References:
- topology: wikipedia
- complex analysis: Evan Chen, The Napkin
topology
compact: every open cover has a finite sub-cover.
complete: limits of cauchy sequences exist
open set (in a metric space): \(\forall x\in U, \exists \varepsilon>0\) such that \(\left\{ y\; : \;d(x,y)<\varepsilon \right\}\subseteq U\).
discrete topology: all sets are open
- is compact iff finite
product topology: minimal amount of open sets to make projections \(\pi_i : \prod_j X_j \mapsto X_i\) continuous for each \(i\).
- continuous: \(f^{-1}(\text{open set}) =\text{open set}\)
open sets in the product topology are unions of sets of the form \[\prod_i U_i\]
- where \(U_i\) is open in \(X_i\) and \(U_i\neq X_i\) for only finitely many \(i\).
Tychonoff’s theorem: product of compact spaces is compact
complex analysis
Derivative: \[\lim_{h\to 0} \frac{f(x+h)-f(x)}{h},\] if it exists, which is a high bar because \(h\to 0\) means complex limit.
holomorphic / entire: complex differentiable on domain / all of \(\mathbb{C}\).
ex: polynomials, \(\exp\), nice compositions of holomorphic functions.
Countour integrals
\[\oint_\alpha f(z) dz = \int_a^{b} f(\alpha(t)) \alpha'(t) dt\]
Theorem. Let \(\gamma\) parameterize the unit circle. Then \[\oint_\gamma \frac{1}{z} dz = 2\pi i.\]
For any integer \(m\neq -1\), \[\oint_\gamma z^{m} dz = 0.\]
Theorem. \(\gamma\) a loop, \(f\) holomorphic
\[\oint_\gamma f(z) dz = 0.\]
Another way of saying this is, if you have two different curves with the same endpoints then the contour integrals will be the same.
Theorem. Let \(D\) be a disk, and \(\gamma\) parameterize the boundary \(t\mapsto R\exp(it)\). Let \(f\) be holomorphic. For any \(a\) in the interior of \(D\) we have
\[f(a) = \frac{1}{2\pi i} \oint_\gamma \frac{f(z)}{z-a} dz.\]
Theorem. Over a disk, a holomorphic function is given exactly by its Taylor Series.
Some important results:
Theorem. Let \(f\) be entire, and suppose \(|f(z)|< 1000\) for all \(z\). Then \(f\) is the constant function.
memomorphic functions
- holomorphic: proto-ex: polynomials
- memomorphic: proto-ex: rational functions
memomorphic functions are basically holomorphic everywhere except for a couple of “poles”.
Definition.
memomorphic function: not identically zero on any open neighborhood.
so basically just has a bunch of isolated poles.
Another way of defining memomorphic functions is in terms of Laurent Series:
For a function to be memormophic it should be holomorphic in neighborhoods of all points except for a “small” number of poles. (this is allowed to be infinite, e.g., like \(1/\sin(z)\) is memomorphic) but the poles must have non-intersecting open balls around them). If \(a\) is a pole then it is required to, in some neighborhood of \(a\) admit a series expansion \[f(z) = \frac{c_{-m}}{(z-a)^{m}} + \cdots + c_0 + c_1 (z-a) + \cdots .\]
So it’s like a Taylor series but we can have some pole crap at the front.
In the expression \(\frac{c}{z^{m}}\) \(c\) is callled the residue, \(m\) the order of the pole (simple pole is \(m=1\))
Definition.
Winding number: Let \(\gamma\) be a loop.
\[\frac{1}{2\pi i}\oint_\gamma \frac{1}{z-p} dz\]
Theorem.
\[\frac{1}{2\pi i}\oint_\gamma f(z) dz = \sum_{\text{pole } p} \text{winding}(\gamma, p) Res(f; p)\]