2 Complex derivatives and analytic functions
2.1 Real linear maps versus complex linear maps
The right abstract way of understanding the differential of a function is as a linear approximation to a function locally. The key difference between real analysis and complex analysis is whether one uses real linear maps or complex linear maps.
Let \(\mathbb {K}\) be a field (for example \(\mathbb {K}= \mathbb {R}\) or \(\mathbb {K}= \mathbb {C}\)), and let \(V\) and \(W\) be vector spaces over \(\mathbb {K}\). A function \(L : V \to W\) is said to be \(\mathbb {K}\)-linear if
Such a function \(L\) is also called a \(\mathbb {K}\)-linear map (or a \(\mathbb {K}\)-linear transformation) between the spaces \(V\) and \(W\).
The complex plane \(\mathbb {C}\cong \mathbb {R}^2\) can be seen either as a 2-dimensional real vector space or as a \(1\)-dimensional complex vector space. In particular, it makes sense to talk about both \(\mathbb {R}\)-linear maps \(\mathbb {C}\to \mathbb {C}\) and \(\mathbb {C}\)-linear maps \(\mathbb {C}\to \mathbb {C}\).
More generally, any complex vector space can be seen as a real vector space (of twice the same dimension), and any complex linear map becomes a real linear map. The converse is not true! Let us elaborate on this in a simple example which will soon be seen to pertain to the difference of complex differentiability and real differentiability.
Remark: Identifying \(\mathbb {C}= \mathbb {R}^2\) (and choosing basis vectors \(1, \mathfrak {i}\in \mathbb {C}\) for \(\mathbb {C}\) seen as a 2-dimensional vector space), we see that an \(\mathbb {R}\)-linear map \(L : \mathbb {C}\to \mathbb {C}\) can be encoded in a \(2 \times 2\) matrix with real entries,
in such a way that
Remark: A \(\mathbb {C}\)-linear map \(L : \mathbb {C}\to \mathbb {C}\) can be encoded in a single complex number \(\lambda \in \mathbb {C}\) (or more pedantically, in a \(1 \times 1\) matrix \([\lambda ] \in \mathbb {C}^{1 \times 1}\)), in such a way that
We can write \(\lambda = \alpha + \mathfrak {i}\beta \), with \(\alpha = \Re \mathfrak {e}(\lambda ), \beta = \Im \mathfrak {m}(\lambda ) \in \mathbb {R}\). We can also write \(z = x + \mathfrak {i}y\) and obtain the expression
In other words, seen as a real-linear map, the complex multiplication by \(\lambda \) corresponds to the matrix
This clearly shows that not every real-linear map \(\mathbb {C}\to \mathbb {C}\) is complex linear. It also gives an explicit set of equations for the entries of the matrix of a real-linear map characterizing complex-linearity, which turn out to be very closely related to the Cauchy-Riemann equations.
Let \(L : \mathbb {C}\to \mathbb {C}\) be a \(\mathbb {R}\)-linear map represented in the basis \(1,\mathfrak {i}\) by the matrix \(M = \left[ \begin{array}{cc} a & b \\ c & d \end{array} \right] \in \mathbb {R}^{2 \times 2}\). Then the following are equivalent:
\(L\) is \(\mathbb {C}\)-linear;
\(b = - c\) and \(a = d\).
Clear from the above discussion.
2.2 Complex derivative
Let \(f \colon A \to \mathbb {C}\) be a complex-valued function defined on a subset \(A \subset \mathbb {C}\) of the complex plane, and let \(z_0 \in A\) be an interior point of the subset.
The \(f\) is said to have a complex derivative
at \(z_0\), if the limit on the right hand side above exists.
(In complex analysis we often drop the epithet “complex” above, and simply call \(f'(z_0)\) the derivative of \(f\) at \(z_0\).)
If a function \(f \colon A \to \mathbb {C}\) has complex derivative \(f'(z_0) = \lambda \in \mathbb {C}\) at a point \(z_0 \in A\), then we can write a linear approximation
where the error term \(\epsilon \) is small near \(z_0\) in the sense that \(\lim _{z \to z_0} \frac{\epsilon (z)}{|z-z_0|} = 0\).
…
If a function \(f \colon A \to \mathbb {C}\) has a complex derivative \(f'(z_0)\) at a point \(z_0 \in A\), then it is continuous at \(z_0\).
…
2.3 Cauchy-Riemann equations
Let \(f \colon A \to \mathbb {C}\) be a function defined on a set \(A \subset \mathbb {C}\), and let \(u \colon A \to \mathbb {R}\) and \(v \colon A \to \mathbb {R}\) be its real and imaginary parts, viewed as real-valued functions of two real variables, \(u(x,y) = \Re \mathfrak {e}\big( f (x + \mathfrak {i}y) \big)\) and \(v(x,y) = \Im \mathfrak {m}\big( f (x + \mathfrak {i}y) \big)\), so that \(f = u + \mathfrak {i}\, v\). If \(f\) has a complex derivative \(f'(z_0)\) at an interior point \(z_0 = x_0 + \mathfrak {i}y_0 \in A\), then \(u\) and \(v\) are differentiable at \((x_0,y_0)\) and their partial derivatives satisfy the Cauchy-Riemann equations
(These equations are equivalent to the differential \(\mathrm{d}f (x_0, y_0) \colon \mathbb {R}^2 \to \mathbb {R}^2\) being \(\mathbb {C}\)-linear when we identify \(\mathbb {R}^2 = \mathbb {C}\).)
We can then write the derivative at \(z_0\) in any of the following ways:
…
2.3.1 Differentiation rules
If two functions \(f, g \colon A \to \mathbb {C}\) have complex derivatives \(f'(z_0), g'(z_0)\) at a point \(z_0 \in A\), then the sum function \(f + g\) has a complex derivative at \(z_0\) given by
If a function \(f \colon A \to \mathbb {C}\) is has a complex derivative \(f'(z_0)\) at a point \(z_0 \in A\) and \(c \in \mathbb {C}\) is a complex number, then the function \(c f\) has complex derivative
at \(z_0\).
…
If two functions \(f, g \colon A \to \mathbb {C}\) have complex derivatives \(f'(z_0), g'(z_0)\) at a point \(z_0 \in A\), then the product function \(f g\) has complex derivative
at \(z_0\).
…
If two functions \(f, g \colon A \to \mathbb {C}\) have complex derivatives \(f'(z_0), g'(z_0)\) at a point \(z_0 \in A\) and \(g(z_0) \ne 0\), then the quotient function \(f / g\) has complex derivative
at \(z_0\).
…
If \(f \colon A \to B \subset \mathbb {C}\) is differentiable at \(z_0 \in A\) and \(g \colon B \to \mathbb {C}\) is differentiable at \(f(z_0) \in B\), then the composition \(g \circ f \colon A \to \mathbb {C}\) is differentiable at \(z_0\), with derivative
…
Suppose that \(f\) is a complex-valued function defined on a subset of the complex plane, which has a nonzero complex derivative \(f'(z_0) \ne 0\) at a point \(z_0\) and which has a local inverse function near \(z_0\) in the sense that there are open sets \(U, V \subset \mathbb {C}\) with \(z_0 \in U\) and \(f(z_0) \in V\), and the restriction of \(f\) to \(U\) is continuous \(U \to V\) with a continuous inverse. Then the local inverse function \(f^{-1} \colon V \to U\) has complex derivative at \(w_0 := f(z_0)\) given by
…
2.3.2 Analytic functions
A function \(f \colon U \to \mathbb {C}\) defined on an open set \(U \subset \mathbb {C}\) is said to be analytic (or holomorphic) if it is complex differentiable at every point \(z_0 \in U\).
Let \(f \colon U \to \mathbb {C}\) be a function defined on an open set \(U \subset \mathbb {C}\), and let \(u \colon U \to \mathbb {R}\) and \(v \colon U \to \mathbb {R}\) be its real and imaginary parts, viewed as real-valued functions of two real variables,
so that \(f = u + \mathfrak {i}\, v\).
Then the following are equivalent:
The functions \(u\) and \(v\) are differentiable at every point in \(U\) and their partial derivatives satisfy the Cauchy-Riemann equations
\begin{align*} \frac{\partial {u}}{\partial {x}} = \frac{\partial {v}}{\partial {y}} \quad \text{and} \quad \frac{\partial {u}}{\partial {y}} = - \frac{\partial {v}}{\partial {x}} \end{align*}in \(U\).
The function \(f\) is analytic.
…
Every function \(f \colon U \to \mathbb {C}\) which is analytic on an open set \(U \subset \mathbb {C}\) is also continuous on \(U\).
…
Every polynomial function \(p \colon \mathbb {C}\to \mathbb {C}\) is analytic.
…
Every rational function \(f \colon U \to \mathbb {C}\) is analytic on its domain of definition \(U \subset \mathbb {C}\).
…
The complex exponential function \(\exp \colon \mathbb {C}\to \mathbb {C}\) is analytic. Its (complex) derivative at \(z \in \mathbb {C}\) is \(\exp '(z) = \exp (z)\).
…
The principal branch of the \(n\)th root function \(z \mapsto \sqrt[n]{z}\) is analytic on its domain \(\mathbb {C}\setminus (-\infty ,0]\).
(Different branch choices can be made to obtain analyticity on other domains, but for \(n \ge 2\), no branch of \(\sqrt[n]{z}\) can be made analytic on all of \(\mathbb {C}\).)
…
2.3.3 Consequences of Cauchy-Riemann equations
Suppose that \(f \colon D \to \mathbb {C}\) is a analytic function on a connected open subset \(D \subset \mathbb {C}\) of the complex plane such that \(f'(z) = 0\) for all \(z \in D\). Then \(f\) is a constant function.
…
Suppose that \(f \colon D \to \mathbb {C}\) is a analytic function on a connected open subset \(D \subset \mathbb {C}\) of the complex plane. If any of the functions \(u = \Re \mathfrak {e}(f) \colon D \to \mathbb {R}\), \(v = \Im \mathfrak {m}(f) \colon D \to \mathbb {R}\), \(|f| \colon D \to \mathbb {R}\), is constant on \(D\), then \(f\) is itself a constant function.
…
Suppose that \(f \colon U \to \mathbb {C}\) is a analytic function on an open subset \(U \subset \mathbb {C}\) of the complex plane. Let \(u,v \colon U \to \mathbb {R}\) denote the real and imaginary parts of \(f\) defined by \(u(x,y) = \Re \mathfrak {e}\big( f(x + \mathfrak {i}y)\big)\) and \(v(x,y) = \Im \mathfrak {m}\big( f(x + \mathfrak {i}y)\big)\). Assume moreover that that \(u\) and \(v\) are twice continuously differentiable (later it will be shown that this assumption holds automatically by the analyticity of \(f\)). Then \(u\) and \(v\) are harmonic functions, i.e., they satisfy
…
Suppose that \(u \colon U \to \mathbb {R}\) is harmonic function on an open subset \(U \subset \mathbb {R}^2\), i.e., a twice continuously differentiable function satisfying \(\frac{\partial ^2}{\partial {x}^2} u + \frac{\partial ^2}{\partial {y}^2} u = 0\) on \(U\). A function \(v \colon U \to \mathbb {R}\) is called a harmonic conjugate of \(u\) if the function
is analytic on \(U\).
Let \(B = \mathcal{B}(z_0; r) \subset \mathbb {C}\) be a disk in the complex plane. Suppose that \(u \colon B \to \mathbb {R}\) is harmonic function on \(B\). Then a harmonic conjugate \(v \colon B \to \mathbb {R}\) of \(u\) in the disk \(B\) exists and is unique up to an additive constant.
…