Arithmetic Compensation Method
Arithmetic Compensation Method
Arithmetic Compensation Method
Vasile Cı̂rtoaje
University of Ploiesti,
Romania
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
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
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
Solution. Let
(1 + 2a)(1 + 2b)(1 + 2c)(1 + 2d)
F (a, b, c, d) = − .
(1 − a)(1 − b)(1 − c)(1 − d)
(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
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 .