Proving $a^b+b^a > 1$ for Positive Real $a,b$

  • Context: MHB 
  • Thread starter Thread starter anemone
  • Start date Start date
  • Tags Tags
    Positive
Click For Summary

Discussion Overview

The discussion centers around the inequality \( a^b + b^a > 1 \) for positive real numbers \( a \) and \( b \). Participants explore various approaches to proving this inequality, including specific cases and general reasoning.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested
  • Mathematical reasoning

Main Points Raised

  • Some participants assert that the inequality \( a^b + b^a > 1 \) holds for positive real numbers \( a \) and \( b \), suggesting various methods of proof.
  • Others challenge the validity of specific arguments presented, questioning the correctness of certain equalities and assumptions made in the proofs.
  • One participant proposes proving the inequality \( x^x > 1 - \sqrt{x} \) for \( 0 < x < 1 \) as a step towards establishing the original inequality.
  • There are claims that if either \( a \) or \( b \) is greater than 1, the inequality is trivially satisfied.
  • Some participants discuss the behavior of \( a^b \) and \( b^a \) when \( a \) and \( b \) are both less than 1, noting that \( a^b \) increases with \( a \) and decreases with \( b \).
  • Concerns are raised about the implications of certain values approaching zero, with discussions on the limits of \( a^b + b^a \) as \( a \) or \( b \) approaches zero.
  • Participants express differing views on the sufficiency of certain proofs, with some asserting that more rigorous justification is needed.

Areas of Agreement / Disagreement

There is no consensus on the validity of the various arguments presented. Multiple competing views remain, with some participants supporting specific proofs while others challenge their correctness.

Contextual Notes

Some arguments rely on specific assumptions about the values of \( a \) and \( b \), and there are unresolved questions regarding the implications of certain limits and the correctness of derived equalities.

anemone
Gold Member
MHB
POTW Director
Messages
3,851
Reaction score
115
If $a$ and $b$ are positive real numbers, show that $a^b+b^a>1$.
 
Mathematics news on Phys.org
anemone said:
If $a$ and $b$ are positive real numbers, show that $a^b+b^a>1$.
$a^b>0,b^a>0 $ are the roots of the following equation :
$(x-a^b)(x-b^a)=0$
or $x^2-x(a^b+b^a)+a^bb^a=0$
let x=1, we have:
$1-(a^b+b^a)+a^bb^a=0$
$\therefore 1+a^bb^a=(a^b+b^a), \,\, and \,\, the\,\, proof \,\,is \,\, done$
 
I don't think that is a correct argument.

For example, $2$ and $3$ are roots of the equation $(x - 2)(x - 3) = x^2 - 5x + 6$ but if one sets $x = 1$ then $1 - 5 + 6 = 0$, a false equality.
 
mathbalarka said:
I don't think that is a correct argument.

For example, $2$ and $3$ are roots of the equation $(x - 2)(x - 3) = x^2 - 5x + 6$ but if one sets $x = 1$ then $1 - 5 + 6 = 0$, a false equality.
note the values of :$a^b\, and \,\,\, b^a$ are variables ,not fixed
 
I though that didn't matter? Your equality was

$$a^b + b^a = 1 + a^b b^a$$

Which is clearly false if, say, $a = 3$ and $b = 4$.
 
mathbalarka said:
I though that didn't matter? Your equality was

$$a^b + b^a = 1 + a^b b^a$$

Which is clearly false if, say, $a = 3$ and $b = 4$.
in fact if a=0 or b=0 then :
$a^b+b^a= 1$
if we want to find $a^b+b^a>100$
a,b also can be found
now for all value of a>0 ,and b>0
we want to find the value of k with
$a^b+b^a>k$
if $a--->0^+ ,or\,\, b--->0^+$
we may cnclude:$a^b+b^a>1$
 
Last edited:
let $g(t)=t^2-(a^b+b^a)t+(a^bb^a)$
here $ a>0 ,\, b>0$
$g(1)=1-(a^b+b^a)+a^bb^a$
now let $g(1)=0$
we have:$1-(a^b+b^a)+a^bb^a=0---(*)$
this time we will find $a,\,and \,\,b$ satisfying (*)
this can be considered as
a linear equation :$1-y+x=0$
here $y=a^b+b^a ,\,\, x=a^bb^a$
so $y=1+x$
$\therefore y=a^b+b^a=1+a^bb^a>1$
and the proof is finished
 
Last edited:
Albert said:
let $g(t)=t^2-(a^b+b^a)t+(a^bb^a)$
here $ a>0 ,\, b>0$
$g(1)=1-(a^b+b^a)+a^bb^a$
now let $g(1)=0$
we have:$1-(a^b+b^a)+a^bb^a=0---(*)$
this time we will find $a,\,and \,\,b$ satisfying (*)
this can be considered as
a linear equation :$1-y+x=0$
here $y=a^b+b^a ,\,\, x=a^bb^a$
so $y=1+x$
$\therefore y=a^b+b^a=1+a^bb^a>1$
and the proof is finished

not correct because g(1) = 0 is wrong pressumption
 
kaliprasad said:
not correct because g(1) = 0 is wrong pressumption
are you sure we can not find solutions
$(a^b,b^a)$ satisfying :
$1-(a^b+b^a)+a^bb^a=0\,\, ?$
 
  • #10
anemone said:
If $a$ and $b$ are positive real numbers, show that $a^b+b^a>1$.
[sp]First, I want to prove the inequality $x^x > 1-\sqrt x$, for $0<x<1.$ Taking logs, this is equivalent to $x\ln x > \ln(1-\sqrt x).$

Let $f(x) = x\ln x - \ln(1-\sqrt x).$ Then $f(x)\to0$ as $x\searrow0$. So to prove the inequality it will be enough to show that $f'(x)>0$ for $0<x<1.$ But $f'(x) = 1+\ln x + \dfrac1{2\sqrt x(1-\sqrt x)} > 1+\ln x + \dfrac1{2\sqrt x}$, because $\frac 1{1 - \sqrt x} > 1.$ So it will be enough to show that $g(x)>0$, where $g(x) = 1+\ln x + \dfrac1{2\sqrt x}$.

Differentiating again, $$g'(x) = \frac1x - \frac1{4x^{3/2}},$$ which has a single zero when $x = \frac1{16}$. This turning point is a local minimum (as you can see by differentiating again), and $g\bigl(\frac1{16}\bigr) = 3-4\ln2 \approx 0.2274 > 0$. It follows that $g(x)>0$ throughout the interval $(0,1)$, as required. Working backwards, this means that the inequality $x^x > 1-\sqrt x\;\; (0<x<1)$ is proved.

Now we can look at the inequality $a^b + b^a > 1.$ It is obviously satisfied if either $a$ or $b$ is greater than $1$, so we may assume that they are both less than $1$. Next, notice that when $a,b \in (0,1)$, $a^b$ increases when $a$ increases, and it decreases when $b$ increases.

Suppose that $b < \frac12.$ Then $b^a > a^a > 1-a^{1/ 2}$ (by the above inequality). But $a^{1/ 2} < a^b$, and so $1-a^{1/ 2} > 1-a^b.$ Thus $b^a > 1-a^b$, so that the inequality $a^b+b^a >1$ is proved.

The same argument shows that the inequality holds when $a < \frac12.$ So we only have to look at the case when both $a$ and $b$ lie between $\frac12$ and $1$. But then $a^b+b^a > a+b >1.$ That completes the proof.[/sp]
 
  • #11
if a or b are above 1 we are done,.so let us assume both < 1
by Bernoulli Inequality -- from Wolfram MathWorld we have$(1+x)^n \ge 1+nx$so $(1+\dfrac{a}{b})^{\dfrac{1}{a}} \ge 1+\dfrac{1}{b}$so $(1+\dfrac{a}{b})^{\dfrac{1}{a}} \gt \dfrac{1}{b}$or $1+\dfrac{a}{b} \gt (\dfrac{1}{b})^a$or $\dfrac{a+b}{b} \gt (\dfrac{1}{b})^a$or $\dfrac{b}{a+b} \lt b^a$similarly

$\dfrac{a}{a+b} \lt a^b$
Adding, we get $1 \lt a^b + b^a$

or $a^b+b^a \gt 1$
 
Last edited:
  • #12
kaliprasad said:
if a or b are above 1 we are done,.so let us assume both < 1
by Bernoulli Inequality -- from Wolfram MathWorld we have$(1+x)^n \ge 1+nx$so $(1+\dfrac{a}{b})^{\dfrac{1}{a}} \ge 1+\dfrac{1}{b}$so $(1+\dfrac{a}{b})^{\dfrac{1}{a}} \gt \dfrac{1}{b}$or $1+\dfrac{a}{b} \gt (\dfrac{1}{b})^a$or $\dfrac{a+b}{b} \gt (\dfrac{1}{b})^a$or $\dfrac{b}{a+b} \gt b^a$similarly

$\dfrac{a}{a+b} \gt a^b$
Adding, we get $1 \lt a^b + b^a$

or $a^b+b^a \gt 1$
Nice proof!
 
  • #13
kaliprasad said:
or $\dfrac{a+b}{b} \gt (\dfrac{1}{b})^a$

or $\dfrac{b}{a+b} \gt b^a$

How is that?
 
  • #14
I like Serena said:
How is that?
Just a typo – a couple of the inequality signs are the wrong way round.
 
  • #15
I like Serena said:
How is that?

I am sorry . There were typo Thanks for pointing. I have corrected the same.
 
Last edited:
  • #16
Opalg said:
[sp]First, I want to prove the inequality $x^x > 1-\sqrt x$, for $0<x<1.$ Taking logs, this is equivalent to $x\ln x > \ln(1-\sqrt x).$

Let $f(x) = x\ln x - \ln(1-\sqrt x).$ Then $f(x)\to0$ as $x\searrow0$. So to prove the inequality it will be enough to show that $f'(x)>0$ for $0<x<1.$ But $f'(x) = 1+\ln x + \dfrac1{2\sqrt x(1-\sqrt x)} > 1+\ln x + \dfrac1{2\sqrt x}$, because $\frac 1{1 - \sqrt x} > 1.$ So it will be enough to show that $g(x)>0$, where $g(x) = 1+\ln x + \dfrac1{2\sqrt x}$.

Differentiating again, $$g'(x) = \frac1x - \frac1{4x^{3/2}},$$ which has a single zero when $x = \frac1{16}$. This turning point is a local minimum (as you can see by differentiating again), and $g\bigl(\frac1{16}\bigr) = 3-4\ln2 \approx 0.2274 > 0$. It follows that $g(x)>0$ throughout the interval $(0,1)$, as required. Working backwards, this means that the inequality $x^x > 1-\sqrt x\;\; (0<x<1)$ is proved.

Now we can look at the inequality $a^b + b^a > 1.$ It is obviously satisfied if either $a$ or $b$ is greater than $1$, so we may assume that they are both less than $1$. Next, notice that when $a,b \in (0,1)$, $a^b$ increases when $a$ increases, and it decreases when $b$ increases.

Suppose that $b < \frac12.$ Then $b^a > a^a > 1-a^{1/ 2}$ (by the above inequality). But $a^{1/ 2} < a^b$, and so $1-a^{1/ 2} > 1-a^b.$ Thus $b^a > 1-a^b$, so that the inequality $a^b+b^a >1$ is proved.

The same argument shows that the inequality holds when $a < \frac12.$ So we only have to look at the case when both $a$ and $b$ lie between $\frac12$ and $1$. But then $a^b+b^a > a+b >1.$ That completes the proof.[/sp]

Thanks for participating, Opalg! Thanks for giving us such a thorough proof and I appreciate that!:)
kaliprasad said:
if a or b are above 1 we are done,.so let us assume both < 1
by Bernoulli Inequality -- from Wolfram MathWorld we have$(1+x)^n \ge 1+nx$so $(1+\dfrac{a}{b})^{\dfrac{1}{a}} \ge 1+\dfrac{1}{b}$so $(1+\dfrac{a}{b})^{\dfrac{1}{a}} \gt \dfrac{1}{b}$or $1+\dfrac{a}{b} \gt (\dfrac{1}{b})^a$or $\dfrac{a+b}{b} \gt (\dfrac{1}{b})^a$or $\dfrac{b}{a+b} \lt b^a$similarly

$\dfrac{a}{a+b} \lt a^b$
Adding, we get $1 \lt a^b + b^a$

or $a^b+b^a \gt 1$

Hey kaliprasad,

This is one very elegant proof, well done and thanks for participating!
 

Similar threads

  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 5 ·
Replies
5
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
Replies
1
Views
2K