How do I evaluate $\lim_{\Delta z \to 0} \frac{1}{\Delta z}\int_{z_1}^{z_1 + \Delta z}f(t)dt$ in order to finish this proof?

Given $f(z)$ is a continuous function on domain $D$, and given that for any contour $C$ connecting $z_1,z_2\in D$

$$ \int_C f(z)dz = \int_{z_1}^{z_2}f(t)dt = F(z_2) - F(z_2) $$

Then $\forall z\in D$, $F'(z) = f(z)$

My work

1.) Take any two points $z_1, z_2 \in D$. Let $z_2 = z_1 +\Delta z$.

2.) Fix any countour $C$ connecting $z_1$ and $z_2$

Then by hypothesis

$$ \int^{z_1 + \Delta z}_{z_1} f(t)dt = F(z_1 + \Delta z) - F(z_1) $$


$$ \frac{1}{\Delta z}\int^{z_1 + \Delta z}_{z_1} f(t)dt = \frac{1}{\Delta z}[F(z_1 + \Delta z) - F(z_1)] $$

which gives the difference quotient on the RHS. Therefore

$$ \lim_{\Delta z \to 0} \frac{1}{\Delta z}\int^{z_1 + \Delta z}_{z_1} f(t)dt = F'(z_1) $$

end of my work

It looks like I'm really close. If this were a real valued function, I could use the IVT for integrals to finish this proof. But since we haven't proved the complex analog of the IVT in my class, I'm unsure as how to make this final leap.

edit: my professor says I can "temporarily fix $t$" in the last line and therefore move $f(t)$ outside of the integrand... Which I see yields the desired result. But I'm pretty skeptical because he couldn't justify why fixing what is already a dummy variable is allowed.

edit 2: The source of this question comes from page 142 of Complex Variables and Applications 8th edition by Brown and Churchill. 3 statements are made and claimed to be equivalent. They show a implies b, b implies c, and c implies a. The proof in this question is me trying to prove b implies a, which is not done in the textbook.


Beware that your function $F$ is not uniquely defined if $f$ is not holomorphic! If $f$ was holomorphic, you could simply use the fact that $F'=f$ to conclude.

To get the continuous case, you could also start by using the continuity of $f$ at $z_1$. Indeed, $$\forall \varepsilon>0,\exists \eta>0, \forall z \in \mathbb C, \quad |z-z_1|<\eta \Rightarrow |f(z)-f(z_1)|<\varepsilon $$ Let $\varepsilon>0$, let such a $eta>0$ ; for all $|\Delta z |< \eta$, $$\left| \frac{1}{\Delta z} \int_{z_1}^{z_1+\Delta z} f(z)\mathrm d z -f(z_1)\right| = \left| \frac{1}{\Delta z} \int_{z_1}^{z_1+\Delta z} (f(z)-f(z_1))\mathrm d z\right| \leq \left|\frac{1}{\Delta z}\int_{z_1}^{z_1+\Delta z_1} \eta\right| = \eta $$ Thus $$\lim_{\Delta z\rightarrow 0} \left|\frac{1}{\Delta z} \int_{z_1}^{z_1+\Delta z} f(z)\mathrm d z -f(z_1)\right| =0$$ and thus $$\lim_{\Delta z\rightarrow 0} \frac{1}{\Delta z} \int_{z_1}^{z_1+\Delta z} f(z)\mathrm d z =f(z_1)$$

| cite | improve this answer | |
  • $\begingroup$ I cannot see how the first equality holds. You brought a constant $f(z_1)$ which was previously subtracted from the integrand on the LHS to be included in the integrand on the RHS. From that step on, I understand completely! Can you just clarify why the equality holds. Thanks! $\endgroup$ – rocksNwaves Oct 31 '18 at 18:03
  • $\begingroup$ I figured out why the equality holds. I just had to play with it. $\endgroup$ – rocksNwaves Nov 1 '18 at 13:54

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.