0% found this document useful (0 votes)
77 views5 pages

Arithmetic Compensation Method

This document presents the Arithmetic Compensation Method, which can be used to prove certain difficult symmetric inequalities. It defines the Arithmetic Compensation Theorem in both short and extended forms, and provides proofs of each. It then applies the theorem to solve three example problems involving inequalities over variables a, b, c, d where their sum is a constant.

Uploaded by

Ntc Dsx Vdu
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
77 views5 pages

Arithmetic Compensation Method

This document presents the Arithmetic Compensation Method, which can be used to prove certain difficult symmetric inequalities. It defines the Arithmetic Compensation Theorem in both short and extended forms, and provides proofs of each. It then applies the theorem to solve three example problems involving inequalities over variables a, b, c, d where their sum is a constant.

Uploaded by

Ntc Dsx Vdu
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Arithmetic Compensation Method

Vasile Cı̂rtoaje
University of Ploiesti,
Romania

The Arithmetic Compensation Method is a powerful tool which can be used


to prove certain difficult symmetric inequalities. Such a problem, which was left
unsolved on the Mathlnks Inequalities Forum, is presented here:
64
Problem. Let a, b, c, d ≥ 0 such that a + b + c + d = 4. For p > 27 , what is
the minimum value of the expression
1 1 1 1
+ + + ?
p − abc p − bcd p − cda p − dab
Arithmetic Compensation Theorem (Short Form). Let s > 0 and let
F (x1 , x2 , · · · , xn ) be a symmetrical continuous function on the compact set in
Rn

S = {(x1 , x2 , . . . , xn ): x1 + x2 + · · · + xn = s, x1 ≥ 0, . . . , xn ≥ 0}.

If
x1 + x2 x1 + x2
F (x1 , x2 , x3 , . . . , xn ) < max{F ( , , x3 , . . . , xn ), F (0, x1 +x2 , x3 , . . . , xn )}
2 2
for all (x1 , x2 , . . . , xn ) ∈ S with x1 > x2 > 0, then
s s
F (x1 , x2 , . . . , xn ) ≤ max F ( , . . . , , 0, . . . , 0)
1≤k≤n k k
for all (x1 , x2 , . . . , xn ) ∈ S.
Proof. Since the function F is continuous on the compact set S, F attains a
maximum value at one or more points of the set. Let (x1 , x2 , . . . , xn ) be such
a maximum point. For the sake of contradiction, assume that there exist two
numbers xi and xj such that xi > xj > 0; for convenience, let us consider i = 1
and j = 2 (hence x1 > x2 > 0). According to the hypothesis, we have
x1 + x2 x1 + x2
F (x1 , x2 , x3 , . . . , xn ) < max{F ( , , x3 , . . . , xn ), F (0, x1 +x2 , x3 , . . . , xn )}.
2 2
But this is false because F is maximal at (x1 , x2 , . . . , xn ), and the theorem is
proved.

1
Arithmetic Compensation Theorem (Extended Form). Let s > 0 and
let F (x1 , x2 , · · · , xn ) be a symmetrical continuous function on the compact set
in Rn

S = {(x1 , x2 , . . . , xn ): x1 + x2 + · · · + xn = s, x1 ≥ 0, . . . , xn ≥ 0}.

If
x1 + x2 x1 + x2
F (x1 , x2 , x3 , . . . , xn ) ≤ max{F ( , , x3 , . . . , xn ), F (0, x1 +x2 , x3 , . . . , xn )}
2 2
for all (x1 , x2 , . . . , xn ) ∈ S with x1 > x2 > 0, then
s s
F (x1 , x2 , . . . , xn ) ≤ max F ( , . . . , , 0, . . . , 0)
1≤k≤n k k

for all (x1 , x2 , . . . , xn ) ∈ S.


Proof. In order to prove this theorem, we will show that among the maximum
points of F there exists at least one point (y1 , y2 , . . . , yn ) such that all yi ∈
{0, ks }, where 1 ≤ k ≤ n. Let (x1 , x2 , . . . , xn ) be a maximum point. Again
suppose by way of contradiction that x1 > x2 > 0. We have considered the case
where the inequality in the hypothesis is strict; we now prove the conclusion for
the equality case in the hypothesis; that is, when
x1 + x2 x1 + x2
F (x1 , x2 , x3 , . . . , xn ) = max{F ( , , x3 , . . . , xn ), F (0, x1 +x2 , x3 , . . . , xn )}
2 2
The function F attains again its maximum value at (y1 , y2 , . . . , yn ) with yi =
xi for i ≥ 3 and either y1 = y2 = x1 +x 2
2
or y1 = 0 and y2 = x1 + x2 . If
there are not two numbers yi and yj such that yi > yj > 0, then the proof is
finished. Otherwise, we iterate the preceding process, eventually in the limiting
case finding a maximum point (z1 , z2 , . . . , zn ) such that all zi ∈ {0, ks }, where
1 ≤ k ≤ n.

Applications
Problem 1. If a, b, c, d ≥ 0 such that a + b + c + d = 4, then
1 1 1 1
+ + + ≤ 1.
5 − abc 5 − bcd 5 − cda 5 − dab
Solution. If at least two of the numbers a, b, c, d are equal to zero, then the
inequality is clearly true. Otherwise, let us denote by F (a, b, c, d) the left hand
side of the inequality. We claim that for a > b > 0, the inequality of the theorem
holds; that is, we claim that
a+b a+b
F (a, b, c, d) < max{F ( , , c, d), F (0, a + b, c, d)} (1)
2 2

Mathematical Reflections 3, (2006) 2


Then, by the short form of the theorem, it follows that
4 4 4
F (a, b, c, d) < max{F (4, 0, 0, 0), F (2, 2, 0, 0), F ( , , , 0), F (1, 1, 1, 1)}.
3 3 3
Since F (4, 0, 0, 0) = F (2, 2, 0, 0) = 54 , F ( 43 , 43 , 43 , 0) = 348
355 , and F (1, 1, 1, 1) = 1,
we see that F (a, b, c, d) ≤ 1, which is the desired inequality.
In order to prove (1), we will assume for the sake of contradiction that there
exist a > b > 0, c > 0 and d ≥ 0 such that
F (a, b, c, d) ≥ F (t, t, c, d) and
F (a, b, c, d) ≥ F (0, 2t, c, d),
a+b
where t = 2 . Write now the inequality F (a, b, c, d) ≥ F (t, t, c, d) in the form
2(5 − tcd) 2 1 1 1 1
− ≥( 2
− )+( 2
− ).
(5 − acd)(5 − bcd) 5 − tcd 5 − t c 5 − abc 5 − t d 5 − abd
Dividing by the positive factor t2 − ab, the inequality becomes
2c2 d2 c d
≥ + .
(5 − acd)(5 − bcd)(5 − tcd) (5 − acd)(5 − t c) (5 − abd)(5 − t2 d)
2

Since
c d c d
+ > + ,
(5 − acd)(5 − t2 c) (5 − abd)(5 − t2 d) 5(5 − t2 c) 5(5 − t2 d)
we get
2c2 d2 c d
> + . (2)
(5 − acd)(5 − bcd)(5 − tcd) 5(5 − t c) 5(5 − t2 d)
2

Similarly, write the inequality F (a, b, c, d) ≥ F (0, 2t, c, d) as follows:


1 1 1 1 1 1 1 1
( − )+( − )+( − )≥ +
5 − abc 5 5 − abd 5 5 − acd 5 − bcd 5 5 − 2tcd
abc abd 2(5 − tcd) 2(5 − tcd)
+ + ≥
5(5 − abc) 5(5 − abd) (5 − acd)(5 − bcd) 5(5 − 2tcd)
c d 2c2 d2 (5 − tcd)
+ ≥ .
5(5 − abc) 5(5 − abd) 5(5 − acd)(5 − bcd)(5 − 2tcd)
Since
5 − tcd 5
≥ ,
5 − 2tcd 5 − tcd
we get
c d 2c2 d2
+ ≥ ,
5(5 − abc) 5(5 − abd) (5 − acd)(5 − bcd)(5 − tcd)
which contradicts (2). This completes the proof. Equality occurs if and only if
a = b = c = d = 1.

Mathematical Reflections 3, (2006) 3


Problem 2. If a, b, c, d ≥ 0 such that a + b + c + d = 4, then
1 1 1 1 15
+ + + ≤ .
4 − abc 4 − bcd 4 − cda 4 − dab 11
Solution. If at least two of the numbers a, b, c, d are equal to zero, then the
inequality is true. Otherwise, as in the preceding problem, we can show that
a+b a+b
F (a, b, c, d) < max{F ( , , c, d), F (0, a + b, c, d)}.
2 2
By the theorem, we have
4 4 4
F (a, b, c, d) < max{F (4, 0, 0, 0), F (2, 2, 0, 0), F ( , , , 0), F (1, 1, 1, 1)}.
3 3 3
Since F (4, 0, 0, 0) = F (2, 2, 0, 0) = 1, F ( 34 , 43 , 43 , 0) = 15 4
11 , and F (1, 1, 1, 1) = 3 ,
the conclusion follows. Equality occurs when one of a, b, c, d equals zero and the
other three equal 34 .
Problem 3. If a, b, c, d ≥ 0 such that a + b + c + d = 1, then

(1 + 2a)(1 + 2b)(1 + 2c)(1 + 2d) 125


≥ .
(1 − a)(1 − b)(1 − c)(1 − d) 8

Solution. Let
(1 + 2a)(1 + 2b)(1 + 2c)(1 + 2d)
F (a, b, c, d) = − .
(1 − a)(1 − b)(1 − c)(1 − d)

We claim that for a > b > 0,


a+b a+b
F (a, b, c, d) < max{F ( , , c, d), F (0, a + b, c, d)}. (3)
2 2
Then, by the extended form of the theorem, we have
1 1 1 1 1 1 1 1 1
F (a, b, c, d) ≤ max{F ( , , 0, 0), F ( , , , 0), F ( , , , )}.
2 2 3 3 3 4 4 4 4
Since F ( 21 , 12 , 0, 0) = −16, F ( 13 , 31 , 13 , 0) = − 125 1 1 1 1
8 , and F ( 4 , 4 , 4 , 4 ) = −16, we
125
get F (a, b, c, d) ≤ − 8 which is the desired inequality.
The inequality (3) is equivalent to

(1 + 2a)(1 + 2b) 1 + 2t 2 1 + 4t
≥ min{( ) , }.
(1 − a)(1 − b) 1−t 1 − 2t

The inequality
(1 + 2a)(1 + 2b) 1 + 2t 2
≥( )
(1 − a)(1 − b) 1−t

Mathematical Reflections 3, (2006) 4


is equivalent to

3(4t − 1)(t2 − ab)


≥ 0, (4)
(1 − t)(1 − a)(1 − b)

and the inequality


(1 + 2a)(1 + 2b) 1 + 4t

(1 − a)(1 − b) 1 − 2t
is equivalent to

3ab(−4t + 1)
≥ 0. (5)
(1 − 2t)(1 − a)(1 − b)

Since (4) is true for t ≥ 41 and (5) is true for t ≤ 41 , the proof is completed.
Equality occurs when one of the numbers a, b, c, d is equal to zero, and the others
equal 13 .

Mathematical Reflections 3, (2006) 5

You might also like