1

If $a,b\in\mathbb{C}$ and $z$ is a non zero complex number.

The the rots of the equation $az^3+bz^2+\bar{b}z+\bar{a}=0$

lie on $x^2+y^2=1$. Where $z=x+iy$

Plan: I am taking $z=x+iy$

$\displaystyle a(x+iy)^3+b(x+iy)^2+\bar{b}(x+iy)+\bar{a}=0$

$a(x^3-iy^3+3ixy(x+iy))+b(x^2-y^2+2ixy)+\bar{b}(x+iy)+\bar{a}=0$

$\displaystyle a(x^3-3xy^2)+b(x^2-y^2)+\bar{b}x+\bar{a}=0$

and $-ay^3+3x^2y+2bxy+\bar{b}x=0$

how do i solve it help me please

jacky
  • 5,194

3 Answers3

2

I now prove the conjecture of Oscar given in his answer. Oscar's conjecture limits the claim of the original proposal by Jacky, and hence it corrects the original question.

The Conjecture of Oscar Lanzi:

If $a,b$ are complex numbers, the equation $$ az^3+bz^2+\bar{b}z+\bar{a}=0.....(1),$$ has at least one root which is uni-modular.

The Proof:

Let $t$ be one root of (1), then $$ at^3+bt^2+\bar{b} t+ \bar{a}=0....(2)$$ Complex conjugation of (2) leads to $$\bar{a}\bar{t}^3+\bar{b}\bar{t}^2+b \bar{t} +a=0....(3).$$ Multiplying Eq. (3) by $t^3$, we get $$\bar{a}(t\bar{t})^3+\bar{b} t({t \bar {t}})^2+ b t^2 (t \bar{t})+a t^3=0...(4). $$ This equation can be re-written as $$ \bar{a} |t|^6+\bar{b} t|t|^4+b t^2 |t|^2+ a t^3=0...(5).$$ Subtracting Eqs. (2) and (5), we get $$\bar{a} (|t|^6-1)+\bar{b} t (|t|^4-1)+b t^2 (|t|^2-1)=0....(6).$$ Remarkably, we get $(|t|^2-1)=0.$ Hence proved that at least one root of Eq. (1) is uni-modular: $(x^2+y^2)=1.$

Z Ahmed
  • 43,235
1

Put $a=1$ and $b=4$. You find roots on the real axis which are not $\pm1$; to wit, $z\in\{-1,(-3\pm\sqrt{5})/2\}$ and only $-1$ meets the claim. You'll have to put constraints on $a$ and $b$ if this is to work.

What always does work is, at least one root will lie on the unit circle. In the above example that would be $-1$.

Oscar Lanzi
  • 39,403
0

I prove that if $a,b$ are complex numbers roots the equation (1) are pairs of the type: $z$ and $1/\bar{z}$ which means all roots are uni-modular or at least one (un-paired one) is so ($x^2+y^2=1$).

Let $$az^3+bz^2+\bar{b}z+\bar{a}=0.....(1).$$ Complex conjugation of Eq. (1) gives $$\bar{a} \bar{z}^3+\bar{b}{\bar{z}^2}+b \bar{z}+a=0....(2).$$ Now if we change $\bar{z}=\frac{1}{z}$ in Eq. (2), we get $$\frac{\bar{a}}{z^3}+\frac{\bar{b}}{z^2}+\frac{b}{z}+\frac{a}{z}=0...(3).$$ Which (3) is nothing but Eq. (1). This proves that Eq. (1) has all roots sucj that $x^2+y^2=1$. In other words roots of Eq. (1), are pairs of the type $z$ and $1/\bar{z}$, which means either all roots are uni-modular or at least one (the unpaired one) is so.

For instance one may check that for $a=1-i$ and $b=i$,we get roots as $-(1/4 + i/4) (i + \sqrt{7}),~(1/4 + i/4) (-i + \sqrt{7}),~ -1$.

Z Ahmed
  • 43,235
  • 1
    It shows that, if $z$ is a root, then so is $1/\overline z$. I don't think you have shown they are the same root. – Empy2 May 28 '19 at 15:39
  • True. By this method $1+i$ and $(1+i)/2$ could be a pair of roots mapped into each other. – Oscar Lanzi May 28 '19 at 15:50
  • So each root either is on the unit circle or is part of a pair of roots $z, 1/z.$ And each root in the pair has the same multiplicity, so there are an even number of roots in pairs. Since the cubic has three roots (an odd number), at least one root is unpaired. – David K May 28 '19 at 18:21
  • So while it is not necessarily true that all roots are on the unit circle as claimed in the question, this proof can be modified to show that at least one root is. – David K May 28 '19 at 18:24
  • Thanks for the encouragement, I think I have proved it now, please see. – Z Ahmed May 29 '19 at 02:51