Showing posts with label inequality. Show all posts
Showing posts with label inequality. Show all posts

Wednesday, July 2, 2008

Proof of Trigonometric Inequalities

After yesterday's inequality induction, it's time for some trigonometric induction. The proof is a little hard to understand at first, but it's definitely worth seeing.

Prove:
xsin(1/x) < 1 when x > 0

This question can definitely fit in the short question-long answer template. However, it doesn't necessarily fit short question-hard answer.

First, we know that x is positive, so we can divide by it without changing the inequality sign:
sin(1/x) < 1/x

Now the argument of the sine function and the right side of the inequality are the same, so let's call them y. All we have to prove now is that sin y < y for every positive y.

First, when y is greater than 1, this is obvious. A sine can never be greater than 1. When y is 1, there is no problem either, because sin 1 is about 0.84 (the argument is in radians, don't forget). The problem begins when 0 < y < 1.

To solve that, let's derivate the inequality we want to prove. We get cos y < 1. As you can see, the rate of growth of sin y is cos y, and the rate of growth of y is linear. When y goes up by 0.5, sin y goes by less than that. This means that sin y is always less than y when 0 < y < 1.

Since y = 1/x, we can conclude that:
sin(1/x) < 1/x (when 1/x > 0, which means x > 0)
xsin(1/x) < 1

Problem solved.
Nadav

nadavs

Tuesday, July 1, 2008

Inequality Induction

Induction is an untouched topic in this blog so far, but I finally found a good question about it. It's not hard, but it's rather challenging.

Prove by induction:
(1 * 3 * 5 * 7 * ... * (2n - 1)) / (1 * 2 * 3 * 4 * ... * n) < 2n/√(2n + 1)

The first step in proving an induction is checking whether it works at all. Let's plug 1 for n:
1 / 1 < 21/√(2 + 1)
1 / 1 < 2 / √3 - correct!

Assume that the inequality is correct for n = k (k is natural)
(1 * 3 * 5 * ... * (2k - 1)) / (1 * 2 * 3 * ... * k) < 2k/√(2k + 1)

Now let's prove that if the inequality is right for n = k, it is right for n = k + 1:
(1 * 3 * 5 * ... * (2k - 1) * (2k + 1)) / (1 * 2 * 3 * ... * k * (k + 1)) < 2k + 1/√(2k + 3)

First, we know that (1 * 3 * 5 * ... * (2k - 1)) / (1 * 2 * 3 * ... * k) is smaller than 2k/√(2k + 1), so it's definitely smaller than 2k + 1/√(2k + 3). For this reason, we can place 2k/√(2k + 1) instead of this big expression:
(2k * (2k + 1))/(k + 1)√(2k + 1) < 2k + 1/√(2k + 3)

Divide by 2k:
(2k + 1)/(k + 1)√(2k + 1) < 2 / √(2k + 3)

Square both sides and cross multiply (k is natural, all positive):
4k2 + 8k + 3 < 4k2 + 8k + 3
3 < 4 - always true

We have shown that if n = k is correct, n = k + 1 is also correct. By checking that n = 1 is correct, we have proven that the inequality is correct for all natural numbers.

Enjoy,
Nadav

nadavs

Sunday, June 29, 2008

Logarithms and Inequalities

Being independent from Yahoo Answers feels great. I can go to my source, find a great question, and post it here. The topic are somewhat repeating, but there are many more questions. Today's question is about logarithms and inequalities.

Solve:
xlog2 x + log1/4 4 < 4


The solution here is very simple, yet it requires some mathematical knowledge, like what is log1/4 4. Well, the answer is -1. When the base and the argument are reciprocals, the logarithm's value is -1. Now we can really start solving.

xlog2 x - 1 < 4

As you can see, there is a log with base 2 in the exponent. To get rid of the exponent, let's take log base 2 from each side. Since the base is bigger than 1, the inequality sign remains as it is.
log2 xlog2 x - 1 < log2 4

Using the log property that says loga xn = nloga x, we can say that:
(log2 x - 1)log2 x < 2

Let t = log2 x

(t - 1)t < 2
t2 - t < 2
t2 - t - 2 < 0
(t - 2)(t + 1) < 0
-1 < t < 2

Plug the real value of t back:
-1 < log2 x < 2

By the definition of logs:
1/2 < x < 4

Hope you liked it,
Nadav

nadavs

Tuesday, June 24, 2008

Exponential Absolute Values

Complex numbers are incredible, but in order to use them we have to remember some definitions. Here is a question involving both complex numbers and exponential inequalities. It's not hard, but it requires some thinking.

Solve for x (x is real):
abs(2 + 3x2 - x - 1 - 12i) > 13

To do that, here is a reminder for the definition of absolute value:
Let z be a complex number.
z = a + bi
abs(z) = √(a2 + b2)

In the question:
a = 2 + 3x2 - x - 1
b = -12

The absolute value is:
√((2 + 3x2 - x - 1)2 + 122)
√(4 + 4*3x2 - x - 1 + (3x2 - x - 1)2 + 144)

Now return to the inequality:
√(4 + 4*3x2 - x - 1 + (3x2 - x - 1)2 + 144) > 13

Since both sides are always positive, we can square both sides without changing the inequality sign:
4 + 4*3x2 - x - 1 + (3x2 - x - 1)2 + 144 > 169

Let t = 3x2 - x - 1. This gives:
4 + 4t + t2 + 144 > 169
t2 + 4t - 21 > 0
(t + 7)(t - 3) > 0
t > 3 or t < -7 Since t is an exponent of 3, it can't be negative, so we're left with only one solution: t > 3.

Switch t back:
3x2 - x - 1 > 3

The base is bigger than 1, so the inequality sign remains the same:
x2 - x - 1 > 1
x2 - x - 2 > 0
(x - 2)(x + 1) > 0
x > 2 or x < -1

And that's it.

Hope you liked it,
Nadav

nadavs