Difference between revisions of "1966 IMO Problems/Problem 2"
GausssWill (talk | contribs) m |
m |
||
(9 intermediate revisions by 3 users not shown) | |||
Line 1: | Line 1: | ||
− | + | ==Problem== | |
− | < | + | Let <math>a</math>, <math>b</math>, and <math>c</math> be the lengths of the sides of a triangle, and <math> \alpha,\beta,\gamma </math> respectively, the angles opposite these sides. Prove that if |
− | + | <cmath> a+b=\tan{\frac{\gamma}{2}}(a\tan{\alpha}+b\tan{\beta}), </cmath> | |
+ | |||
+ | the triangle is isosceles. | ||
+ | |||
+ | |||
+ | ==Solution== | ||
+ | |||
+ | We'll prove that the triangle is isosceles with <math>a=b</math>. | ||
+ | We'll prove that <math>a=b</math>. Assume by way of contradiction WLOG that <math>a>b</math>. | ||
+ | First notice that as <math>\gamma = \pi -\alpha-\beta</math> then and the identity <math>\tan\left(\frac \pi 2 - x \right)=\cot x</math> our equation becomes: | ||
+ | <cmath>a+b=\cot \frac{\alpha +\beta}{2}\left(a\tan \alpha + b\tan \beta \right)</cmath><cmath>\iff a\tan\frac{\alpha +\beta}{2}+b\tan \frac{\alpha +\beta}{2}=a\tan \alpha + b\tan \beta </cmath> | ||
+ | <cmath>\iff a\left(\tan \alpha -\tan \frac{\alpha +\beta}{2}\right)+b\left(\tan \beta -\tan \frac{\alpha +\beta}{2} \right)=0 </cmath> | ||
+ | Using the identity <math>\tan (A-B)=\frac {\tan A-\tan B}{1+\tan A\tan B}</math> <math> \iff \tan A-\tan B=\tan(A-B)(1+\tan A\tan B)</math> | ||
+ | and inserting this into the above equation we get: | ||
+ | <cmath>\iff a\tan \frac{\alpha -\beta}{2}\left(1+\tan \alpha \tan \frac{\alpha +\beta}{2}\right)+b\tan \frac{\beta -\alpha}{2}\left(1+\tan \beta \tan \frac{\alpha +\beta}{2} \right)=0 </cmath> | ||
+ | <cmath>\underbrace{\iff}_{\tan -A=-\tan A}a\tan \frac{\alpha -\beta}{2}\left(1+\tan \alpha \tan \frac{\alpha +\beta}{2}\right)-b\tan \frac{\alpha -\beta}{2}\left(1+\tan \beta \tan \frac{\alpha +\beta}{2} \right)=0 </cmath> | ||
+ | <cmath>\iff \tan \frac{\alpha -\beta}{2}\left(a-b+\tan \frac{\alpha +\beta}{2}(a\tan\alpha -b\tan \beta) \right)=0 </cmath> | ||
+ | Now, since <math>a>b</math> and the definitions of <math>a,b,\alpha,\beta</math> being part of the definition of a triangle, <math>\alpha >\beta</math>. | ||
+ | Now, <math>\pi >\alpha -\beta >0</math> (as <math>\alpha+\beta +\gamma = \pi</math> and the angles are positive), <math>\tan \frac{\alpha -\beta}{2}\neq 0</math>, and furthermore, <math>\tan \frac{\alpha+\beta}{2}>0</math>. By all the above, <cmath>\left(a-b+\tan \frac{\alpha +\beta}{2}(a\tan\alpha -b\tan \beta) \right)>0</cmath> | ||
+ | Which contradicts our assumption, thus <math>a\leq b</math>. By the symmetry of the condition, using the same arguments, <math>a\geq b</math>. Hence <math>a=b</math>. | ||
+ | |||
+ | |||
+ | ==Solution 2== | ||
+ | |||
+ | First, we'll prove that both <math>\alpha</math> and <math>\beta</math> are acute. | ||
+ | At least one of them has to be acute because these are angles | ||
+ | of a triangle. We can assume that <math>\alpha</math> is acute. We want | ||
+ | to show that <math>\beta</math> is acute as well. For a proof by | ||
+ | contradiction, assume <math>\beta \ge \frac{\pi}{2}</math>. | ||
+ | |||
+ | From the hypothesis, it follows that | ||
+ | <math>(a + b) \tan \frac{\alpha + \beta}{2} = a \tan \alpha + b \tan \beta</math>. | ||
+ | |||
+ | From <math>\alpha < \frac{\pi}{2} \le \beta</math> it follows that <math>a < b</math>. So, | ||
+ | |||
+ | <math>b \tan \beta = (a + b) \tan \frac{\alpha + \beta}{2} - a \tan \alpha > | ||
+ | 2a \tan \frac{\alpha + \beta}{2} - a \tan \alpha \ge | ||
+ | a (2 \tan \left( \frac{\alpha}{2} + \frac{\pi}{4} \right) - \tan \alpha) =</math> | ||
+ | |||
+ | <math>2a \left( \frac{\tan \frac{\alpha}{2} + 1}{1 - \tan \frac{\alpha}{2}} - | ||
+ | \frac{\tan \frac{\alpha}{2}}{1 - \tan^2 \frac{\alpha}{2}} \right) = | ||
+ | 2a \cdot \frac{\tan^2 \frac{\alpha}{2} + \tan \frac{\alpha}{2} + 1} | ||
+ | {1 - \tan^2 \frac{\alpha}{2}} > 0</math> | ||
+ | |||
+ | because the numerator is <math>> 0</math> (because <math>Y^2 + Y + 1 > 0</math> for any real | ||
+ | <math>Y</math>), and the denominator is also <math>> 0</math> (because <math>\alpha < \frac{\pi}{2}</math>, | ||
+ | so <math>\tan \frac{\alpha}{2} < \tan \frac{\pi}{4} = 1</math>). | ||
+ | |||
+ | It follows that <math>\tan \beta > 0</math>, so it can not be that | ||
+ | <math>\beta \ge \frac{\pi}{2}</math>. | ||
+ | |||
+ | Now, we will prove that | ||
+ | <math>(a + b) \tan \frac{\alpha + \beta}{2} = a \tan \alpha + b \tan \beta</math> | ||
+ | implies <math>\alpha = \beta</math>. | ||
+ | |||
+ | Replace <math>a = \sin \alpha \cdot 2R</math> and <math>b = \sin \beta \cdot 2R</math> | ||
+ | (in fact, we don't care that <math>R</math> is the radius of the circumscribed | ||
+ | circle), and simplify by <math>2R</math>. We get | ||
+ | |||
+ | <math>(\sin \alpha + \sin \beta) \tan \frac{\alpha + \beta}{2} = | ||
+ | \sin \alpha \tan \alpha + \sin \beta \tan \beta</math>. | ||
+ | |||
+ | This becomes | ||
+ | |||
+ | <math> \left( \sin \frac{\alpha + \beta}{2} \tan \frac{\alpha + \beta}{2} \right) | ||
+ | \cdot \cos \frac{\alpha - \beta}{2} = | ||
+ | \frac{1}{2}(\sin \alpha \tan \alpha + \sin \beta \tan \beta)</math> | ||
+ | |||
+ | We will show that the function <math>f(x) = \tan x \sin x</math> is convex on | ||
+ | the interval <math> \left( 0, \frac{\pi}{2} \right)</math>. Indeed, the first | ||
+ | derivative is <math>f'(x) = \frac{\sin x}{\cos^2 x} + \sin x</math>, and the | ||
+ | second derivative is <math>f''(x) = \frac{\cos^4 x - \cos ^2 x + 2}{\cos^3 x}</math>. | ||
+ | |||
+ | We have <math>f''(x) > 0</math> on <math> \left( 0, \frac{\pi}{2} \right)</math> since the | ||
+ | numerator is <math>> 0</math> (because <math>Y^2 - Y + 1 > 0</math> for any real <math>Y</math>), and | ||
+ | the denominator is <math>> 0</math> on the interval | ||
+ | <math> \left( 0, \frac{\pi}{2} \right)</math>. It follows that <math>f(x) = \tan x \sin x</math> | ||
+ | is convex on the interval <math> \left( 0, \frac{\pi}{2} \right)</math>. | ||
+ | |||
+ | Using the convexity we have | ||
+ | <math>f \left( \frac{x + y}{2} \right) \le \frac{1}{2} (f(x) + f(y))</math>. In | ||
+ | our case, we have | ||
+ | |||
+ | <math>\frac{1}{2}(\sin \alpha \tan \alpha + \sin \beta \tan \beta) = | ||
+ | \left( \sin \frac{\alpha + \beta}{2} \tan \frac{\alpha + \beta}{2} \right) | ||
+ | \cdot \cos \frac{\alpha - \beta}{2} \le | ||
+ | \frac{1}{2}(\sin \alpha \tan \alpha + \sin \beta \tan \beta) \cdot | ||
+ | \cos \frac{\alpha - \beta}{2}</math>. | ||
+ | |||
+ | We can simplify by <math>\sin \alpha \tan \alpha + \sin \beta \tan \beta</math> | ||
+ | because it is positive (because both <math>\alpha, \beta</math> are acute!), | ||
+ | and we get | ||
+ | |||
+ | <math>1 \le \cos \frac{\alpha - \beta}{2}</math>. This is possible only when | ||
+ | <math>\cos \frac{\alpha - \beta}{2} = 1</math>, i.e. <math>\alpha = \beta</math>. | ||
+ | |||
+ | [Solution by pf02, September 2024] | ||
+ | |||
+ | |||
+ | ==See Also== | ||
+ | {{IMO box|year=1966|num-b=1|num-a=3}} |
Latest revision as of 19:17, 10 November 2024
Contents
Problem
Let , , and be the lengths of the sides of a triangle, and respectively, the angles opposite these sides. Prove that if
the triangle is isosceles.
Solution
We'll prove that the triangle is isosceles with . We'll prove that . Assume by way of contradiction WLOG that . First notice that as then and the identity our equation becomes: Using the identity and inserting this into the above equation we get: Now, since and the definitions of being part of the definition of a triangle, . Now, (as and the angles are positive), , and furthermore, . By all the above, Which contradicts our assumption, thus . By the symmetry of the condition, using the same arguments, . Hence .
Solution 2
First, we'll prove that both and are acute. At least one of them has to be acute because these are angles of a triangle. We can assume that is acute. We want to show that is acute as well. For a proof by contradiction, assume .
From the hypothesis, it follows that .
From it follows that . So,
because the numerator is (because for any real ), and the denominator is also (because , so ).
It follows that , so it can not be that .
Now, we will prove that implies .
Replace and (in fact, we don't care that is the radius of the circumscribed circle), and simplify by . We get
.
This becomes
We will show that the function is convex on the interval . Indeed, the first derivative is , and the second derivative is .
We have on since the numerator is (because for any real ), and the denominator is on the interval . It follows that is convex on the interval .
Using the convexity we have . In our case, we have
.
We can simplify by because it is positive (because both are acute!), and we get
. This is possible only when , i.e. .
[Solution by pf02, September 2024]
See Also
1966 IMO (Problems) • Resources | ||
Preceded by Problem 1 |
1 • 2 • 3 • 4 • 5 • 6 | Followed by Problem 3 |
All IMO Problems and Solutions |