How do I find the maximum and minimum values of the following?

$$a^2\sin^2\theta + b^2\csc^2\theta$$ Is the max value $\infty$?

I tried to find the minimum value by using A.M$\geq$G.M. inequality (is there any other way?) and got it as 2|a||b| which seem to work for all the positive values when a>b (I checked it on Desmos by putting different values of a and b) but when a$\le$b it just don't match in the graph. So, what am I doing wrong here? And what should be the correct answer?

  • 2
    $\begingroup$ Think of it as $rx+sx^{-1}$ where $-1\le x\le1$, and $r,s$ are constants. $\endgroup$ – Gerry Myerson Aug 2 '19 at 7:17
  • 2
    $\begingroup$ @GerryMyerson You mean $0\le x\le1$. $\endgroup$ – J.G. Aug 2 '19 at 7:36
  • $\begingroup$ @J.G., yes, thanks. $\endgroup$ – Gerry Myerson Aug 2 '19 at 9:25

The function is periodic with period $\pi$ so let us consider $ 0\le x \le \pi.$

Case-1: When $a>b$ in $$f(x)=a^2\sin^2 x+ b^2 \csc^2 x~~~(1)$$ then $$f'(x)=2a^2 \sin x \cos x-2b^2 \cot x \csc^2 x.= 2 \cos x(a^2 \sin^4 -b^2)/\sin^2 x ~~~(2)$$ $f'(x)=0,$ gives $\cos x=0$ or $\sin^4 x = b^2/a^2 \Rightarrow \sin^2 x= b/a \Rightarrow \sin x = \pm \sqrt{b/a} \Rightarrow x_1 = \pi/2, ~ \mbox{or} ~ x_{2,3}=\pm \sin^{-1} \sqrt{b/a}$. Next $$f''(x)= b^2(4 \cot^2x \csc^2 x + 2 \csc^2 x)+2a^2(\cos^2 x- \sin^2 x)~~~~~~(3).$$ Get $f''(x_1)=-2(a^2-b^2)< 0,$ so local max. at $x=\pi/2$. Use $\sin^2 x=b/a, \cos^2 x=(a-b)/a$ etc. in (3) to get $f''(x_{2,3})=8a(a-b)>0,$ so two local minima at $x=x_{2,3}.$

Hence when $a>b$, the local max. and min. are given as $$f_{max}=f(\pi/2)=a^2+b^2, f_{min}f(x_{2,3})=2ab.$$

Case-2: When $a<b$, then $x_{2,3}$ do not exist and $f''(\pi/2)=-2(a^2-b^2)>0,$ so there exists only one local min. and $f_{min}=f(\pi/2)=a^2+b^2$

In both the cases the function $f(x)$ is positive and unbounded which can becomes $\infty$ at $x=0, \pi.$

  • $\begingroup$ Thanks, I guessed the correct answer myself which matches with yours but could you give a much easier proof of it or explain it a bit? I don't seem to get it after the "cos x = 0" line. $\endgroup$ – Aditya Kshitiz Aug 2 '19 at 11:40
  • $\begingroup$ @ Aditya Kshitz I have added more stuff now you can follow it. $\endgroup$ – Dr Zafar Ahmed DSc Aug 2 '19 at 12:17
  • $\begingroup$ Okay now it makes sense but shouldn't f max be infinity other than a^2 +b^2 when a>b? And why doesn't A.M≥G.M. inequality only give the result with a>b and not b>a. (i.e. only 2ab not a^2 + b^2) $\endgroup$ – Aditya Kshitiz Aug 2 '19 at 15:17
  • $\begingroup$ @Aditya Kshitz Please note that $infunuty$ can never be answer usually. Your function has local max and two local minima. These minima are also global . In second case there is no local max. Mentioning that the function can become infinitely large is correct rather than saying max is $\infty$Like in the equation $1/x=0$ one does not sat that $x=\infty$ is a root. $\endgroup$ – Dr Zafar Ahmed DSc Aug 3 '19 at 3:00

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.