Here is Prob. 2, Chap. 6, in the book Principles of Mathematical Analysis by Walter Rudin, 3rd edition:

Suppose $f \geq 0$, $f$ is continuous on $[a, b]$, and $\int_a^b f(x) \ \mathrm{d} x = 0$. Prove that $f(x) = 0$ for all $x \in [a, b]$.

My Attempt:

Let $c$ and $d$ be any points in $[a, b]$ such that $c \leq d$. As $f(x) \geq 0$ for all $x \in [a, b]$ and as $f$ is continuous on $[a, b]$, so $f(x) \geq 0$ for all $x \in [c, d]$ and $f$ is continuous on $[c, d]$; therefore $f \in \mathscr{R}$ on $[c, d]$ and $$ \int_c^d f(x) \ \mathrm{d} x \geq \int_c^d \hat{0}(x) \ \mathrm{d} x = 0$$ by Theorem 6.12 (b) in Baby Rudin, where $\hat{0}$ denotes the zero function on $[c, d]$. That is, $$ \int_c^d f(x) \ \mathrm{d} x \geq 0 \tag{0}$$ for any points $c$ and $d$ such that $a \leq c \leq d \leq b$.

Suppose that there is a point $p \in [a, b]$ such that $f(p) > 0$. Then, as $f$ is continuous at $p$, so, for any real number $\varepsilon$ such that $$ 0 < \varepsilon < \frac{ f(p) }{2}, $$ we can find a real number $\delta > 0$ such that $$ \lvert f(x) - f(p) \rvert < \varepsilon < \frac{ f(p) }{2} $$ for all $x \in [a, b]$ for which $ \lvert x-p \rvert < \delta$.

Therefore, we can conclude that $$ \frac{ f(p) }{2} < f(x) < \frac{3 f(p) }{2} \tag{1} $$ for all $x \in I$, where $$I \colon= [a, b] \cap \left[ p- \frac{\delta}{2}, p+ \frac{\delta}{2} \right]. $$ Let's put $I \colon= [u, v]$. Then we see that $$ \begin{align} \int_a^b f(x) \ \mathrm{d} x &= \int_a^u f(x) \ \mathrm{d} x + \int_u^v f(x) \ \mathrm{d} x + \int_v^b f(x) \ \mathrm{d} x \\ & \qquad \qquad \mbox{ [ by an extension of Theorem 6.12 (c) in Baby Rudin ] } \\ &\geq \int_u^v f(x) \ \mathrm{d} x \qquad \mbox{ [ by (0) above ] } \\ &\geq \int_u^v \frac{f(p)}{2} \ \mathrm{d} x \qquad \mbox{ [ using (1) and Theorem 6.12 (b) in Baby Rudin ] } \\ &= \frac{f(p)}{2} (v-u) \\ &> 0, \qquad \mbox{ [ by (2) and our choice of $u$ and $v$ above ] } \end{align} $$ which contradicts our hypothesis that $\int_a^b f(x) \ \mathrm{d} x = 0$.

Is this proof sound enough in terms of its logic and rigor? If so, then is it also lucid enough in its presentation?

  • $\begingroup$ Your proof seems great. I would just suggest to avoid the introduction of $\epsilon$ which is not necessary. $\endgroup$ – mathcounterexamples.net Jul 26 '17 at 16:43
  • 1
    $\begingroup$ This should be the proof of the contrapositive, not a proof by contradiction. $\endgroup$ – egreg Jul 26 '17 at 17:40
  • $\begingroup$ @egreg The author of this proof is proceeding by contradiction; namely assuming that there is a point $p$ where $f(p)>0$, and then deducing a contradiction of what we assumed about the integral of $f$. $\endgroup$ – Alex Ortiz Jul 26 '17 at 18:52
  • 2
    $\begingroup$ @AOrtiz That's exactly what I objected to: the proof is unnecessarily by contradiction. If $f$ takes on a positive value, the integral is positive. $\endgroup$ – egreg Jul 26 '17 at 19:26

Your proof is fine. A little wordy and perhaps using more notation than you need, but it's correct.

I would have phrased it thusly:

Proof. Suppose $c\in [a,b]$ is such that $f(c)>0$. By continuity of $f$, there is a closed interval $[r,s]$ of $c$ where $f>f(c)/2>0$. By monotonicity of the integral, $$ \int_a^bf \ge \int_r^sf\ge \int_r^sf(c)/2>0, $$ which is a contradiction by what we assumed about $\int_a^b f$. $\qquad\square$


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.