## Holomorphic functions #1

When we do analysis with real-valued functions of a single variable, differentiability at a point $a \in \mathbb{R}$ says that there must exist a linear map (often called a “linearization” or “linear approximation”) $L$ such that the limit $\frac{f(a+\Delta x) - f(a) - L_a(\Delta x)}{\Delta x} = 0$ as $\Delta x \to 0$. So we can basically see that this map, when supplied with $\Delta x$, gives us a (loosely speaking) “close” estimate of the difference between $f(a+\Delta x)$ and $f(a)$. In this single-variable estimation, this linear map is represented by a $1 \times 1$ matrix, in other words a real number, and is denoted $f'(a)$. Continuing with this idea, a function $f : \mathbb{R}^n \to \mathbb{R}^m$ is differentiable if there is a linear map (this time from $\mathbb{R}^n \to \mathbb{R}^m$) which satisfies a similar limit. This linear map turns out to be the left-multiplication transformation associated to an $m \times n$ matrix called the Jacobian and if $m=1$, that is, if our function is real-valued, it is a row matrix, which is the function’s so-called “gradient” $\nabla f$ (in this way, a Jacobian can be thought of just as a matrix whose rows are the gradients of the component functions $f^{(i)}$ for $1 \leq i \leq m$).

In complex analysis, new functions are examined; namely functions $f : D \to \mathbb{C}$. For our purposes the set $D$ is a nonempty, open, path-connected subset of $\mathbb{C}$ (these are often referred to as domains). Intuitively then, such a function $f$ can be thought of as a transformation of a subset of the complex plane. Since two quantities (usually the modulus and the argument of the resulting complex number) are associated to each point in $D$ by this function, the graph is 4-dimensional (as opposed to the 2-dimensional graphs of single-variable analysis), therefore it is quite difficult to represent it properly without resorting to 2D color maps, and so on. Often, an image of a grid in the complex plane is shown, and then another diagram shows the same grid after it is transformed by the function. We first say what it means for such a function to be differentiable at a point $z_0 = x_0 + iy_0 \in D$.

Definition. Suppose $f : D \to \mathbb{C}$, given by $f(z) = u(x,y) + iv(x,y)$, is a complex-valued function of a complex variable $z=x+iy$. ($u$ and $v$ are the real and imaginary parts of $f$, respectively). Then we say $f$ is differentiable at $z_0 = x_0 + iy_0 \in D$ if the limit of $\frac{f(z+\Delta z) - f(z)}{\Delta z}$ exists as $\Delta z \to 0$.

Definition. If $f$ is differentiable on all of $D$ it is said to be holomorphic on $D$. If it is differentiable in some neighbourhood of $z_0$ it is said to be holomorphic at $z_0$.

So in any case we see that holomorphicity is a stronger condition than differentiability, as it requires differentiability not only at a given point, but at least on some neighbourhood around that point. Some authors use the term analytic or regular rather than holomorphic, but I disagree with this. The word “analytic” should be reserved for functions which have power series expansions on that set/neighbourhood.

Now that we know what it takes for a function to be complex-differentiable, we now ask whether there is some connection with the more familiar world of real-valued functions. There is: a certain condition on the functions $u$ and $v$ turns out to be linked to complex differentiability. These are called the Cauchy-Riemann equations, and I will speak about this in a follow-up post. I am currently reviewing a few notions of differentiability from multivariable calculus that are needed.

just another guy trying to make the diagrams commute.
This entry was posted in analysis and tagged , , , , . Bookmark the permalink.

### 2 Responses to Holomorphic functions #1

1. crobert says:

“The word “analytic” should be reserved for functions which have power series expansions on that set/neighbourhood.”

…but every holomorphic function does have a power series (locally).

2. dx7hymcxnpq says:

Exactly, but I think it’s good to keep two separate names for the two properties, which (nontrivially) happen to imply one another. At least while you’re talking about elementary complex analysis – since after that you can probably assume everyone knows what’s going on. lol