Can Jensen's Inequality Prove 1/a+1/b+1/c+1/d>1?

  • MHB
  • Thread starter anemone
  • Start date
In summary, the solution to this problem is to use Jensen's Inequality. This solution is much simpler than the other two solutions and will give the student a better understanding of the mathematics involved.
  • #1
anemone
Gold Member
MHB
POTW Director
3,883
115
Let $a,\,b,\,c,\,d>0$ such that $a<2$, $a+b<6$, $a+b+c<12$ and $a+b+c+d<24$.

Prove that $\dfrac{1}{a}+\dfrac{1}{b}+\dfrac{1}{c}+\dfrac{1}{d}>1$.
 
Mathematics news on Phys.org
  • #2
anemone said:
Let $a,\,b,\,c,\,d>0$ such that $a<2---(1)$, $a+b<6---(2)$, $a+b+c<12---(3)$ and $a+b+c+d<24---(4)$.

Prove that $\dfrac{1}{a}+\dfrac{1}{b}+\dfrac{1}{c}+\dfrac{1}{d}>1$.
from (1):$\dfrac {1}{a}>\dfrac {1}{2}---(5)$
using $AP\geq GP$
from (2)$6>a+b\geq 2\sqrt{ab}$
$\therefore ab< 9, or,\,\, \dfrac {1}{ab}>\dfrac {1}{9}$
$\dfrac {1}{a}+\dfrac {1}{b}\geq 2\sqrt {\dfrac {1}{ab}}>\dfrac {2}{3}--(6)$
from (3)$12>a+b+c\geq 3\sqrt[3]{abc}$
$\therefore abc< 64, or,\,\, \dfrac {1}{abc}>\dfrac {1}{64}$
$\dfrac {1}{a}+\dfrac {1}{b}+\dfrac {1}{c}\geq 3\sqrt [3]{\dfrac{1}{abc}}>\dfrac {3}{4}---(7)$
to be continued----

 
Last edited:
  • #3
other solution :
range of $a:1<a<2$
range of $b:1<b<4$
range of $c: 1<c<6$
range of $d : 1<d<12$
so $\dfrac {1}{a}+\dfrac{1}{b}+\dfrac{1}{c}+\dfrac {1}{d}>\dfrac {6+3+2+1}{12}=1$
for if anyone of $a,b,c,d\leq 1$ then there is no need to prove


 
Last edited:
  • #4
Albert said:
from (1):$\dfrac {1}{a}>\dfrac {1}{2}---(5)$
using $AP\geq GP$
from (2)$6>a+b\geq 2\sqrt{ab}$
$\therefore ab< 9, or,\,\, \dfrac {1}{ab}>\dfrac {1}{9}$
$\dfrac {1}{a}+\dfrac {1}{b}\geq 2\sqrt {\dfrac {1}{ab}}>\dfrac {2}{3}--(6)$
from (3)$12>a+b+c\geq 3\sqrt[3]{abc}$
$\therefore abc< 64, or,\,\, \dfrac {1}{abc}>\dfrac {1}{64}$
$\dfrac {1}{a}+\dfrac {1}{b}+\dfrac {1}{c}\geq 3\sqrt [3]{\dfrac{1}{abc}}>\dfrac {3}{4}---(7)$
to be continued----


Hi Albert,

I hope you would remember to finish the proof, but I will wait! :)

Albert said:
other solution :
range of $a:1<a<2$
range of $b:1<b<4$
range of $c: 1<c<6$
range of $d : 1<d<12$
so $\dfrac {1}{a}+\dfrac{1}{b}+\dfrac{1}{c}+\dfrac {1}{d}>\dfrac {6+3+2+1}{12}=1$
for if anyone of $a,b,c,d\leq 1$ then there is no need to prove

Hmm...I cannot say this proof is not correct, but I have the feeling that something is amiss...:confused: chances are I could be wrong though.

I will post back for the solution using the Jensen's Inequality method, not today but soon, I promise.
 
Last edited:
  • #5
I am sure my second solution is correct ,and nothing is amiss
I usually use simpler way to solve a math problem , this will give my students a better understanding
in fact there is something amiss in my first solution , for using $AP\geq GP$,
the ranges $0<a,b,c,d \leq 1$ have also taken into consideration
within this range ,the statement is always true there is no need to prove
 
  • #6
Albert said:
other solution :
range of $a:1<a<2$
range of $b:1<b<4$
range of $c: 1<c<6$
range of $d : 1<d<12$
so $\dfrac {1}{a}+\dfrac{1}{b}+\dfrac{1}{c}+\dfrac {1}{d}>\dfrac {6+3+2+1}{12}=1$
for if anyone of $a,b,c,d\leq 1$ then there is no need to prove

a = 1.2, b= 4.7 is a contradiction
 
  • #7
kaliprasad said:
a = 1.2, b= 4.7 is a contradiction
$\dfrac {1}{1.2}+\dfrac {1}{4.7}>1$ and there is no need to take c,d into consideration
when anyone of a,b,c,d is bigger than 1 and very close to 1 then its reciprocal very close to 1
for example if you take b=4 then $\dfrac {1}{b}=0.25$ ,and $a$ must bigger then $\dfrac {4}{3}>1.2 $ for $\dfrac {3}{4}=0.75$ ,the sum of $\dfrac {1}{a}+\dfrac {1}{b}=1$
may be you can help me to find better ranges for a,b,c,d
 
Last edited:
  • #8
Albert said:
$\dfrac {1}{1.2}+\dfrac {1}{4.7}>1$ and there is no need to take c,d into consideration
when anyone of a,b,c,d is bigger than 1 and very close to 1 then its reciprocal very close to 1
for example if you take b=4 then $\dfrac {1}{b}=0.25$ ,and $a$ must bigger then $\dfrac {4}{3}>1.2 $ for $\dfrac {3}{4}=0.75$ ,the sum of $\dfrac {1}{a}+\dfrac {1}{b}=1$
may be you can help me to find better ranges for a,b,c,d

what I meant that I have provided a case which contradicts your claim if b < 4 and your solution is inituitive and not correct mathemetically This is no way to provide a solution based on gut feeling.
 
  • #9
if a,b,c,d all should be taken into consideration ,then a,b,c,d must be as bigger as possible ,
that is 1/a, 1/b, 1/c and 1/d as smaller as possible (and "a" affects most)
for example even if we let b=5 ,then "a" very close to 5/4 =1.25>1.2
I admit my solution is not strict ,but sometimes by observation we can make the original problem easier for studens to understand
 
Last edited:
  • #10
Solution of other using Jensen's Inequality:

Applying Jensen's Inequality to the function $f(x)=\dfrac{1}{x}$ for $x>0$ gives

$\begin{align*}\dfrac{1}{a}+\dfrac{1}{b}+\dfrac{1}{c}+\dfrac{1}{d}&=\dfrac{1}{2}f\left(\dfrac{a}{2}\right)+\dfrac{1}{4}f\left(\dfrac{b}{4}\right)+\dfrac{1}{6}f\left(\dfrac{c}{6}\right)+\dfrac{1}{12}f\left(\dfrac{d}{12}\right)\\&>f\left(\dfrac{a}{4}+\dfrac{b}{16}+\dfrac{c}{36}+\dfrac{d}{144}\right)\\&=\dfrac{144}{36a+9b+4c+d}\\&=\dfrac{144}{27a+5(a+b)+3(a+b+c)+(a+b+c+d)}\\&>\dfrac{144}{27(2)+5(6)+3(12)+24}=1\end{align*}$

and we're hence done.
 

FAQ: Can Jensen's Inequality Prove 1/a+1/b+1/c+1/d>1?

How can I prove that 1/a+1/b+1/c+1/d>1?

To prove this inequality, we must first understand that it is only valid for positive values of a, b, c, and d. We can then use the properties of inequality to manipulate the equation and prove it to be true.

What are the key steps to proving this inequality?

The key steps to proving this inequality are: 1) Rewrite the expression as a single fraction, 2) Multiply both sides by the common denominator, 3) Simplify the resulting expression, and 4) Use the properties of inequality to show that the simplified expression is greater than 1.

Can I use any values for a, b, c, and d in this inequality?

No, this inequality is only valid for positive values of a, b, c, and d. If any of these values are negative, the inequality will not hold true.

How does this inequality relate to other mathematical concepts?

This inequality is a specific case of the Cauchy-Schwarz inequality, which states that for any two sequences of real numbers, the sum of their products is less than or equal to the product of their corresponding sums.

Can this inequality be extended to more than four variables?

Yes, this inequality can be extended to n variables. The general form is: 1/a1 + 1/a2 + ... + 1/an > n/(a1 + a2 + ... + an).

Similar threads

Replies
1
Views
909
Replies
3
Views
985
Replies
1
Views
1K
Replies
1
Views
935
Replies
8
Views
2K
Replies
1
Views
884
Back
Top