This can be thought of as roughly the converse of Cauchy’s theorem. Interestingly, it’s common knowledge in analysis that ‘if it’s differentiable, then it’s continuous, and if it’s continuous, then it’s integrable’. However, Morera’s theorem intriguingly uses integration to determine the differentiability of a function, which is truly a remarkable theorem.
F(z):=∫z0zf(w)dw
Define the complex path integral of f from a fixed point z0∈R to any point z∈R as a function of z∈R as shown above. First, let’s check if it’s well-defined. Assuming the premise that ∫Cf(z)dz=0 for all closed paths C, regardless of which path w0:z→z0 is fixed,
∫z0zf(w)dw+∫w0f(u)du=0
it’s evident that integrating F(z) over any path from z0 to z always yields the same value. Thus, F is confirmed to be a function determined solely by the choice of z.
By the fundamental property of complex path integrals, since hF(z+h)−F(z)=h1∫zz+hf(w)dw,
===hF(z+h)−F(z)−f(z)h1∫zz+hf(w)dw−h1hf(z)h1∫zz+hf(w)dw−h1∫zz+hf(z)dwh1∫zz+h(f(w)−f(z))dw
Given that f is continuous, for a given ε>0,
∣h∣<δ⟹∣f(z+h)−f(z)∣<ε
there exists a δ that satisfies.
ML Lemma: For a positive number M that satisfies ∣f(z)∣≤M and the length &VariableDoubleVerticalBar;C&VariableDoubleVerticalBar; of L,
∫Cf(z)dz≤ML
By the ML Lemma,
hF(z+h)−F(z)−f(z)=∣h∣1∫zz+h(f(w)−f(z))dw<∣h∣1ε∣h∣=ε
Therefore,
f(z)=h→0limhF(z+h)−F(z)=F′(z)
Thus, f is the derivative of some functionF. In complex analysis, if it’s differentiable once, it’s infinitely differentiable; hence, if F is differentiable, so is f.
■
Osborne (1999). Complex variables and their applications: p92. ↩︎