Sie sind auf Seite 1von 4

Inequalities

Brian Hamrick
December 8, 2008

Introduction

Inequality problems are somewhat more concrete than other problems and they fall to simpler techniques, with perhaps smaller leaps of faith. Oftentimes you can immediately judge whether youre on the right track or not by whether or not your work is making it simpler and whether or not what you have is still true. However, like functional equations, you need to be careful about the direction of implication, since these theorems only go one way. Hence, while youll be working backwards, you should always write up your proof forwards and conrm that every step actually works. Too often, people mix up the implications and claim that they have solved the problem, when theyve actually done absolutely nothing.

2
2.1

Basic Techniques
Substitution

Oftentimes youll have an unfortunate constraint to deal with. The most important substitutions eliminate constraints. For example, if you have three nonnegative real numbers b c a x, y, z such that xyz = 1, you can make the substitution x = , y = , z = and you obtain b c a a new inequality in a, b, c with no constraints. Another common substitution is that when you have the sides of a triangle, a, b, c, you can write a = x + y, b = y + z, c = z + x (if you need convincing, try drawing the incircle). When you have these common constraints, try these common substitutions.

2.2

Homogenization

We say an inequality is homogeneous if every (nonzero) term is of the same degree. For example, x2 0 and x2 + y 2 2xy are homogeneous while x2 x + 1 and xy + x + y 1 are not. Homogenization is a technique to take a non-homogeneous statement that we want to prove and nd an equivalent homogeneous statement that we can prove instead. We do this by multiplying by some power of our constraint. For example, if we know that xyz = 1, we can homogenize inequalities by multiplying terms by x1/3 y 1/3 z 1/3 = 1 to increase the degree of that term by 1. This is useful because many of the theorems only apply to homogeneous inequalities.

2.3

Isolated Fudging

Suppose you have an inequality where the left side is a sum of three terms, and the right is x , the 1. Well, what happens if you show that the rst term on the left is at most x+y+z y z second is at most , and the third is at most ? Well then clearly the sum x+y+z x+y+z is at most 1, so youre done. This is a clever trick and is used mostly with cyclic sums in ar three variables, and youll often compare the terms to r , but watch out for other a + br + c r things to fudge to.

2.4

Dumbassing

Dumbassing is a tried and true method and is the reason that test writers are hesitant to put three variable symmetric inequalities on an olympiad. All you have to do is multiply everything out, homogenize using your constraint, and then nd the appropriate applications of AM-GM and Schur. Try not to use it too much today, but be sure to practice it (its about equivalent to trig bashing a geometry problem).

3
3.1

The Theorems For Today


The Trivial Inequality

Well start out with something simple, but incredibly useful. That is the fact that x2 0 for all real x. Youll see how this is used in the rst problem.

3.2

AM-GM

This is the inequality that youll use the most. Ill start with the statement. For any nonnegative reals x1 , x2 , . . . , xn , a1 + a2 + + an n a1 a2 an n with equality if and only if x1 = x2 = = xn . In addition, there is a weighted version, which states that for nonnegative reals x1 , x2 , . . . , xn and arbitrary weights w1 , w2 , . . . , wn with w1 + w2 + + wn = 1, w1 x1 + w2 x2 + wn xn xw1 xw2 xwn 1 2 n , with equality if and only if x1 = x2 = = xn . Major signs for when youll use this include when you have conditions for either the sum or product of the variables and the other on the other side of the inequality.

3.3

The Cauchy-Schwarz Inequality


(a2 + a2 + + a2 )(b2 + b2 + + b2 ) (a1 b1 + a2 b2 + + an bn )2 n 2 1 n 2 1

This one works for all reals a1 , a2 , . . . , an , b1 , b2 , . . . , bn . It states that

a1 a2 an = = = . You can also remember this as The b1 b2 bn product of the sum of the squares is greater than or equal to the square of the sum of the products. with equality if and only if

3.4

Jensens Inequality

This inequality is useful especially in geometric inequalities where trigonometry is involved. It states that for a convex function f (x), real numbers x1 , x2 , . . . , xn , and weights w1 , w2 , . . . , wn with w1 + w2 + + wn = 1, w1 f (x1 ) + w2 f (x2 ) + + wn f (xn ) f (w1 x1 + w2 x2 + + wn xn ). This works for functions which are convex only on a certain domain, but to be safe you should only use positive weights if its only convex on an interval. The inequality also holds for concave functions, but with the inequality sign ipped ( rather than ).

3.5

Schurs Inequality

This inequality is considerably stronger than either of the rst two, but takes a similar form. It says that for three nonnegative reals x, y, z and a positive , x (x y)(x z) + y (y x)(y z) + z (z x)(z y) 0 . Schurs inequality is often dicult to motivate, since the dierent values of give dierent summations and the negative signs make it dicult to recognize the various forms.

Problems
1. Prove the Cauchy-Schwarz inequality using only that x2 0. 2. (Nesbitt Inequality) If a, b, c > 0, prove that a b c 3 + + . b+c c+a a+b 2 3. (Korea 1998) Let x, y, z > 0 with x + y + z = xyz. Show that 1 + 1 + x2 1 1+ 3 y2 + 1 3 . 2 2 1+z

4. (G.C.Giri) If a, b, c > 0, prove that 1 1 1 a8 + b 8 + c 8 + + a b c a3 b 3 c 3 . 5. If a, b, c are positive real numbers satisfying a2 + b2 + c2 = 1, nd the minimal value of S= a2 b 2 b 2 c 2 c 2 a2 + 2 + 2 . c2 a b

6. (Tournament of Towns 1997) Let a, b, c be positive numbers such that abc = 1. Prove that 1 1 1 + + 1 a+b+1 b+c+1 c+a+1 . 7. If x + y + z = 1 for some nonnegative numbers x, y, z, prove that 0 xy + yz + zx 2xyz 7 . 27

8. (1996 SL A1)Let a, b, c be positive real numbers such that abc = 1. Prove that ab bc ca + 5 + 5 1. a5 + b5 + ab b + c5 + bc c + a5 + ca When does equality hold? 9. (IMO 2001 2 / SL A6) Prove that for all positive real numbers a, b, c, a b c + + 1. a2 + 8bc b2 + 8ac c2 + 8ab

Das könnte Ihnen auch gefallen