A great deal of content borrowed from https://web.stanford.edu/~chriseur/notes_pdf/Eur_ComplexAnalysis_Notes.pdf
Some useful notation:
- \({\mathbb{D}}_r(a) \coloneqq\left\{{z\in {\mathbb{C}}{~\mathrel{\Big\vert}~}0\leq {\left\lvert {z-a} \right\rvert}< a}\right\}\) an open disc about \(a\)
- \(\mkern 1.5mu\overline{\mkern-1.5mu{\mathbb{D}}\mkern-1.5mu}\mkern 1.5mu_r(a) \coloneqq\left\{{z\in {\mathbb{C}}{~\mathrel{\Big\vert}~}0\leq {\left\lvert {z-a} \right\rvert} \leq a}\right\}\) a closed disc about \(a\).
- \({\mathbb{D}}^*(a) \coloneqq\left\{{z\in {\mathbb{C}}{~\mathrel{\Big\vert}~}0 < {\left\lvert {z-a} \right\rvert} < r}\right\}\) a punctured disc about \(a\).
- \(\Delta \coloneqq{\mathbb{D}}_1(0)\) the standard unit disc
- \(\mkern 1.5mu\overline{\mkern-1.5mu\Delta \mkern-1.5mu}\mkern 1.5mu\coloneqq\mkern 1.5mu\overline{\mkern-1.5mu{\mathbb{D}}\mkern-1.5mu}\mkern 1.5mu_1(0)\) the closed unit disc
- \(\Delta^* \coloneqq{\mathbb{D}}_1^*(0)\) the punctured unit disc.
- \(\Omega\) an open simply-connected subset of \({\mathbb{C}}\).
- \({\mathcal{O}}(\Omega), \mathop{\mathrm{Hol}}(\Omega), \mathop{\mathrm{Hol}}(\Omega, {\mathbb{C}})\) the holomorphic functions \(f:\Omega \to {\mathbb{C}}\) equipped with the structure of a \({\mathbb{C}}{\hbox{-}}\)algebra..
Greatest Hits
Things to know well:
- Estimates for derivatives
- Cauchy's theorem
- MVT for harmonic/holomorphic functions
- Cauchy's integral formula
- Cauchy's inequality
- Morera's theorem
- Liouville's theorem
- Rouche's theorem
- The Schwarz reflection principle
- The Schwarz lemma
- Casorati-Weierstrass
- Conformal maps
-
Automorphisms of the disc and plane
- The Cayley transformation and other Mobius transformations
- Conformal map questions
- The identity principle
- Picard theorems
- The open mapping theorem
- Computing residues
- Jordan’s lemma
- The Cauchy-Riemann equations
- The argument principle
- The Riemann mapping theorem
- Riemann's removable singularity theorem
For just the statements of most of these theorems: see this doc.
Common tricks
- Virtually any time: consider \(1/f(z)\) and \(f(1/z)\).
- Set \(w=e^z\).
- If \(f\) has no zeros, put it in the denominator! No one can stop you!
- If \(f\) is holomorphic in a neighborhood of \({\mathbb{D}}\) and \({\left\lvert {f} \right\rvert} = 1\) on \({{\partial}}{\mathbb{D}}\), then \(f\) is a finite Blaschke product.
- If \(\Omega\) is connected, \(f\) admits a log and exponential, so try setting \(f^{1\over n} = \exp\qty{{1\over n}\log(f)}\).
Holomorphic
-
To show a function is holomorphic,
- Use Morera’s theorem
- Find a primitive (sufficient but not necessary)
- Express \(f\) as a convergent power series
- Holomorphic functions have isolated zeros.
Arithmetic
Some silly arithmetic tricks:
- Absolutely essential: \({\left\lvert {f} \right\rvert}^2 = f\mkern 1.5mu\overline{\mkern-1.5muf\mkern-1.5mu}\mkern 1.5mu\).
- \(z\) is purely imaginary \(\iff \mkern 1.5mu\overline{\mkern-1.5muz\mkern-1.5mu}\mkern 1.5mu = -z\).
- \(z\in {\mathbb{R}}\iff \mkern 1.5mu\overline{\mkern-1.5muz\mkern-1.5mu}\mkern 1.5mu = z\).
- \(\log\qty{{\left\lvert {z} \right\rvert}} = {1\over 2}\log\qty{{\left\lvert {z} \right\rvert}^2} = {1\over 2}\log\qty{x^2 + y^2}\), which is easier to differentiate.
- To prove \(a=b\), try \(a/b = 1\) or \(a-b=0\).
- \(\int_0^{2\pi} e^{i(m-n)\theta}\,d\theta= \chi_{m=n}\cdot 2\pi\).
Showing a function is constant (or zero)
-
Show \(f' = 0\).
- Can write \(f=u+iv\) and show \(u_x, u_y = 0\) and apply CR.
- Show \({\left\lvert {f} \right\rvert}=0\) on the boundary and apply the MMP.
- Show that \(f\) attains a minimum or maximum on the interior of a domain where it is nonzero.
-
Show that \(f\) is entire and bounded.
- If you additionally want to show \(f\) is zero, show \(\lim_{z\to\infty} f(z) = 0\).
- Useful trick: show that either \({\left\lvert {f} \right\rvert} \geq M\) or \({\left\lvert {f} \right\rvert} \leq M\), then by Liouville on \(f\) or \(1/f\) respectively, \(f\) must be constant.
- Similar trick: show either \(e^f\) or \(e^{-f}\) is bounded.
- If the function is periodic, just bound it on a fundamental domain.
-
Show that \(f({\mathbb{C}})\) is not an open set (e.g. \({\mathbb{R}}\) or \({{\partial}}{\mathbb{D}}_r(0)\), and apply the open mapping theorem.
- More generally, the image can be dimension 0 or 2, but never 1.
- E.g. if \(\operatorname{im}(f) \subseteq {\mathbb{R}}\) or \({\left\lvert {f} \right\rvert} = R\) is constant.
-
A holomorphic function with a non-isolated zero is identically zero.
- How to use: show \(f-g\) has uncountably many zeros
-
Show that \(f\) omits at least 2 values and apply little Picard.
- E.g. if \(f\) misses an open set, or \({\left\lvert {f} \right\rvert} \geq M\) or \({\left\lvert {f} \right\rvert} \leq M\).
- Define \(g\coloneqq e^f\), then \({\left\lvert {g} \right\rvert} = e^{\Re(f)}\) and if \(g\) is constant then \(f\) is constant.
-
Show any of the following are constant:
- \(u = \Re(f)\)
- \(v = \Im(f)\)
- \({\left\lvert {f} \right\rvert}\)
- \(\operatorname{Arg}(f)\)
- Show that \(f\) preserves \({{\partial}}{\mathbb{D}}\), so \({\left\lvert {f(z)} \right\rvert} = 1\) when \({\left\lvert {z} \right\rvert} = 1\), and has no zeros in \({\mathbb{D}}\).
- To show \(f(z) = g(z)\) infinitely often, show \(f(z)/g(z)\) (or \(f(1/z)/g(1/z)\)) has an essential singularity and apply Picard or Casorati.
Singularities
-
Let \(z_0\) be a singularity of \(f\). To show \(z_0\) is…
- Removable: show that \(\lim_{z\to z_0} f(z)\) is bounded.
- A pole of some order: show \(\lim_{z\to z_o}f(z) = \infty\).
- A pole of order \(m\): write \(f(z) = (z-z_0)^mg(z)\) where \(g(z_0)\neq 0\) (or check the Laurent expansion directly).
- Can also check that \({\partial}_z^k f(z_0) \neq 0\) for \(k<m\) but \({\partial}_z^m f(z_0) = 0\).
- Essential: show that \(\lim_{z\to z_0} f(z)\) doesn’t exist (e.g. if it’s oscillatory).
- Alternatively, show \(z_0\) is neither removable nor a pole, or that \(f\) has a Laurent expansion about \(z_0\) with infinitely many negative terms.
- It can be useful to take a specific sequence \(\left\{{z_k}\right\}\to z_0\).
- \(f\) and \(f'\) have the same poles.
Zeros
- To show that a zero \(z_0\) is order \(n\), show that \(f^{(<n)}(z_0) = 0\) but \(f^{(n)}(z_0) \neq 0\).
- Getting rid of zeros: divide by a Blaschke product.
-
To count zeros:
- Rouche’s theorem
- The argument principle
-
If \(f(z_0)\neq 0\), by continuity there is some neighborhood where \(f\) is nonzero.
- Conversely, if \(f\) is holomorphic at \(z_0\) is a zero, there is punctured neighborhood of \(z_0\) where \(f\) is nonzero.
Estimating
- To prove \(a\leq b\), try showing \({a\over b} \leq 1\) and reason about \({\mathbb{D}}\), or show \(b-a\geq 0\),
- To bound a rational function, use the reverse triangle inequality: \begin{align*} {\left\lvert {a\pm b} \right\rvert} \geq {\left\lvert { {\left\lvert {a} \right\rvert} - {\left\lvert {b} \right\rvert}} \right\rvert} \implies {1\over {\left\lvert {a\pm b} \right\rvert}} \leq {1\over {\left\lvert {{\left\lvert {a} \right\rvert} - {\left\lvert {b} \right\rvert} } \right\rvert} } .\end{align*}
-
Bounding a derivative using the original function: Cauchy’s formula.
- Also works to bound a function in terms of its integral, e.g. over a compact set like a curve.
-
If \({\left\lvert {f} \right\rvert} = M\) on \({{\partial}}\Omega\), then if (importantly) \(f\neq 0\) in \(\Omega\) then \({\left\lvert {f} \right\rvert} = M\) on all of \(\mkern 1.5mu\overline{\mkern-1.5mu\Omega\mkern-1.5mu}\mkern 1.5mu\) by apply the MMP to \(f\) and \(1/f\).
- Why \(f\neq 0\) is necessary: take \(f(z) = z\).
- To show that a sequence of harmonic functions converge on e.g. a disc or rectangle, find good estimates on the boundary and apply the MMP.
- For real analysis: if \(f' < M\), apply the mean value theorem to show \(f\) is Lipschitz: \({\left\lvert {f(x) - f(y)} \right\rvert} = {\left\lvert {f'(\xi)} \right\rvert} {\left\lvert {x-y} \right\rvert} < M{\left\lvert {x-y} \right\rvert}\).
- To show \({\left\lvert {f} \right\rvert} \leq {\left\lvert {g} \right\rvert}\): if you have a factor of \(z\) to play with, try to apply Schwarz to \(f/g\) to get \({\left\lvert {f/g} \right\rvert}\leq {\left\lvert {z} \right\rvert}\).
Polynomials
-
\(f\) is polynomial when:
- \(f^{(n)} =0\) for every \(n\) large enough (e.g. using Cauchy’s inequality)
- \(f\) is entire and only has poles at \(\infty\).
Series
- A common trick: \begin{align*} \frac{1}{z-w}=\frac{1}{(z-a)\left(1-\frac{w-a}{z-a}\right)}=\sum_{n=0}^{n} \frac{(w-a)^{n}}{(z-a)^{n+1}} . \end{align*}