1
$\begingroup$

Is there any nice way to solve this equation without wolfram?

$\log_{\sqrt{x}}2+\log_6x^x=4$

Thanks.

$\endgroup$
  • 2
    $\begingroup$ Can you please share the source of the problem $\endgroup$ – lab bhattacharjee Aug 30 '14 at 17:06
  • $\begingroup$ @labbhattacharjee - it's a question i found in one of my old notebook of exersices no source known. $\endgroup$ – dave Aug 30 '14 at 17:08
  • $\begingroup$ Since wolfram alpha doesn't find an exact solution, I don't think there are any such solutions, and thus it probably can't be solved in a 'nice' way. $\endgroup$ – Ragnar Aug 30 '14 at 17:16
  • $\begingroup$ It's equivalent to, with $t=\log_2 x$, $$ \log_62 \cdot t^2 2^t-4t+2=0$$ Unlikely to yield except to numerical methods. $\endgroup$ – Macavity Aug 30 '14 at 17:16
  • $\begingroup$ @dave Then you probably copied it wrong, unless your question is meant to use the Lambert W-function, or if you only want an approximation. $\endgroup$ – Thomas Andrews Aug 30 '14 at 17:24
1
$\begingroup$

Rewriting $$\log_{\sqrt{x}}2+\log_6x^x=4$$ leads to the equation $$f(x)=x \log ^2(x)-4 \log (6) \log (x)+\log (4) \log (6)=0$$ which cannot be solved analytically ($f'(x)=0$ or $f''(x)=0$ cannot be solved either).

Plotting the function shows two roots, the first close to $x=1.5$, the second close to $x=4.0$. For the solution, Newton is the simplest root-finder. Strating from a "reasonable" guess $x_0$, it will update it according to $$x_{n+1}=x_n-\frac{f(x_n)}{f'(x_n)}$$

For the first root, using $x_0=1.5$, the successive iterates are $1.45386$, $1.45521$ which is the solution for six significant figures. Limiting to a single iteration, we should have obtained $$x \simeq \frac{6 \log \left(\frac{9}{8}\right) \log (6)-3 \log (384)}{3 \log ^2\left(\frac{3}{2}\right)-\log (147456)}$$

For the second root, using $x_0=4.0$, the successive iterates are $3.91886$, $3.91689$ which is the solution for six significant figures. Limiting to a single iteration, we should have obtained $$x \simeq \frac{\log \left(\frac{4096}{81}\right)+\log (6) \log (64)}{\log ^2(4)+\log \left(\frac{8}{3}\right)}$$ As you can see, all of this could be done using a pocket calculator.

$\endgroup$
  • $\begingroup$ How did you evaluate $\log_{\sqrt{x}}2$ ? Seems to me that there's missing a factor $2$ in your equations (when compared with the other answer). $\endgroup$ – Han de Bruijn Aug 31 '14 at 9:45
  • $\begingroup$ @HandeBruijn $$\log_{\sqrt{x}}a=\frac{\log (a)}{\log \left(\sqrt{x}\right)}=\frac{2 \log (a)}{\log (x)}$$ I double checked and I did not find anything wrong. Take care : some $\log(2)$ became $\log(4)$. Please tell me where I made mistakes. I really thank you for your attention. Cheers :-) $\endgroup$ – Claude Leibovici Aug 31 '14 at 14:10
  • $\begingroup$ OK, now I see there's nothing wrong indeed (that $\log(4)$ as you say). $\endgroup$ – Han de Bruijn Aug 31 '14 at 16:37
1
$\begingroup$

I am pretty sure the isn't solvable with Algebra II, unless you resort to the mess that is complex logs. However: $$ log_\sqrt{x}2+log_6x^x=4\Rightarrow log_\sqrt{x}6+x(log_26)(log_6x)=4(log_26)\Rightarrow \frac{2}{log_6x}+x(log_26)(log_6x)=4(log_26)\Rightarrow2+x(log_6x)^2-4(log_6x)(log_26)=0 $$ which is a quadratic in $log_6x$ except for the pesky x coefficient. Using the quadratic formula anyway, and evaluating the logs, $$ log_6x\approx\frac{10.34\pm\sqrt{106.91-8x}}{2x} $$ This is definitely solvable. However, The only way I can come up with a numerical solution is graphing the two sides.

$\endgroup$

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.