4
$\begingroup$

In physics, a resistor is an electrical component that inhibits current. We represent its ability to inhibit current with a value called its resistance, represented with $R$. ($R$ is always positive.)

There are two ways to combine two resistors to make another resistor: in parallel, or in series. If combined in series, $R_{\mathrm{new}} = R_1 + R_2$. If combined in parallel, $\frac1{R_{\mathrm{new}}} = \frac1{R_1} + \frac1{R_2}$.

Now the actual puzzle:

Given 4 different resistors, let $L$ be the largest resistance you can make with those resistors and let $S$ be the smallest resistance you can make with those resistors. Show that $L \geq 16S$.

$\endgroup$
2
  • 3
    $\begingroup$ I've edited your question to add an explanation of the "rules", since not everyone knows resistance formulas. $\endgroup$
    – Deusovi
    Commented Jul 22, 2016 at 12:57
  • 4
    $\begingroup$ Isn't that rather a math's question than a puzzle? Seems straight forward calculation is all that is needed... $\endgroup$
    – BmyGuest
    Commented Jul 22, 2016 at 14:48

4 Answers 4

12
$\begingroup$

If you put all four resistors in series, you get a resistance $R_s=R_1+R_2+R_3+R_4$. If you put all four in parallel, you get a resistance $R_p$ such that $\frac1{R_p}=\frac1{R_1}+\frac1{R_2}+\frac1{R_3}+\frac1{R_4}$.

Now $\frac{L}S\ge\frac{R_s}{R_p}=\left(R_1+R_2+R_3+R_4\right)\left(\frac1{R_1}+\frac1{R_2}+\frac1{R_3}+\frac1{R_4}\right)$.

By AM-GM, $$\left(R_1+R_2+R_3+R_4\right)\left(\frac1{R_1}+\frac1{R_2}+\frac1{R_3}+\frac1{R_4}\right)\\\ge\left(4\sqrt[4]{R_1R_2R_3R_4}\right)\left(4\frac1{\sqrt[4]{R_1R_2R_3R_4}}\right)=16$$ as desired. Equality holds iff $R_1=R_2=R_3=R_4$.

$\endgroup$
4
$\begingroup$

First consider the case of two resistors.

By the Arithmetic-Geometric Mean Inequality Theorem, given two positive reals $a, b$, $\sqrt{ab}\leqslant \frac{a+b}2$. Applying that theorem both to the resistances and to their reciprocals,

$\sqrt{ab}\leqslant \dfrac{a+b}2=\dfrac{R_s}2$

$\dfrac1{\sqrt{ab}}\leqslant \dfrac{1/a+1/b}2$

$\Rightarrow \sqrt{ab}\geqslant \dfrac2{1/a+1/b}=2R_p(a,b)$

$\Rightarrow 4R_p(a,b)\leqslant R_s$

where $R_p(a,\dots)$ denotes the resistance of resistors $a,\dots$ in parallel. Moving on now to four resistors, we have

$16R_p(a,b,c,d)\leqslant 4(R_p(a,b)+R_p(c,d))\leqslant a+b+c+d$


Proof of the Arithmetic-Geometric Mean Inequality Theorem for two reals $a, b$. If $a\leqslant b$, then

$a(b-a)\leqslant b(b-a)$

$\Rightarrow ab-a^2\leqslant b^2-ab$

$\Rightarrow 2ab\leqslant a^2+b^2$

$\Rightarrow 4ab\leqslant a^2+2ab+b^2=(a+b)^2$

$\Rightarrow ab\leqslant (\frac{a+b}2)^2$

$\Rightarrow \sqrt{ab}\leqslant \frac{a+b}2$

If $b\leqslant a$ the same result follows. Therefore it is true for all real $a, b$.

$\endgroup$
2
$\begingroup$

Actually there is no need to appeal to the AM-GM inequality.

Given that $L=R_1+R_2+R_3+R_4$

and that $\frac{1}{S}=\frac{1}{R_1}+\frac{1}{R_2}+\frac{1}{R_3}+\frac{1}{R_4}$,
we obtain
$$ \frac{L}{S}=(R_1+R_2+R_3+R_4)\left(\frac{1}{R_1}+\frac{1}{R_2}+\frac{1}{R_3}+\frac{1}{R_4}\right)=4+\frac{R_2+R_3+R_4}{R_1}+\frac{R_1+R_3+R_4}{R_2}+\frac{R_1+R_2+R_4}{R_3}+\frac{R_1+R_2+R_3}{R_4} $$ Then we use the function inequality valid for $x>0$ $$ f(x)=x+\frac{1}{x}= 2+(\sqrt{x}-\frac{1}{\sqrt{x}})^2\geq 2 $$ Apply this then to $x = \frac{R_1}{R_2}, \frac{R_1}{R_3},\dots$ in the expression above and one obtains $$ \frac{L}{S}\geq 16 $$

$\endgroup$
-1
$\begingroup$

Sorry, but the inequality seems wrong...

The allowed uses of the resistors are not made explicit in the -original-question.

If someone gives me four resistors, there is a resistance value I can make -with those resistors- that the other answers do not account for. By using only the resistor legs I can make a very low resistance.

This low resistance is independent of the resistor-values. So any inequality describing the situation should include the specific resistor values. Therefore it must be wrong.

Perhaps a trick question to separate engineers from theoreticians?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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