I am trying to see if this inequality is correct or not:


Given that: $C_1\geq C_2$, $\hat{C_1}\geq \hat{C_2}$, $C_1\geq\hat{C_1}$,$C_2\geq\hat{C_2}$.

My attempt:

$2^{2C_1+2\hat{C_1}}+2^{2C_2+2\hat{C_1}}-2^{2\hat{C_1}}\geq 2^{2\hat{C_1}+2C_1}+2^{2\hat{C_2}+2C_1}-2^{2C_1}$

$2^{2C_2+2\hat{C_1}}-2^{2\hat{C_1}}\geq 2^{2\hat{C_2}+2C_1}-2^{2C_1}$

$2^{2C_1}+2^{2C_2+2\hat{C_1}}\geq 2^{2\hat{C_1}}+2^{2\hat{C_2}+2C_1}$

  • $\begingroup$ Get rid of the 2 in front of each C and call them a b c d. It will be a lot easier to work with. $\endgroup$ – marty cohen May 30 '18 at 11:37
  • $\begingroup$ @Lee Please recall that if the OP is solved you can evaluate to accept an answer among the given, more details here meta.stackexchange.com/questions/5234/… $\endgroup$ – gimusi Jun 22 '18 at 21:06

Consider your last line $$ 2^{2C_1}+2^{2C_2+2\hat{C_1}}\geq 2^{2\hat{C_1}}+2^{2\hat{C_2}+2C_1} $$ Select $C_1 = 5$, $\hat{C_1} = 2$, $C_2 = 2$,$\hat{C_2} = 1$.

This satisfies the given conditions: $C_1\geq C_2$, $\hat{C_1}\geq \hat{C_2}$, $C_1\geq\hat{C_1}$,$C_2\geq\hat{C_2}$.

Plugging this into the eq: $$ 2^{2\cdot 5}+2^{2\cdot 2+2 \cdot 2}\geq 2^{2\cdot 2}+2^{2\cdot 1+2 \cdot 5} $$ we see that this is violated since the numbers are $$ 2^{2\cdot 5}+2^{2\cdot 2+2 \cdot 2} = 1280\geq 2^{2\cdot 2}+2^{2\cdot 1+2 \cdot 5} = 4112 $$ So the inequality does not hold.

This didn't materialize out of thin air, here is some more explanation how I "made this up":

With some real $a$, multiply both sides of your last line with $2^{4 a}$ which doesn't change the inequality. You get $$ 2^{2(C_1+2a)}+2^{2(C_2+a)+2(\hat{C_1} +a)}\geq 2^{2(\hat{C_1}+2a)}+2^{2(\hat{C_2}+a)+2(C_1+a)} $$ Shifting all variables by $a$ also doesn't change the given conditions. So in the shifted variables, you have $$ 2^{2a +2C_1}+2^{2C_2+2\hat{C_1}}\geq 2^{2a +2\hat{C_1}}+2^{2\hat{C_2}+2C_1} $$ which gives that
$$ 2^{2a}\geq \frac{2^{2\hat{C_2}+2C_1} - 2^{2C_2+2\hat{C_1}}}{2^{2{C_1}} - 2^{2\hat{C_1}}} $$ is required. Now you can choose some values on the RHS and then compute an $a$ which actually violates this.

I chose $C_1 = 4$, $\hat{C_1} = 1$, $C_2 = 1$,$\hat{C_2} = 0$ - remember these are the shifted variables. Then $$ 2^{2a}\geq \frac{2^{8} - 2^{4}}{2^{8} - 2^{2}} \simeq 0.95 $$ or $a > -0.035$. So this can be violated by, say, $a = -1$. Then we have the original values $C_1 = 5$, $\hat{C_1} = 2$, $C_2 = 2$,$\hat{C_2} = 1$ which are used above.

  • $\begingroup$ Thank you @Andreas $\endgroup$ – Lee May 31 '18 at 6:00

The inequality is not true in general, indeed we have that

$$\frac{2^{2C_1} +2^{2C_2} -1}{2^{2C_1}}\geq\frac{2^{2\hat{C_1}}+2^{2\hat{C_2}}-1}{2^{2\hat{C_1}}}\iff 1 +4^{C_2-C_1}-4^{-C_1}\ge1 +4^{\hat C_2-\hat C_1}-4^{-\hat C_1}$$ $$\iff 4^{C_2-C_1}-4^{-C_1}\ge 4^{\hat C_2-\hat C_1}-4^{-\hat C_1}$$ $$\iff 4^{-C_1}(4^{C_2}-1)\ge 4^{-\hat C_1}(4^{\hat C_2}-1)$$

now let consider

$$C_2=\hat C_2=\hat C_1>0\implies 4^{-C_1} (4^{C_2}-1)\ge 4^{-C_2}(4^{C_2}-1)\implies 4^{-C_1}\ge 4^{-C_2}$$

which is not true since $4^{-x}$ is decreasing and $C_1\ge C_2$.

  • $\begingroup$ Thank you @gimusi $\endgroup$ – Lee May 31 '18 at 6:00
  • $\begingroup$ @Lee You are welcome! Bye $\endgroup$ – gimusi May 31 '18 at 6:02

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.