Commit 04ffa56d authored by Jan-Bernhard Kordaß's avatar Jan-Bernhard Kordaß
Browse files

Added first lecture on harmonic maps.

parent ac43f39b
\chapter{Tosion Invariants}
\chapter{Tosion Invariants [Roman Sauer]}
Torsion invariants fall into a class of so-called ``secondary invariants'' of topological spaces in the sense that they are only defined if a certain class of ``primary invariants'' (e.g. Betti numbers) vanish.
Often they reveal more subtle geometric information.
......@@ -156,6 +156,185 @@ Thus, this number only depends on the homotopy class of $f$.
Hence, the diagonal matrix entries of each $C_i^{CW}(g)$ are zero and thus $\Lambda(g) = \Lambda(f) = 0$.
\chapter{Harmonic Maps [Andy Sanders]}
\section{Basics of harmonic maps}
\subsection{Background differential geometry}
Let $E \to M$ be an $\R$-vector bundle over $M$ (second countable, hausdorff manifold) of rank $r$.
A \CmMark{connection} $\nabla$ on $E$ is an $\R$-linear map
\nabla \colon \Omega^0(E) \to \Omega^0(\T^{*}M \otimes_{\R} E) =: \Omega^1(M,E),
s \mapsto \nabla_{\blank} s
where $\Omega^0(E)$ denotes smooth sections in $E$, such that
\item $\nabla_{X+Y}s = \nabla_Xs + \nabla_Ys$,
\item $\nabla_X(s+s') = \nabla_X s + \nabla_Xs'$
\item $\nabla_{fX} s = f\nabla_Xs$
\item $\nabla_X(fs) = f\nabla_Xs + X(f) s$.
Let $q$ be an inner product on $E$.
We say that $\nabla$ is a \CmMark{metric connection} for $q$, if for all $s,t \in \Omega^0(E)$ we have
\dop q(s,t) = q(\nabla s,t) + q(s, \nabla t).
Let $(M,g)$ be a riemannian manifold with tangent bundle $E = \T M$ and Levi-Civita connection $\nabla$ of $g$.
Let $X,Y \in \Omega^0(M)$ be vector fields, i.e. $X = X^i \frac{\partial}{\partial x^i}$ and $Y = Y^j \frac{\partial}{\partial x^j}$ in local co-ordinates.
(Abbreviate $\partial_i$ for $\frac{\partial}{\partial x^i}$.)
\nabla_XY = \nabla_{X^i\partial_i} Y^i\partial_i
= X^i(\nabla_{\partial_i}Y^i\partial_i)
= X^i(\partial_iY^i\partial_i + Y^i\nabla_{\partial_i}\partial_i)
= X^i(\partial_iY^i\partial_i + Y^i\Gamma_{ij}^k\partial_i)
where $\Gamma_{ij}^k = g^{km}(\partial_ig_{im} + \partial_j g_{im} - \partial_mg_{ij})$ for $g_{ij} = g(\partial_i,\partial_j)$ and $g^{km}$ is the $km$-entry of $g^{-1}$.
Out of $E$ one can build another bundle $E^{*} = \Hom(E,\R)$ and given another vector bundle $F$, one can build $\Hom(E,F)$,
Let $(E,\nabla) \to M$ be a vector bundle with a connection over $M$.
The space of \CmMark{$p$-forms} on $m$ with values in $E$ is the $C^{\infty}(M)$-module $\Omega^p(M,E) = \Omega^0(M,\bigwedge^p\T^{*}M \otimes E)$.
Elements $\alpha$ in $\Omega^p(M,E)$ have representations as linear combination of $\alpha_{i_1,\cdots,i_p}\dop x^{i_1} \wedge \cdots \wedge \dop x^{i_p} \otimes (s_1, \cdots s_p)$.
The exterior covariant derivative is the map given by extension of
\dop^{\nabla} \colon \Omega^p(M,E) & \to \Omega^{p+1}(M,E),\\
\alpha \otimes u & \mapsto \dop^{\nabla}(\alpha \otimes u) = \dop \alpha \otimes u + (-1)^p \alpha \wedge \nabla u
for $\alpha \in \bigwedge^p\T^{*}M$, $u \in \Omega^0(E)$.
We want to define an inner product on $\Omega^p(M,E)$.
For this, fix a metric $g$ on $M$ and let $(E,\nabla,q) \to M$ be a vector bundle with metric and connection over $M$.
\left< \alpha \otimes u, p \otimes v\right>
= \int_M g(\alpha,p) q(u,v) \dop v_g
is a number.
(For this integral to be finite, assume $M$ is compact or work with compactly supported sections.)
The \CmMark{exterior covariant codifferential}\footnote{non-standard notation} is the formal $L^2$-adjoint of $d$
\delta^{\nabla} \colon \Omega^p(M,E) \to \Omega^p(M,E)
such that $\left< \dop^{\nabla}(\alpha \otimes u), \beta \otimes v\right> = \left<\alpha \otimes u, \delta^{\nabla}(\beta \otimes v)\right>$.
An integration by parts arguement shows that $\delta^{\nabla}$ exists and, when $\nabla$ is a metric connection, then
\delta^{\nabla} \colon \Omega^1(M,E) \to \Omega^0(M,E),
\alpha \otimes u \mapsto -\tr_g(\nabla^{\T^{*} \otimes E} \alpha \otimes u),
where for $\Omega^1(M,E) \to \Omega^0(M, \T^{*}M \otimes \T^{*}M \otimes E)$, we can take a trace with the metric by choosing an orthonormal basis.
A \CmMark{harmonic $p$-form} with values in $E$ is an element $\omega_i \in \Omega^p(M,E)$ such that $\delta^{\nabla} = \delta^{\nabla} \omega = 0$.
As a matter of fact this is equivalent to $\Delta \omega = 0$ for $\Delta := \delta^{\nabla} \circ \dop^{\nabla} + \dop^{\nabla} \circ \delta^{\nabla}$ (Consider $\left<\Delta \omega, \omega\right>$ and utilize the obvious stuff).
\subsection{Definition of harmonic maps of 1st variation formula}
Let $(M,g)$ and $(N,h)$ be two riemannian manifolds and let $f \colon M \to N$ be a smooth map.
Then $\dop f \colon \T M \to \T N$ is an element $\dop f \in \Omega^0(\Hom(\T M, f^{*}\T N)) = \Omega^0(\T^{*}M \otimes f^{*}\T N)$.
Next, the metrics $g,h$ induce a metric on $\T^{*}M \otimes f^{*}\T N$.
The energy density of $f \colon M \to N$ is $e(f) := \frac{1}{2} \left< \dop f, \dop f\right>_{\T^{*}M \otimes f^{*}\T N} = \frac{1}{2} \|\dop f\|^2$.
Choose co-ordinates $\{x^i\}$ in $M$ and $\{y^i\}$ in $N$.
With respect to these, we have
\frac{1}{2} \|\dop f \|^2 = \frac{1}{2}y^{ij} \partial_if^{*}\partial_jf^{\beta}h_{\alpha\beta}(f).
The \CmMark{Dirlichlet energy} is given by
E \colon C^2_0(M,N) \to \R,
f \mapsto \int_M e(f) \dop V_g.
A \CmMark{critical map} (or \CmMark{stationary map}) is a map $f \colon M \to N$ such that for all compactly supported $F \colon M \times (-\varepsilon, \varepsilon) \to N$ $C^2$-map (variation of $f$) with $F(x,0) = f(x)$ we have that
\delta E(\nu) := \left.\frac{\dop}{\dop t} E(F) \right|_{t = 0} = 0
for $\nu = \frac{\dop}{\dop t} F|_{t = 0} \in \Omega^0(f^{*}\T N)$.
The \cref{eq:first-variation} is called \CmMark{first variation in the direction of $\nu$}.
The map $f \colon (M,g) \to (N,h)$ is called \CmMark{harmonic}, if it is a critical point for the Dirlichlet energy.
Let $\dop f \in \Omega^1(M, f^{*}\T N)$ then $\nabla \dop f \in \Omega^0(M, \T^{*}M \otimes \T^{*}M \otimes E)$.
The \CmMark{second fundamental form} of $f$ is $\nabla \dop f := B_f$, which is a symmetric $2$-tensor on $M$.
The \CmMark{tension field} of $f$ is the trace of $B_f$: $\tau(f) := \tr_g(B_f) \in \Omega^0(M,f^{*}\T N)$.
\begin{thm*}[1st variation of $E$]
Let $F \colon M \times (\varepsilon, \varepsilon) \to N$ a variation of $f$ and let $\nu = \frac{\dop}{\dop t}F|_{t = 0}$.
\delta E(\nu) = \frac{\dop}{\dop t}E(F)|_{t = 0} = - \int_M \left<\tau(f), \nu\right> \dop v_g.
The variation $F \colon M \times (-\varepsilon,\varepsilon) \to N$ yields a pullback connection on $F^{*}\T N$, which shows
\frac{\dop}{\dop t}E(F)|_{t = 0} & = \frac{1}{2} \int_M \frac{\dop}{\dop t}\left<\dop F, \dop F\right> \dop V_g|_{t = 0}
= \int_M \left<\nabla_{\frac{\partial}{\partial t}}\dop F, \dop F\right> \dop V_g|_{t = 0}\\
& = \int_M \left<\nabla^{f^{*}\T N}\nu, \dop f\right> \dop V_g
\overset{(*)}{=} \int_M \left<\nu, \delta^{\nabla^{f^{*}\T N}} \dop f\right> \dop V_g\\
& = -\int_M \left< \nu, \tr_g(\nabla \dop f) \right> \dop V_g
= - \int_M \left< \nu, \tau(f)\right> \dop V_g,
where $(*)$ follows by a calculation in local co-ordinates.
\begin{cor*}[Fundamental theorem of the calculus of variations]
A $C^2$-map $f \colon (M,g) \to (N,h)$ is harmonic if and only if $\tau(f) = 0$.
What does $\tau(f) = 0$ look like?
Fix local co-ordinates $\{x^i\}$ on $M$ and $\{y^j\}$ on $N$.
Then $\dop f = \partial_if^{alpha} \dop x^i \otimes \frac{\partial}{\partial y^{\alpha}}$ and thus
\nabla \dop f
& = \nabla \partial_i f^{\alpha} \dop x^i \otimes \frac{\partial}{\partial y^{\alpha}}
= \partial_j\partial_i f^{\alpha} \dop x^j \otimes \dop x^i \otimes \frac{\partial}{\partial y^{\alpha}}
+ \partial_if^{\alpha} \nabla \dop x^i \otimes \frac{\partial}{\partial y^{\alpha}}\\
& = A + \partial_i f^{\alpha}(\nabla \dop x^i \otimes \frac{\partial}{\partial y^{\alpha}} + \dop x^i \otimes \nabla \frac{\partial}{\partial y^{\alpha}})\\
& = A + \partial_i f^{\alpha}( -\Gamma^i_{jk} \dop x^i \otimes \dop x^k \otimes \frac{\partial}{\partial y^{\alpha}} + \dop x^i \partial_j f^{\beta} \Gamma^{\gamma}_{\alpha\beta} \frac{\partial}{\partial y^{\gamma}})\\
& = \partial_i \partial_jf^{\gamma} \Gamma_{ij}^k \partial_k f^{\gamma} + \Gamma_{\alpha\beta}^{\gamma}(f) \partial_jf^{\alpha}\partial_if^{\beta)} \dop x^i \otimes \dop x^j \otimes \frac{\partial}{\partial y^j}.
Thus $\tau(f) = (\Delta_gf^{\gamma} + \Gamma_{\alpha\beta}^{\gamma}(f) \partial_if^{\alpha}\partial_jf^{\beta}g^{ij})$.
%%% Local Variables:
%%% mode: latex
%%% TeX-master: "skript-rtg-lectures-ws1617"
% \includegraphics[scale=0.45]{kit-logo.jpg}
% \hfill
......@@ -7,14 +7,16 @@
{\Large \textsc{Notes}}\\[0.8cm]
{\huge \textsc{RTG Lectures}}\\[0.8cm]
\textsc{Prof. Dr. R. Sauer and Dr. A. Sanders}
{\huge \textsc{RTG Lectures}}\\[1.2cm]
{\Large \textsc{Torsion Invariants}}\\
\textsc{Prof. Dr. R. Sauer}\\[1.2cm]
{\Large\textsc{Harmonic maps}}\\
\textsc{Dr. A. Sanders}
\textsc{Winter 2016/17}\\[0.5cm]
\textsc{Torsion Invariants | Harmonic maps}\\[0.7cm]
\textsc{Winter 2016/17}
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment