Justifying the Assumption $\Delta_{h}f(z) \rightarrow f'(z), \Delta_{h}g(z) \rightarrow g'(z)$ uniformly on $\text{K} \subset \Omega$?

119 Views Asked by At

In "Micheal's E. Taylor's" Complex Analysis, i'm having trouble computing the Partial Derivatives, my initial approach leading up to where i'm stuck in is detailed in $(0)$. Specifically in $(4)$, i'm having trouble verifying my set-up of the partial derivatives is valid.

$(0)$

Let $\Omega \subset \mathbb{C}$, be open and assume $K \subset \Omega$ is compact, assume $f,g \in C(\Omega)$ and in $(1.2)$, the following assumption can be made:

$(0.1)$ $$\triangle_{h}f(z) \rightarrow f'(z), \triangle_{h}g(z) \rightarrow g'(z).$$, $$ \text{Remark} : (0.1), and \, (0.2)\,\text{are} \, \text{Uniformaly on K}$$ One has to show that in $(0.2)$

$(0.2)$ $$ \triangle_{h}(f(z)g(z)) = \triangle_{h}f'(z)g(z) + f(z)g'(z)$$

$$\text{Remark}$$

The $f(z)$, observed within $(0.1)$, is formally defined as the following in $(1.3)$, one can make subtle operations by exploiting the Lapalcian Opeator on the RHS side of $(1.)$

$(1.3)$

$$(1) \, \, \, \, \, \, \triangle{h}f(z) = \frac{1}{h}(f(z+h) - f(z))$$ $$(2) \, \, \, \, \, \, \nabla^2_{h}f(z) = \nabla^2_{h}(\frac{1}{h}(f(z+h) - f(z)))$$ $$(3) \, \, \, \, \, \, \nabla_{} \cdot \nabla_{h}f(z) =\nabla_{} \cdot \nabla_{h} (\frac{1}{h}(f(z+h) - f(z)))$$

$$\text{Lemma}:$$

To verify the Holomorphcity of $f$, one can use the Cauchy-Riemann Equations, formally developed: in $(1.4)$

$(1.4)$

$ \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, $Let $u, v: \left\{ {\left({x, y}\right) \in \mathbb{R^2} }\, \middle\vert \, {x+iy = z \in D }\right\} \to \mathbb{R}$ be two real-valued functions:

$$u(x,y) = Re(f(z))$$ $$u(x,y) = Im(f(z))$$

$ \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, $The $f$ is complex-differentiable in $\text{D}$ if and only if: $$(1): \quad {\partial u}/{\partial x} = {\partial v}/{\partial y}$$ $$(2): \quad {\partial u}/{\partial y} = - {\partial v}/{\partial x}$$

If the conditions are true then: $$f' \left({z}\right) = {\partial f}/{\partial x} \left({z}\right) = -i {\partial f}/{\partial y} \left({z}\right)$$

Applying our recent development in $(1.4)$, first we can denote the real and complex parts of our function in $(3.)$ as follows:

$(3.)$ $$ \nabla_{} \cdot \nabla_{h}f(v+iy)=\nabla_{} \cdot \nabla_{h} (\frac{1}{h}(f(z)+h) - f(z))$$

$$\nabla_{} \cdot \nabla_{h}f(z)=\nabla_{} \cdot \nabla_{h} (\frac{1}{h}(f(z)+h) - f(z))$$

$$\text{Remark}$$: $ \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, $The Real and Complex Parts of our function can be denoted as follows: $$\nabla_{} \cdot \nabla_{h}(\text{Re}(f(z))$$

$$\nabla_{} \cdot \nabla_{h}(\text{Im}(f(z))$$

$$\text{Lemma}$$

$ \, \, \, \, \, \, \, \, \, \, \, $$ \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, \, $Applying,$(1.)$ and $(2.)$ one should be able to observe the following in $(4)$

$(4)$

$$ \partial_{u}(\nabla_{} \cdot \nabla_{h}f(v+iy))= \partial_{v}(\nabla_{} \cdot \nabla_{h}f(v+iy))$$

$$ \partial_{u}(\nabla_{} \cdot \nabla_{h}f(v+iy))= -\partial_{v}(\nabla_{} \cdot \nabla_{h}f(v+iy))$$

$$\partial_{u}(\nabla_{} \cdot \nabla_{h} (\frac{1}{h}(f(u+iy)+h) - f(u+iy))) = \partial_{v}(\nabla_{} \cdot \nabla_{h} (\frac{1}{h}(f(v+iy)+h) - f(v+iy))) $$

$$\partial_{u}(\nabla_{} \cdot \nabla_{h} (\frac{1}{h}(f(u+iy)+h) - f(u+iy))) = -\partial_{v}(\nabla_{} \cdot \nabla_{h} (\frac{1}{h}(f(v+iy)+h) - f(v+iy))) $$

$$\partial_{u}( \partial_{\text{zzhh}} (\frac{1}{h}(f(u+iy)+h) - f(u+iy))) = \partial_{v}( \partial_{\text{zzhh}} (\frac{1}{h}(f(v+iy)+h) - f(v+iy)))$$

$$\partial_{u}( \partial_{zzhh} (\frac{1}{h}(f(u+iy)+h) - f(u+iy))) = -\partial_{\text{v}}( \partial_{\text{zzhh}} (\frac{1}{h}(f(v+iy)+h) - f(v+iy)))$$

1

There are 1 best solutions below

1
On BEST ANSWER

Claim For each $c$ in the domain of $f$ and each $\varepsilon>0$ there is a disc $B = B(c,\delta)$ such that $$\tag{1}\left|\frac{f(z)-f(w)}{z-w} - f'(c)\right| < \varepsilon$$ for every distinct $z,w\in B$.

Proof Begin by writing, for $z,w$ in a disc $B'$ with center $c$ $$f(z) - f(w) - f'(c)(z-w) = \int_{[z,w]} (f'(\xi)-f'(c))d\xi.$$ where $[z,w]$ is the straight line segment that joins $z$ and $w$. We can use this and the standard estimate${}^1$ to obtain $$\left|\frac{f(z)-f(w)}{z-w} - f'(c)\right| \leqslant |f'-f'(c)|_{[z,w]} \leqslant |f'-f'(c)|_{B'}$$ where the right hand side is the maximum of $f'-f'(c)$ over $B'$. Because $f'$ is continuous at $c$, it suffices we take $B'$ small enough to guarantee that $(1)$ holds.


$1$. By this I mean that $\left|\int_{\gamma} f dz\right|\leqslant L(\gamma)|f|_\gamma.$