Difference between revisions of "1974 IMO Problems/Problem 5"
| (One intermediate revision by the same user not shown) | |||
| Line 4: | Line 4: | ||
==Solution== | ==Solution== | ||
| + | |||
Note that <cmath>2 = \frac{a}{a+b}+\frac{b}{a+b}+\frac{c}{c+d}+\frac{d}{c+d} > S > \frac{a}{a+b+c+d}+\frac{b}{a+b+c+d}+\frac{c}{a+b+c+d}+\frac{d}{a+b+c+d} = 1.</cmath> We will now prove that <math>S</math> can reach any range in between <math>1</math> and <math>2</math>. | Note that <cmath>2 = \frac{a}{a+b}+\frac{b}{a+b}+\frac{c}{c+d}+\frac{d}{c+d} > S > \frac{a}{a+b+c+d}+\frac{b}{a+b+c+d}+\frac{c}{a+b+c+d}+\frac{d}{a+b+c+d} = 1.</cmath> We will now prove that <math>S</math> can reach any range in between <math>1</math> and <math>2</math>. | ||
| Line 15: | Line 16: | ||
Notice that if we treat the numerator and denominator each as a quadratic in <math>y</math>, we will get <math>1 + \frac{g(x)}{2y^2 + 5xy + 2x^2}</math>, where <math>g(x)</math> has a degree lower than <math>2</math>. This means taking <math>\lim_{y\to\infty} 1 + \frac{g(x)}{2y^2 + 5xy + 2x^2} = 1</math>, which means <math>S</math> can be brought arbitrarily close to <math>1</math>. Therefore, we are done. | Notice that if we treat the numerator and denominator each as a quadratic in <math>y</math>, we will get <math>1 + \frac{g(x)}{2y^2 + 5xy + 2x^2}</math>, where <math>g(x)</math> has a degree lower than <math>2</math>. This means taking <math>\lim_{y\to\infty} 1 + \frac{g(x)}{2y^2 + 5xy + 2x^2} = 1</math>, which means <math>S</math> can be brought arbitrarily close to <math>1</math>. Therefore, we are done. | ||
<cmath> </cmath> | <cmath> </cmath> | ||
| + | |||
~Imajinary | ~Imajinary | ||
| + | |||
| + | |||
| + | ==Remarks (added by pf02, October 2025)== | ||
| + | |||
| + | 1. Strictly speaking, the solution given above is incomplete. | ||
| + | It shows that <math>S</math> can be arbitrarily close to <math>1</math> and to <math>2</math>, | ||
| + | but it does not show that <math>S</math> can take <math>\mathbf{any}</math> value | ||
| + | in <math>(1, 2)</math>. This is not difficult to prove, but it is not | ||
| + | obvious or trivial, so it should be mentioned. | ||
| + | |||
| + | Here is an argument: Let <math>S_1</math> be given by <math>a_1, b_1, c_1, d_1</math> | ||
| + | and let <math>S_2</math> be given by <math>a_2, b_2, c_2, d_2</math>. (Think of <math>S_1</math> | ||
| + | close to <math>1</math> and <math>S_2</math> close to <math>2</math>). Let <math>a = (1 - t) a_1 + t a_2</math>, | ||
| + | and similarly for <math>b, c, d</math>. Let <math>S(t)</math> be given by <math>a, b, c, d</math>. | ||
| + | Clearly <math>S(t)</math> is a continuous function of <math>t</math>, <math>S(0) = S_1</math> and | ||
| + | <math>S(1) = S_2</math>. By the [[Intermediate Value Theorem]] for any | ||
| + | <math>S \in [S_1, S_2]</math> there is <math>t</math>, such that <math>S(t) = S</math>. | ||
| + | |||
| + | 2. Motivation for the solution: Take <math>a, b, c, d</math> of the same | ||
| + | magnitude, for example <math>a = b = c = d = 1</math>. Then <math>S = 1.333\dots\ .</math> | ||
| + | Now take <math>a, b</math> being of higher magnitude than <math>c, d</math>, for example | ||
| + | <math>a = b = 1000</math>, <math>c = d = 1</math>. We get <math>S = 1.001\dots\ .</math> And, take | ||
| + | <math>a, c</math> of higher magnitude than <math>b, d</math>, for example <math>a = c = 1000</math>, | ||
| + | <math>b = d = 1</math>. We get <math>S = 1.997\dots\ .</math> | ||
| + | |||
| + | Take a few more examples if desired. This suggests <math>1 < S < 2</math>, | ||
| + | and that we can get values as close to <math>1</math> and <math>2</math> as we want. | ||
| + | |||
| + | 3. I will give another solution below. It is computationally more | ||
| + | involved, but conceptually simpler, and it uses no calculus. | ||
| + | |||
| + | |||
| + | ==Solution 2== | ||
| + | |||
| + | Compute, but group terms to make the computations easier. | ||
| + | Add 1st and 3rd terms, 2nd and 4th terms, and group <math>a, c</math> | ||
| + | and <math>b, d</math>: | ||
| + | |||
| + | <math>S = \frac{a}{a+b+d} + \frac{b}{a+b+c} + \frac{c}{b+c+d} + \frac{d}{a+c+d} =</math> | ||
| + | |||
| + | <math>\frac{2ac + (a + c)(b + d)}{ac + (a + c)(b + d) + (a + c)^2} + | ||
| + | \frac{2bd + (a + c)(b + d)}{bd + (a + c)(b + d) + (b + d)^2}</math> | ||
| + | |||
| + | Showing that <math>S < 2</math> amounts to showing that | ||
| + | |||
| + | <math>4acbd + 3(ac + bd)(a + c)(b + d) + 2ac(b + d)^2 + 2bd(a + c)^2 + | ||
| + | 2(a + c)^2(b + d)^2 + (a + c)(b + d)[(a + c)^2 + (b + d)^2] <</math> | ||
| + | |||
| + | <math>2acbd + 2(ac + bd)(a + c)(b + d) + 2ac(b + d)^2 + 2bd(a + c)^2 + | ||
| + | 4(a + c)^2(b + d)^2 + 2(a + c)(b + d)[(a + c)^2 + (b + d)^2]</math> | ||
| + | |||
| + | This reduces to | ||
| + | |||
| + | <math>-2acbd - (ac + bd)(a + c)(b + d) + 2(a + c)^2(b + d)^2 + | ||
| + | (a + c)(b + d)[(a + c)^2 + (b + d)^2] > 0</math> | ||
| + | |||
| + | This is clearly true because the negative terms cancel with | ||
| + | parts from the positive terms: <math>-2acbd</math> is cancelled by terms | ||
| + | from <math>2(a + c)^2(b + d)^2</math> and <math>-(ac + bd)(a + c)(b + d)</math> is | ||
| + | cancelled by terms from <math>(a + c)(b + d)[(a + c)^2 + (b + d)^2]</math> | ||
| + | |||
| + | Showing that <math>S > 1</math> amounts to showing that | ||
| + | |||
| + | <math>4acbd + 3(ac + bd)(a + c)(b + d) + 2ac(b + d)^2 + 2bd(a + c)^2 + | ||
| + | 2(a + c)^2(b + d)^2 + (a + c)(b + d)[(a + c)^2 + (b + d)^2] ></math>the negative terms cancel with | ||
| + | parts from the positive terms: <math>-2acbd</math> is cancelled by terms | ||
| + | from <math>2(a + c)^2(b + d)^2</math> and <math>-(ac + bd)(a + c)(b + d)</math> is | ||
| + | cancelled by terms from <math>(a + c)(b + d)[(a + c)^2 + (b + d)^2]</math> | ||
| + | |||
| + | <math>acbd + (ac + bd)(a + c)(b + d) + ac(b + d)^2 + bd(a + c)^2 + | ||
| + | 2(a + c)^2(b + d)^2 + (a + c)(b + d)[(a + c)^2 + (b + d)^2]</math> | ||
| + | |||
| + | This reduces to | ||
| + | |||
| + | <math>3acbd + 2(ac + bd)(a + c)(b + d) + ac(b + d)^2 + bd(a + c)^2 > 0</math> | ||
| + | |||
| + | This is clearly true because all the terms are <math>> 0</math>. | ||
| + | |||
| + | Now show that any number <math>S \in (1, 2)</math> is the result of the sum | ||
| + | of the fractions in the statement of the problem for some | ||
| + | <math>a, b, c, d</math>. First, look for <math>a = c, b = d</math> which would yield | ||
| + | <math>S</math>. We are looking for <math>a, b</math> such that | ||
| + | <math>S = \frac{2a}{a + 2b} + \frac{2b}{2a + b}</math>. After some easy | ||
| + | computations this becomes | ||
| + | |||
| + | <math>(4 - 2S)\left(\frac{a}{b}\right)^2 - (5S - 4)\left(\frac{a}{b}\right) + | ||
| + | (4 - 2S) = 0</math> | ||
| + | |||
| + | This yields <math>\frac{a}{b} = \frac{5S - 4 \pm \sqrt{9S^2 + 24S - 48}}{4 - 2S}</math> | ||
| + | |||
| + | Remembering that <math>S > 0</math>, we see that we obtain <math>\frac{a}{b}</math> real and | ||
| + | positive when <math>S \in \left[\frac{4}{3}, 2\right)</math>. | ||
| + | |||
| + | Now, look for <math>a = b, c = d</math> which would yield | ||
| + | <math>S</math>. We are looking for <math>a, c</math> such that | ||
| + | <math>S = \frac{2a}{2a + c} + \frac{2c}{a + 2c}</math>. After some easy | ||
| + | computations this becomes | ||
| + | |||
| + | <math>(2S - 2)\left(\frac{a}{c}\right)^2 - (8 - 5S)\left(\frac{a}{c}\right) + | ||
| + | (2S - 2) = 0</math> | ||
| + | |||
| + | This yields <math>\frac{a}{c} = \frac{8 - 5S \pm \sqrt{9S^2 - 48S + 48}}{4S - 4}</math> | ||
| + | |||
| + | Remembering that <math>S > 0</math>, we see that we obtain <math>\frac{a}{c}</math> real and | ||
| + | positive when <math>S \in \left(1, \frac{4}{3}\right]</math>. | ||
| + | |||
| + | Thus, any <math>S \in (1, 2)</math> is the result of the sum of the fractions in the | ||
| + | statement of the problem for some <math>a, b, c, d</math>. | ||
| + | |||
| + | [Solution by pf02, October 2025] | ||
| + | |||
== See Also == {{IMO box|year=1974|num-b=4|num-a=6}} | == See Also == {{IMO box|year=1974|num-b=4|num-a=6}} | ||
Latest revision as of 16:32, 14 October 2025
Problem 5
Determine all possible values of
where
are arbitrary positive numbers.
Solution
Note that
We will now prove that
can reach any range in between
and
.
Choose any positive number
. For some variables such that
and
, let
,
, and
. Plugging this back into the original fraction, we get
The above equation can be further simplified to
Note that
is a continuous function and that
is a strictly increasing function. We can now decrease
and
to make
tend arbitrarily close to
. We see
, meaning
can be brought arbitrarily close to
.
Now, set
and
for some positive real numbers
. Then
Notice that if we treat the numerator and denominator each as a quadratic in
, we will get
, where
has a degree lower than
. This means taking
, which means
can be brought arbitrarily close to
. Therefore, we are done.
~Imajinary
Remarks (added by pf02, October 2025)
1. Strictly speaking, the solution given above is incomplete.
It shows that
can be arbitrarily close to
and to
,
but it does not show that
can take
value
in
. This is not difficult to prove, but it is not
obvious or trivial, so it should be mentioned.
Here is an argument: Let
be given by
and let
be given by
. (Think of
close to
and
close to
). Let
,
and similarly for
. Let
be given by
.
Clearly
is a continuous function of
,
and
. By the Intermediate Value Theorem for any
there is
, such that
.
2. Motivation for the solution: Take
of the same
magnitude, for example
. Then
Now take
being of higher magnitude than
, for example
,
. We get
And, take
of higher magnitude than
, for example
,
. We get
Take a few more examples if desired. This suggests
,
and that we can get values as close to
and
as we want.
3. I will give another solution below. It is computationally more involved, but conceptually simpler, and it uses no calculus.
Solution 2
Compute, but group terms to make the computations easier.
Add 1st and 3rd terms, 2nd and 4th terms, and group
and
:
Showing that
amounts to showing that
This reduces to
This is clearly true because the negative terms cancel with
parts from the positive terms:
is cancelled by terms
from
and
is
cancelled by terms from
Showing that
amounts to showing that
the negative terms cancel with
parts from the positive terms:
is cancelled by terms
from
and
is
cancelled by terms from
This reduces to
This is clearly true because all the terms are
.
Now show that any number
is the result of the sum
of the fractions in the statement of the problem for some
. First, look for
which would yield
. We are looking for
such that
. After some easy
computations this becomes
This yields
Remembering that
, we see that we obtain
real and
positive when
.
Now, look for
which would yield
. We are looking for
such that
. After some easy
computations this becomes
This yields
Remembering that
, we see that we obtain
real and
positive when
.
Thus, any
is the result of the sum of the fractions in the
statement of the problem for some
.
[Solution by pf02, October 2025]
See Also
| 1974 IMO (Problems) • Resources | ||
| Preceded by Problem 4 |
1 • 2 • 3 • 4 • 5 • 6 | Followed by Problem 6 |
| All IMO Problems and Solutions | ||