bfloat(sqrt(%i))?



Raymond Toy wrote:
> On 6/9/10 10:02 AM, Richard Fateman wrote:
>   
>> Raymond Toy wrote:
>>
>> Going back to the original question,  if we cannot choose the "positive"
>> s,  can we choose a "principal" root?   Sometimes, but you
>> do not have any assurance that this is the one of interest.
>>     
>
> I think principal root is nice.  If it's not the root of interest, you
> should do something else, like solve(x^2=%i,x).
>   

Sometimes you cannot identify the principal root, if there are symbols 
involved.
sqrt((x-y)^2)    has principal root x-y   if x>y.

>   
>> Here is another reason, or at least food for thought.
>> Mathematica 7.0   returns (-1)^(1/4).
>>
>> evaluating it numerically however gives 0.707107 + 0.707107 I.
>>     
>
> What exactly does this mean?  If (-1)^(1/4) is meant to be any (or all)
> of the 4 roots of -1,
Maybe.
>  then it seems that 0.7+0.7I is the wrong answer.
> But if it is the right answer, then (-1)^(1/4) really means the
> principal root.
>
> Unless numerical evaluation is meant to produce the principal root if
> there is ambiguity.
>   
If you leave expressions unevaluated until every symbol has a numeric 
value, I think you are OK.
Any intermediate "simplification" can be problematical.

> How is this meant to interact with say, solve(x^2+3*x+1,x) which returns
>
> [x = -(sqrt(5)-3)/2, x = (sqrt(5)+3)/2].
>
> Does that mean sqrt(5) is +/- 2.23 so there are 4 possible roots to this
> quadratic?  (I didn't actually check that this produces 4 values).
>   
Actually, it only produces 2. a different choice interchanges the roots.
That is partly why the solution to a quadratic  should be a pair of 
equations, and separating them is risky.
( unless they are numbers, when you are probably OK.)

rjf