4 
expressions together as a set of solutions.

Or you can pick a particular solution (in the complex plane) and then 
instead of having the 4 solutions that circulate around, you have fixed 
them all in place. But then you should remove (-1)^(1/4) in all the 
roots in the solution set.