Topics in Inequalities - Theorems and Techniques Hojoo ... - Index of
Topics in Inequalities - Theorems and Techniques Hojoo ... - Index of
Topics in Inequalities - Theorems and Techniques Hojoo ... - Index of
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
S<strong>in</strong>ce it is homogeneous, we may normalize to u + v + w = 3. We are now required to show that<br />
where G(t) = t x y , where t > 0. S<strong>in</strong>ce<br />
x<br />
y<br />
G(u) + G(v) + G(w)<br />
3<br />
G(u) + G(v) + G(w)<br />
3<br />
≥ 1,<br />
Similarly, we may deduce that M is monotone <strong>in</strong>creas<strong>in</strong>g on (−∞, 0).<br />
≥ 1, we f<strong>in</strong>d that G is convex. Jensen’s <strong>in</strong>equality shows that<br />
( )<br />
u + v + w<br />
≥ G<br />
= G(1) = 1.<br />
3<br />
We’ve learned that the convexity <strong>of</strong> f(x) = x λ (λ ≥ 1) implies the monotonicity <strong>of</strong> the power means.<br />
Now, we shall show that the convexity <strong>of</strong> x ln x also implies the power mean <strong>in</strong>equality.<br />
Second Pro<strong>of</strong> <strong>of</strong> the Monotonicity. Write f(x) = M (a,b,c) (x). We use the <strong>in</strong>creas<strong>in</strong>g function theorem. By<br />
the lemma 3, it’s enough to show that f ′ (x) ≥ 0 for all x ≠ 0. Let x ∈ R − {0}. We compute<br />
f ′ (x)<br />
f(x) = d<br />
dx (ln f(x)) = − 1 ( a x<br />
x 2 ln + b x + c x )<br />
+ 1 1<br />
3 (ax ln a + b x ln b + c x ln c)<br />
3 x<br />
1<br />
3 (ax + b x + c x )<br />
or<br />
x 2 f ′ (x)<br />
f(x)<br />
( a x + b x + c x<br />
= − ln<br />
3<br />
)<br />
+ ax ln a x + b x ln b x + c x ln c x<br />
a x + b x + c x .<br />
To establish f ′ (x) ≥ 0, we now need to establish that<br />
( a<br />
a x ln a x + b x ln b x + c x ln c x ≥ (a x + b x + c x x + b x + c x )<br />
) ln<br />
.<br />
3<br />
Let us <strong>in</strong>troduce a function f : (0, ∞) −→ R by f(t) = t ln t, where t > 0. After the substitution p = a x ,<br />
q = a y , r = a z , it becomes<br />
( ) p + q + r<br />
f(p) + f(q) + f(r) ≥ 3f<br />
.<br />
3<br />
S<strong>in</strong>ce f is convex on (0, ∞), it follows immediately from Jensen’s <strong>in</strong>equality.<br />
As a corollary, we obta<strong>in</strong> the RMS-AM-GM-HM <strong>in</strong>equality for three variables.<br />
Corollary 4.2.1. For all positive real numbers a, b, <strong>and</strong> c, we have<br />
√<br />
a2 + b 2 + c 2<br />
≥ a + b + c ≥ 3√ 3<br />
abc ≥<br />
3<br />
3<br />
1<br />
a + 1 b + 1 .<br />
c<br />
Pro<strong>of</strong>. The Power Mean <strong>in</strong>equality states that M (a,b,c) (2) ≥ M (a,b,c) (1) ≥ M (a,b,c) (0) ≥ M (a,b,c) (−1).<br />
Us<strong>in</strong>g the convexity <strong>of</strong> x ln x or the convexity <strong>of</strong> x λ<br />
the power means for n positive real numbers.<br />
(λ ≥ 1), we can also establish the monotonicity <strong>of</strong><br />
Theorem 4.2.2. (Power Mean <strong>in</strong>equality) Let x 1 , · · · , x n > 0. The power mean <strong>of</strong> order r is def<strong>in</strong>ed by<br />
( ) 1<br />
M (x1,··· ,x n)(0) = n√ x<br />
r r r<br />
x 1 · · · x n , M (x1,··· ,x n)(r) = 1 + · · · + x n<br />
(r ≠ 0).<br />
n<br />
Then, M (x1,··· ,x n) : R −→ R is cont<strong>in</strong>uous <strong>and</strong> monotone <strong>in</strong>creas<strong>in</strong>g.<br />
We conclude that<br />
Corollary 4.2.2. (Geometric Mean as a Limit) Let x 1 , · · · , x n > 0. Then,<br />
(<br />
n√ r ) 1<br />
r<br />
x1<br />
r<br />
+ · · · + x n x1 · · · x n = lim<br />
.<br />
r→0 n<br />
Theorem 4.2.3. (RMS-AM-GM-HM <strong>in</strong>equality) For all x 1 , · · · , x n > 0, we have<br />
√<br />
x12 + · · · + x n<br />
2<br />
n<br />
≥ x 1 + · · · + x n<br />
n<br />
≥ n√ n<br />
x 1 · · · x n ≥<br />
1<br />
x 1<br />
+ · · · + 1 .<br />
x n<br />
46