micromass said:
Well, how would you solve [itex]x^2=2+i[/itex]?? By saying [itex]x=\sqrt{2+i}[/itex]?? How helpful...
The square root symbol has nothing to do with finding roots of polynomials. I can easily find the roots of [itex]x^n=-1[/itex]:
[tex]\cos(\pi k /n)+i\sin(\pi k /n),~k\in \{0,...,n-1\}[/tex]
but by your method, you would only find [itex]x=\sqrt[n]{x}[/itex] (even if we would allow the square root on complex numbers). That's not really what we want, is it??
A Direct Application of the Square-Root Operator
[ By using the Binomial Theorem]
[tex]{(}{2}{+}{i}{)}^{1/2}[/tex] ------------ [Expression 1]
[tex]{=}{2}^{1/2}{(}{1}{+}{i/2}{)}^{1/2}[/tex]
[tex]{=}{2}^{1/2}{(}{1}{+}{\Sigma}{(}{-1}{)}^{r-1}\frac{{(}{2r-2}{)}{!}}{{2}^{3r-1}{r}{!}{(}{r-1}{)}{!}}{i}^{r}{)}[/tex] ----------- [Expression 2]
[In the above expression r=1,2,3,4... to infinity]
=[1+1/32-5/2048+21/65536-429/8388608 ……………]+i [1/4-1/128+7/8192-33/262144+...]
=1.4553 + 0.3435 i
The convergence of the two series may be established by considering[for each series] the absolute values of the terms and then by applying D’Alembert’s ratio test or some other suitable test of convergence. The general term given in the third step(relation(2)) of the calculation proves useful in this respect. [One may consider the ratio between alternate terms[their absolute values] in the summation in Expression 2 [tex]{\mid}\frac{{u}_{r+2}}{{u}_{r}}{\mid}[/tex] as r tends to infinity]
Suppose we are to find:
[tex]{{(}2+i{)}}^{1/5}[/tex]
We find the first root by using the binomial theorem . Let the root be a+ib
[tex]{y}^{5}{=}{2+i}[/tex]
[tex]{y}^{4}{=}\frac{2+i}{a+ib}[/tex]
After rationalizing the RHS we take the fourth root by applying the binomial theorem again.
The process is repeated---- we do not get more than five roots by this method.
De Moivre’s method is much more convenient for these evaluations. But the application of the square root and other roots [their validity]can be seen from the above calculations.