Let $r>0$, solve this equation $$\cos{\left(\dfrac{\pi}{3}-\dfrac{\pi}{3r}\right)}=\sqrt{\dfrac{11}{r^2}-2}$$

I have found that $r=2$ is a solution, as $$LHS=\cos{\dfrac{\pi}{6}}=\dfrac{\sqrt{3}}{2}$$ and $$RHS=\sqrt{\dfrac{11}{4}-2}=\dfrac{\sqrt{3}}{2},$$ so $r=2$ is a root.

How can other solutions be found?

  • 4
    $\begingroup$ Plot these two functions and you'll see that there are exactly 2 real solutions. One (the one you found) is positive. The other is negative and I doubt there's an analytic form for it. $\endgroup$ – John Barber May 14 at 4:18

If you consider the function$$f(r)=\cos{\left(\dfrac{\pi}{3}-\dfrac{\pi}{3r}\right)}-\sqrt{\dfrac{11}{r^2}-2}$$ its domain is restrited to $-\sqrt{\frac{11}{2}} \leq r \leq \sqrt{\frac{11}{2}}$.

If you did what @John Barber commented, you must have noticed that the negative solution is close to the left bound where the function value is $\approx 0.077$; this means that the root is quite close to the bound (just above).

So, let $r=t-\sqrt{\frac{11}{2}} $ and expand the function a very truncated series around $t=0$; this should give $$f(t)=\cos \left(\frac{11+\sqrt{22}}{33} \pi \right)-2 \sqrt[4]{\frac{2}{11}} \sqrt{t}+O\left(t\right)$$ Ignoring the higher order terms, then $$t\sim \frac{1}{4} \sqrt{\frac{11}{2}} \cos ^2\left(\frac{11+\sqrt{22}}{33} \pi \right)\approx 0.00347582\implies r\approx -2.34173$$ while the exact solution is $-2.34180$.

Using the expansion up to $O\left(t^{3/2}\right)$ and solving the quadratic in $\sqrt t$ would give $t\approx 0.00341749$ which implies $r\approx -2.34179$.

I should precise here that, using Newton method without any step size control, we face serious problem since, at least for a time, many ieterates are complex numbers.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.