Suppose that $0 <a,b < 1$ and $a+b=1$

Today, I did some investigation of the expression $$a^{\frac{1}{b}}+b^{\frac{1}{a}}$$ and it seems that the maximum is at $a=b=\frac{1}{2}$ where the expression is equal to $\frac {1}{2}$.

I would like to conjecture that we have $$a^{\frac{1}{b}}+b^{\frac{1}{a}}\leq \frac {1}{2}$$

Is this true?

  • $\begingroup$ @Macavity Where did you find greater value? $\endgroup$ – Grešnik Jun 24 at 16:15

Consider the concave function $f(t) = \sqrt[t]{1-t}$ (shown below.) Using Jensen’s inequality,

$$f(a) + f(b) \leqslant 2f(\tfrac12) = \tfrac12$$


To show $f$ is concave, it is enough to show $g=\log f$ is concave, as $t\mapsto e^t$ is convex and increasing. Perhaps the easiest way for this is to note the Taylor series for $x \in (0, 1)$ for $g= -\sum_{n \geqslant 0} \frac{x^n}{n+1} $ which implies all coefficients of $g’’$ are going to be negative as well $\implies g’’<0$.

  • $\begingroup$ I didn´t know that concavity is saved with compositions. $\endgroup$ – Grešnik Jun 24 at 16:35
  • $\begingroup$ Not exactly, but there are some compositions with monotone functions which provide deterministic results. For e.g. if $h$ is convex, decreasing, and $g$ is concave, then $h(g)$ is convex. Helps identify convexity sometimes when derivatives are more difficult, as is the present case. $\endgroup$ – Macavity Jun 24 at 16:45
  • $\begingroup$ @Macavity I think, to prove by hand that $f$ is a concave function it's not so easy. $\endgroup$ – Michael Rozenberg Jun 24 at 17:29
  • $\begingroup$ @MichaelRozenberg Have added a note above which explicitly shows $\log f$ Is concave, which is enough to conclude $f$ is. $\endgroup$ – Macavity Jun 24 at 18:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.