Real numbers satisfy
and none of them lies in the open interval . Find the maximum value of .
Solutions — 2
Solution 1
Solution: By changing to , the condition (1) remains valid and the value of changes sign. Since the ordering of the numbers is irrelevant and they can not be of the same sign because of (1), we can without the loss of generality assume , and , and find the maximum value of .
We can transform (1) equivalently to
with . In the following, we shall use the well known fact about the function : It is an increasing bijection mapping onto (the trivial proof is omitted here). Applying this, from
we get , hence . Therefore our aim is to maximize the positive expression .
We will call valid every triple satisfying . One of the valid triples is , with such that , i. e., . For this triple, we have . We shall show this is the desired maximum. The proof will be based on the following lemma: Whenever and are valid triples with the same first component , and , we have . If this lemma is true, then, with respect to the symmetry, the similar conclusion holds for the valid triples and with the same second component . Hence, any valid triple can be replaced by and then by , and the value of increases or remains the same during this process, concluding . (Moreover, this also implies that is the only valid triple for which the maximum value is reached.)
To prove the lemma, notice that from
and from (which is implied by the given assumption ), we have , hence . Therefore , and from
using the estimates
we obtain , which is equivalent to the desired . Answer. The maximum value of is .
Solution 2
Second solution. We start in the same way as in the previous solution, reducing to the case , , and with maximizing . From (1) we can express as the solution of the quadratic equation
Since the product of the roots of this equation equals 1, the value of lying outside of is, with respect to , , given by
Therefore we are left to maximize
One can easily check that the expression inside of the last bars is positive. It remains
to show that
This inequality is equivalent (after rearranging, checking the positivity of both sides, canceling the square root, and simple calculations) to
which is true, since it is the sum of the inequalities
and these are trivially valid for .