Following problem is from Exercise $14.15$ of F. Bass' Real Analysis.

Let $(\Omega,\Sigma,\mu)$ be a measure space with $\mu(\Omega)<\infty$ and $f$ be a real-valued integrable function. Define $$g(x)=\int_\Omega|f(y)-x|~d\mu(y)$$ for $x\in\mathbb{R}$. Show that $g$ is absolutely continuous. Find $g^\prime(x)$ and prove that $g(x_0)=\inf_{x\in\mathbb{R}}g(x)$ if and only if $$\mu(\{y:f(y)>x_0\})=\mu(\{y:f(y)<x_0\}).$$

According to the definition of absolutely continuous functions, I can prove $g$ is absolutely continuous. But I'm puzzled by the second question about derivatives. Any help is appreciated!

  • $\begingroup$ If $\mu ( \Omega) = \infty,$ then $g(x) = \infty$ for all $x\ne 0.$ $\endgroup$ – zhw. Oct 26 '17 at 4:43
  • $\begingroup$ Alright then. So the original statement should assume that $\Omega$ has finite measure. $\endgroup$ – Jay Oct 26 '17 at 4:52


Proceeding formally you would guess that $g'(x) = -\int_\Omega \operatorname{sgn}(f(y)-x)\,d\mu(y)$. Can you prove that by integrating this guess you recover $g$? (Try Fubini.) If so then it confirms both the absolute continuity and the formula for $g'$.

For the part about the minimum, it might help to observe that if the above guess is right, then the last statement holds for $x_0$ iff $g'(x_0) = 0$. And you would expect the derivative of $g$ to vanish at the minimum of $g$.


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.