4

The problem is: How many zeros of the polynomial $$ f(z)=z^4+3z^2+z+1 $$ lie in the right half-plane?

To solve this, we use the argument principle, $$ \text{number of zeros}=\frac{1}{2\pi i}\int_{\partial\Omega}\frac{f'(z)}{f(z)}dz. $$ Here $$ \partial\Omega=\{z=iy, -R\leq y\leq R\}\cup\{ z=Re^{i\theta},-\frac{\pi}{2}\leq \theta\leq\frac{\pi}{2}\}. $$ It claims that $$ \int_{-iR}^{iR}\frac{f'(z)}{f(z)}dz=0. $$ But I don't know why above equality holds.

Q-Y
  • 1,589

1 Answers1

2

We are going to prove that there is exactly one zero in the first quadrant. Since the polynomial has real coefficients it will have another zero in the fourth quadrant and therefore two zeros in the right half-plane. Let be $$ \gamma _1 \left( t \right):\left\{ \begin{gathered} x(t) = t \hfill \\ y(t) = 0 \hfill \\ \end{gathered} \right.\,\,\,\,\,\,\,0 \leqslant t \leqslant R $$ Let be $C=\gamma_1+\gamma_2-\gamma_3$. $$ \gamma _2 \left( t \right):\left\{ \begin{gathered} x(t) = R\cos (t) \hfill \\ y(t) = R\sin (t) \hfill \\ \end{gathered} \right.\,\,\,\,\,\,\,0 \leqslant t \leqslant \frac{\pi } {2} $$ Let be $$ \gamma _3 \left( t \right):\left\{ \begin{gathered} x(t) = 0 \hfill \\ y(t) = it \hfill \\ \end{gathered} \right.\,\,\,\,\,\,\,0 \leqslant t \leqslant R $$ We notice that there are no zeros on $\gamma_1$ since the coefficients are positive. Moreover we have that there are no zeros on $\gamma_3$ also. Indeed, we have that $$ f\left( {it} \right) = t^4 - 3t^2 + 1 + it $$ which, of course is never zero for any $t\in \mathbb R$. let be $\gamma=\gamma_1+\gamma_2-\gamma_3$. Now, by argument principle, we have that $$ N = \frac{1} {{2\pi i}}\int\limits_C {\frac{{f'(z)}} {{f(z)}}} dz = \frac{{\Delta _\gamma \left( {\arg \left( {f(z)} \right)} \right)}} {{2\pi }} $$ where $N$ is the number of zeros inside $\gamma$. We have that $$ \frac{{\Delta _\gamma \left( {\arg \left( {f(z)} \right)} \right)}} {{2\pi }} = \frac{{\Delta _{\gamma _1 } \left( {\arg \left( {f(z)} \right)} \right)}} {{2\pi }} + \frac{{\Delta _{\gamma _2 } \left( {\arg \left( {f(z)} \right)} \right)}} {{2\pi }} - \frac{{\Delta _{\gamma _3 } \left( {\arg \left( {f(z)} \right)} \right)}} {{2\pi }} $$ Now, $$ \frac{{\Delta _{\gamma _1 } \left( {\arg \left( {f(z)} \right)} \right)}} {{2\pi }} = 0 $$ because $f(x)>0$ on $\gamma_1$. We notice that we can write $\gamma_2=Re^{it}$ with $0 \leq t \leq \pi/2$ and therefore we have that $$ \begin{gathered} \frac{{\Delta _{\gamma _2 } \left( {\arg \left( {f(z)} \right)} \right)}} {{2\pi }} = \Delta _{\gamma _2 } \left( {\arg \left[ {R^4 e^{4it} \left( {1 + \frac{{3R^2 e^{2it} + \operatorname{Re} ^{it} + 1}} {{R^4 e^{4it} }}} \right)} \right]} \right) = \hfill \\ = \frac{{4\left( {\frac{\pi } {2}} \right) + o\left( {\frac{1} {R}} \right)}} {{2\pi }}\,,\,\,\,\,\,\,\left( {R \to + \infty } \right) = \hfill \\ \hfill \\ = 1 + o\left( {\frac{1} {R}} \right),\,\,\,\,\,\left( {R \to + \infty } \right) \hfill \\ \hfill \\ \end{gathered} $$ Finally, $$ \begin{gathered} \frac{{\Delta _{\gamma _3 } \left( {\arg \left( {f(z)} \right)} \right)}} {{2\pi }} = \frac{{\arg \left( {f(iR)} \right) - \arg \left( {f(0)} \right)}} {{2\pi }} = \hfill \\ \hfill \\ = \frac{{\arg \left( {R^4 - 3R^2 + 1 + iR} \right)}} {{2\pi }} = \hfill \\ \hfill \\ = \frac{1} {{2\pi }}\arg \left[ {R^4 \left( {1 - \frac{{3R^2 - 1 - iR}} {{R^4 }}} \right)} \right] = 0 + o\left( {\frac{1} {R}} \right)\,,\,\,\,\,\,\,\left( {R \to + \infty } \right) \hfill \\ \end{gathered} $$ Therefore we have that $N=1$. Thus there are two zeros in the right half-plane.

  • why does the net change of arg(f) on $\gamma_3$ is just $arg(())−arg((0))$?Could it be that $arg(())−arg((0))=2n\pi$ – Ariel So Dec 23 '19 at 19:09
  • In my notation arg stand for its principal value. – Luca Goldoni Ph.D. Dec 24 '19 at 07:27
  • My point is that we shouldn't only take a look at the principle value change between the beginning point and the endpoint.The case could be that along the imaginary axis, the f(z) might go around the origin and then go back to the real axis again. If you only consider the beginning point and endpoint, it definitely shows no change. – Ariel So Dec 24 '19 at 18:39
  • In general, You are right but since along $\gamma_3$ we have that $f(z)$ it is $Im(f(z))>0$ (as long as R>0) it follows that $f(z)$ never turn around the origin. – Luca Goldoni Ph.D. Dec 25 '19 at 09:45
  • 1
    Thank you. I think now the answer is complete. – Ariel So Dec 25 '19 at 17:34
  • Ty for this detailed answer. Can you please explicit what you mean by $\Delta_{\gamma}$ ? – Athena Jan 10 '23 at 13:11