ncg

bachelorthesis in physics
git clone git://popovic.xyz/ncg.git
Log | Files | Refs

commit a4d2f621591e31344cda27e813e358f87b908ab8
parent c708e7589668f5f78a3704d70d0776d8b65a494a
Author: miksa <milutin@popovic.xyz>
Date:   Wed, 28 Apr 2021 13:19:05 +0200

almost done week7

Diffstat:
Mpdfs/week5.pdf | 0
Apdfs/week7.pdf | 0
Msrc/week5.tex | 9+++++----
Asrc/week7.tex | 577+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
4 files changed, 582 insertions(+), 4 deletions(-)

diff --git a/pdfs/week5.pdf b/pdfs/week5.pdf Binary files differ. diff --git a/pdfs/week7.pdf b/pdfs/week7.pdf Binary files differ. diff --git a/src/week5.tex b/src/week5.tex @@ -327,10 +327,11 @@ Then there is a map $d:A\rightarrow \Omega _D ^1 (A)$, $d = [D, \cdot]$. Begin \begin{align*} - a(a_k[D, b_k])b &= a_k'(Db_k - b_k D) b = \\ - &= a_k'(Db_k b - b_k D b) = a_k(Db_k b - b_k Db -Db_k +Db_kb)=\\ - &= a_k'([D, b_kb] - b_k D b + D b_k + \cdots - \cdots) = \\ - &=\sum_k a_k' [D, b_k'] + a(a_k[D, b_k])b &= aa_k(Db_k - b_k D) b = \\ + &= aa_k(Db_k b - b_k D b) = aa_k(Db_k b - b_k Db - b_kbD +b_kbD)=\\ + &= aa_k(Db_kb-b_kbD + b_k b D - b_k D b) = \\ + &= aa_k [D, b_kb] + aa_k b [D, b]=\\ + &= \sum _k a_k' [D, b_k'] \end{align*} \end{MyExercise} diff --git a/src/week7.tex b/src/week7.tex @@ -0,0 +1,577 @@ +\documentclass[a4paper]{article} + +\usepackage[T1]{fontenc} +\usepackage[utf8]{inputenc} + +\usepackage{mathptmx} + +\usepackage{subcaption} +\usepackage[shortlabels]{enumitem} +\usepackage{amsmath,amssymb} +\usepackage{amsthm} +\usepackage{bbm} +\usepackage{graphicx} +\usepackage[colorlinks=true,naturalnames=true,plainpages=false,pdfpagelabels=true]{hyperref} +\usepackage[parfill]{parskip} + +\usepackage{pst-node} +\usepackage{tikz-cd} + +\usepackage{tikz} +\usetikzlibrary{patterns,decorations.pathmorphing,positioning} + +\usepackage[framemethod=TikZ]{mdframed} + +\tikzstyle{titlered} = + [draw=black, thick, fill=white,% + text=black, rectangle, + right, minimum height=.7cm] + +\newcounter{exercise} + +\renewcommand*\theexercise{Exercise~\arabic{exercise}} + +\makeatletter +\mdfdefinestyle{exercisestyle}{% + outerlinewidth=1em,% + outerlinecolor=white,% + leftmargin=-1em,% + rightmargin=-1em,% + middlelinewidth=1.2pt,% + roundcorner=5pt,% + linecolor=black,% + backgroundcolor=blue!5, + innertopmargin=1.2\baselineskip, + skipabove={\dimexpr0.5\baselineskip+\topskip\relax}, + skipbelow={-1em}, + needspace=3\baselineskip, + frametitlefont=\sffamily\bfseries, + settings={\global\stepcounter{exercise}}, + singleextra={% + \node[titlered,xshift=1cm] at (P-|O) % + {~\mdf@frametitlefont{\theexercise}~};},% + firstextra={% + \node[titlered,xshift=1cm] at (P-|O) % + {~\mdf@frametitlefont{\theexercise}~};}, +} +\makeatother + +\newenvironment{MyExercise}% +{\begin{mdframed}[style=exercisestyle]}{\end{mdframed}} + +\theoremstyle{definition} +\newtheorem{definition}{Definition} + +\theoremstyle{definition} +\newtheorem{question}{Question} + +\theoremstyle{definition} +\newtheorem{example}{Example} + +\theoremstyle{theorem} +\newtheorem{theorem}{Theorem} + +\theoremstyle{theorem} +\newtheorem{lemma}{Lemma} + +\theoremstyle{theorem} +\newtheorem{proposition}{Proposition} + +\newtheorem*{idea}{Proof Idea} + + +\title{Notes on \\ Noncommutative Geometry and Particle Physics} +\author{Popovic Milutin} +\date{Week 7: 23.04 - 27.04} + +\begin{document} + +\maketitle +\tableofcontents + +\section{Classification of Finite Real Spectral Triples} + +Here we classify finite real spectral triples modulo unitary equivalence with +\textit{Krajewski Diagrams}. We extend $\Lambda$-decorated graphs to the case of +real spectral triples (grading and real structure). + +\textbf{The Algebra:}Like before: +\begin{align} + A\simeq \bigoplus_{i=1}^N M_{n_i}(\mathbb{C}) \;\;\;\;\;\;\; \text{with} \;\;\; \hat{A} = \{\textbf{n}_1, \dots, \textbf{n}_N\} +\end{align} +Where $\textbf{n}_i$ are irreducible representation of $A$ on +$\mathbb{C}^{n_i}$ + +\textbf{The Hilbertspace:}Faithful irreducible representation on $A$ are the +direct sum of $\mathbb{C}^{n_i}$'s, which act on $A$ by left block-diagonal +matrix multiplication. +\begin{align} + \bigoplus_{i=1}^N \mathbb{C}^{n_i} +\end{align} +Furthermore we need a representation of $A^\circ$ on $H$ that commutes with +$A$. That is +\begin{align} + A^\circ \simeq &\bigoplus_{i=1}^N M_{n_i}(\mathbb{C})^\circ \\ + \text{with} \;\;\; &\hat{A}^\circ = \{\textbf{n}_1^\circ, \dots, + \textbf{n}_N^\circ\} \\ + \text{and} \;\;\; &\bigoplus_{i=1}^N \mathbb{C}^{n_i\circ} +\end{align} +And we need the multiplicity space $V_{ij}$ of $\mathbb{C}^{n_i} \otimes +\mathbb{C}^{n_j\circ}$. +Thus making the Hilbertspace: +\begin{align} + H=\bigoplus_{i,j=1}^N \mathbb{C}^{n_i} \otimes \mathbb{C}^{n_j\circ} + \otimes V_{ij} +\end{align} +\begin{itemize} + \item $\textbf{n}_i$, $\textbf{n}_j^\circ$ form a grid + \item if there is a node at $(\textbf{n}_i$, $\textbf{n}_j^\circ)$ then + $\mathbb{C}^{n_i} \otimes \mathbb{C}^{n_j\circ}$ is nonzero in $H$. + \item multiplicity implies multiple nodes +\end{itemize} + +\begin{example} + $A = \mathbb{C} \oplus M_2 (\mathbb{C})$, two options of the Hilbertspace. + \begin{figure}[h!] \centering + \begin{tikzpicture}[ + dot/.style = {draw, circle, inner sep=0.06cm}, + no/.style = {}, + ] + \node[no](a) at (0,0) [label=left:$\textbf{1}^\circ$] {}; + \node[no](b) at (0, -1) [label=left:$\textbf{2}^\circ$] {}; + \node[no](c) at (1, 0.5) [label=above:$\textbf{1}$] {}; + \node[no](d) at (2, 0.5) [label=above:$\textbf{2}$] {}; + \node[dot](d0) at (1,0) [] {}; + \node[dot](d0) at (2,-1) [] {}; + + \node[no](a1) at (6,0) [label=left:$\textbf{1}^\circ$] {}; + \node[no](b2) at (6, -1) [label=left:$\textbf{2}^\circ$] {}; + \node[no](c2) at (7, 0.5) [label=above:$\textbf{1}$] {}; + \node[no](d2) at (8, 0.5) [label=above:$\textbf{2}$] {}; + \node[dot](d0) at (7,0) [] {}; + \node[dot](d0) at (8,0) [] {}; + + + \end{tikzpicture} + \end{figure} + +The first diagram corresponds to $H_1 = \mathbb{C} \oplus M_2(\mathbb{C})$, +to the second $H_2 = \mathbb{C} \oplus \mathbb{C}^2$. +\end{example} + +\begin{MyExercise} + \textbf{Let $J$ be an anti-unitary operator on a finite-dimensional Hilbert space. + Show that $J^2$ is an unitary operator + } \newline + + Straight forward, say $J:\; H \rightarrow H$, then let $\xi_1, \xi_2 \in + H$: + + \centering + \begin{align} + <J^2 \xi_1, J^2 \xi_2> &= <J(J\xi_1), J(J\xi_2)> =\\ + &= <J\xi_2, J\xi_1> = <\xi_1, \xi_2> + \end{align} +\end{MyExercise} + +\textbf{The real Structure:} $J:\; H \rightarrow H$. +\begin{lemma} + \label{lemma} + Let $J$ be an anti-unitary operator on a finite-dimensional Hilbertspace + $H$ with $J^2 = \pm 1 $ + \begin{enumerate} + \item If $J^2 = 1 \;\; \Rightarrow \;\; \exists$ an ONB $\{e_k\}$ of $H$\\ + with $Je_k = e_k$. + \item If $J^2 = -1 \;\; \Rightarrow \;\; \exists$ an ONB $\{e_k, f_k\}$ of $H$\\ + with $Je_k = f_k$ and consequently $Jf_k = -e_k$. + \end{enumerate} +\end{lemma} +\begin{proof} + \textbf{1.} $J^2 = 1$\newline + + $v\in H$ and set: + \begin{align} + e_1 := + \begin{cases} + c (v + Jv)\;\;\; \text{if}\;\;\; Jv \neq -v \\ + iv\;\;\; \text{if}\;\;\; Jv = -v + \end{cases} + \end{align} + Where $c$ is a normalization constant, then take $Je_1$ + \begin{align} + &J(v + Jv) = Jv + J^2v= v + Jv \;\;\;\; \text{and} \\ + &J(iv) = -iJv = iv\\ + &\Rightarrow Je_1 = e_1 + \end{align} + Take $v'\perp e_1$ making: + \begin{align} + <e_1 , Jv'> = <J^2 v', Je_1> = <v' , Je_1>= <v', e_1> =0 + \end{align} + Construct $e_2 \perp e_1$ with $v'$: + \begin{align} + e_2 := + \begin{cases} + c (v' + Jv')\;\;\; \text{if}\;\;\; Jv' \neq -v' \\ + iv'\;\;\; \text{if}\;\;\; Jv' = -v' + \end{cases} + \end{align} + Do this $k$ times and get $\{e_k\}$ ONB of $H$ for $J^2 = 1$. + \newline + + \textbf{2.} $J^2 = -1$\newline + $v \in H$ and set $e_1 = cv$, $c$ normalization constant. + Then we set $f_1 = Je_1$ with $f_1 \perp e_1$, this is automatically the + case because: + \begin{align} + <f_1, e_1> &= <Je_1, e_1> = -<Je_1 , J^2e_1> =\\ + &= -<Je_1, e_1> = -<f_1, e_1> + \end{align} + this only holds for 0. Then take some $v' \perp e_1, f_1$ and set\\ + $e_2 =c 'v'$ and $f_2 = Je_2 \perp e_2, f_1, e_1$. + \begin{align} + &<e_1, f_2> = <e_1, Je_2> = -<J^2e_1, Je_2> = -<e_2, Je_1> = -<e_2, + f_1>=0\\ + &<f_1, f_2> = <Je_1, Je_2> = <e_2, e_1> = 0. + \end{align} + Do this $k$ times and get $\{e_k, f_k\}$ ONB of $H$ for $J^2 = -1$ + +\end{proof} + +Apply Lemma \ref{lemma} to the real structure $J$ on a spectral triple. $J$ +implements right action of $A$ on $H$ with +\begin{align} + a^\circ = Ja^* J^{-1} +\end{align} +and satisfying $[a, b^\circ]=0$. With the block form of $A$, this implies +\begin{align} + J(a^*_1 \oplus \cdots \oplus a_N^*) = (a^\circ_1 \oplus \cdots \oplus + a_N^\circ)J. +\end{align} +With this we can conclude that the Krajewski diagram for a real finite spectral +triple is symmetric along the diagonal.$J$ hast then the following bilinear +mapping: +\begin{align} + J:\;\; \mathbb{C}^{n_i} \otimes \mathbb{C}^{n_j\circ} \otimes V_{ij} + \rightarrow \mathbb{C}^{n_j} \otimes \mathbb{C}^{n_i\circ} \otimes V_{ji}. +\end{align} + +\begin{proposition} + \label{proposition} + Let $J$ be a real structure on a finite real spectral triple $(A, H , D; + J)$. + \begin{enumerate} + \item If $J^2 = 1$ (K0-dimension 0, 1, 6, 7) $Rightarrow \;\; \exists$ + an ONB $\{e_k^{(ij)}\}$\\ with $e_k^{(ij)} \in \mathbb{C}^{n_i} \otimes + \mathbb{C}^{n_j\circ} \otimes V_{ij}$ such that + \begin{align} + Je_k^{(ij)} = e_k^{(ij)} \;\;\; (i, j = 1,\dots,N;\; k=1,\dots + dim(V_{ij})) + \end{align} + \item If $J^2 = -1$ (KO-dimension 2, 3, 4, 5) $\Rightarrow \;\; \exists$ + ONB $\{e_k^{(ij)}, f_k^{(ji)}\}$ \\ + with $e_k^{(ij)} \in \mathbb{C}^{n_i} \otimes \mathbb{C}^{n_j\circ} + \otimes V_{ij}$ and $f_k^{(ji)} \in \mathbb{C}^{n_j} \otimes + \mathbb{C}^{n_i\circ} \otimes V_{ji}$ such that + \begin{align} + Je_k^{(ij)} = f_k^{(ji)} \;\;\; (i\leq j=1,\dots, N;\; + k=1,\dots,dim(V_{ji})). + \end{align} + \end{enumerate} +\end{proposition} +\begin{proof} + Similar to Lemma \ref{lemma}. +\end{proof} + +For whatever unknown reasons this implies that in the case of KO-dimension 2, +3, 4, 5, diagonals $H_ii$ need to have even multiplicity. + +\textbf{The finite Dirac Operator:} Is a mapping between $H_{ij}$ to $H_{kl}$ + +\begin{align} + D_{ij,kl}: \; \mathbb{C}^{n_i} \otimes + \mathbb{C}^{n_j\circ}\otimes V_{ij} \rightarrow \mathbb{C}^{n_k} \otimes + \mathbb{C}^{n_l\circ}\otimes V_{kl} +\end{align} +We have $D_{kl,ij} = D^*_{ij, kl}$. And in the diagram we have a line between +the nodes $(\textbf{n}_i, \textbf{n}_j^\circ)$ and $(\textbf{n}_l, +\textbf{n}_k^\circ)$. But instead of drawing directional lines draw a single +undirected line that represents both $D_{ij, kl}$ and the adjoint $D_{kl, ij}$. + +\begin{lemma} + The conditions $JD = \pm DJ$ and $[[D,a], b^\circ] = 0$ imply that the + connections in the diagram run only vertically or horizontally and thereby + the diagonal symmetry between the nodes is preserved. +\end{lemma} +\begin{proof} + The condition $JD = \pm DJ$ has the following commutative diagram. + +\[ +\begin{tikzcd} + \mathbb{C}^{n_i\circ}\otimes \mathbb{C}^{n_j\circ}\otimes V_{ij} + \arrow[r,"D"] \arrow[d,swap,"J"] & + \mathbb{C}^{n_k\circ}\otimes \mathbb{C}^{n_l\circ}\otimes V_{kl} \arrow[d,"J"] \\ +\mathbb{C}^{n_j\circ}\otimes \mathbb{C}^{n_i\circ}\otimes V_{ji} \arrow[r,"\pm D"] & +\mathbb{C}^{n_l\circ}\otimes \mathbb{C}^{n_k\circ}\otimes V_{lk} +\end{tikzcd} +\] + Relating $D_{ij, kl}$ to $D_{ji, lk}$ and maintaining diagonal symmetry. + Wit the condition $[[D, a], b^\circ]=0$ for the diagonal elements $a = + \lambda_1\mathbb{I}_{n_1}\oplus \cdots \oplus \lambda_N \mathbb{I}_{n_N} + \in A$ and $b = \mu_1\mathbb{I}_{n_1}\oplus \cdots \oplus \mu_N + \mathbb{I}_{n_N} \in A$, with some $\lambda _i , \mu _i \in \mathbb{C}$, we + can commute: + \begin{align} + D_{ij, kl} (\lambda _i - \lambda _k)(\bar{\mu}_j - \bar{\mu}_l)= 0 + \end{align} + $\forall \lambda _i , \mu _j \in \mathbb{C}$, thus $D_ij, kl = 0$ for + $i\neq j$ or $j\neq i $. +\end{proof} +\textbf{The Grading:} $\gamma : \; H \rightarrow H$ each node gets labeled by a $+$ +or a $-$ sign. + +\begin{itemize} + \item D only connects nodes with different signs + \item If $(\textbf{n}_i, \textbf{n}_j^\circ)$ has a $\pm$ sing then + $(\textbf{n}_j, \textbf{n}_i^\circ)$ has a $\mp$, $\varepsilon''$ sign\\ + according to $J\gamma = \varepsilon'' \gamma J$ +\end{itemize} + +\begin{definition} + A Krajewski Diagram of KO-dimension $k$ is an ordered pair $(\Gamma, + \Lambda)$ where $\Gamma$ is a finite graph and $\Lambda$ is a set of + positive integers with a labeling: + + \begin{itemize} + \item of $v \in \Gamma^{(0)}$ of vertices by elements $\iota (v) = + (n(v), m(v))\; \in \; \Lambda \times \Lambda$, an edge from $v$ to + $v'$ implies that either $n(v) = n(v')$ or $m(v) = m('v)$ or both + \item of $e = (v_1, v_2) \in \Gamma^{(1)}$ edges with non-zero + operators $D_e$ and their adjoints $D_e^*$: + \begin{align*} + &D_e:\mathbb{C}^{n(v_1)} \rightarrow + \mathbb{C}^{n(v_2)}\;\;\;\;\; &\text{if} \;\;\;\; m(v_1) = m(v_2)\\ + &D_e:\mathbb{C}^{m(v_1)} \rightarrow + \mathbb{C}^{m(v_2)}\;\;\;\;\; &\text{if} \;\;\;\; n(v_1) = n(v_2) + \end{align*} + \end{itemize} + Together with an involutive graph automorphism $j:\Gamma \Rightarrow + \Gamma$ such that the following conditions hold: + \begin{enumerate} + \item every row or column in $\Gamma \times \Gamma$ has non-empty + intersection with $\iota(\Gamma)$ + \item for each vertex $v$ we have $n(j(v)) = m(v)$ + \item for each edge $e$ we have $D_e = \epsilon' D_{j(e)}$ + \item if KO dimension $k$ is even, then the vertices are labeled by + $\pm 1$ and the edges only connect opposite signs. The signs at $v$ + and $j(v)$ differ by a factor of $\epsilon$ + \item if the K0-dimension is 2, 3, 4, 5 then the inverse image of + $\iota$ of the diagonal elements in $\Lambda \times \Lambda$ + contains an even number of vertices of $\Gamma$ + \end{enumerate} +\end{definition} +With this definition we can label different vertices by the same element in +$\Lambda \times \Lambda$ (accounting for the multiplicities in $V_{ij}$) +\newline + +\textbf{Diagram:} To sum it up we have the following diagram +\begin{itemize} + \item Node at $(\textbf{n}_i, \textbf{n}_j^\circ)$ for each vertex with that label + \item Operators $D_e$ add up to $D_{ij,kl}$ connecting nodes $(\textbf{n}_i, + \textbf{n}_j^\circ)$ with $(\textbf{n}_k, \textbf{n}_l^\circ)$ + \begin{align} + D_{ij, kl} = \sum\limits_{\substack{e=(v_1, v_2) \in \Gamma^{(1)} + \\ \iota(v_1) = (\textbf{n}_i, \textbf{n}_j)\\ + \iota(v_2)=(\textbf{n}_k, \textbf{n}_l)}} D_e + \end{align} + \item only vertical or horizontal connections +\end{itemize} + +\begin{theorem} + There is a one-to-one correspondence between finite real spectral triples + $(A, H, D; J, \gamma)$ + of K0-dimension $k$ modulo unitary equivalence and Krajewski diagrams of + KO-dimension $k$ in the following way: + + \begin{align} + & A = \bigoplus_{n \in \Lambda} M_n(\mathbb{C})\\ + & H = \bigoplus_{v \in \Gamma^{(0)}} \mathbb{C}^{n(v)} \otimes + \mathbb{C}^{m(v)\circ}\\ + & D = \sum_{e\in \Gamma^{(1)}} D_e + D_e^* + \end{align} + The real structure $J:H\rightarrow H$ is given as as in Proposition + \ref{proposition} with a basis dictated by a graph automorphism $j: \Gamma + \rightarrow \Gamma$. The grading $\gamma$ is difened by setting $\gamma = + \pm 1$ on $\mathbb{C}^{n(v)} \otimes \mathbb{C}^{m(v)\circ} \subset H$ + according to the labeling $\pm$ of the vertex $v$. +\end{theorem} + +\begin{example} + $A = M_n(\mathbb{C})$ with $\hat{A} = {\textbf{n}}$. We have the following + Krajewski diagram. + \begin{figure}[h!] \centering + \begin{tikzpicture}[ + dot/.style = {draw, circle, inner sep=0.06cm}, + no/.style = {}, + ] + \node[no](a) at (0,0) [label=left:$\textbf{n}^\circ$] {}; + \node[no](c) at (0.25, 0.25) [label=above:$\textbf{n}$] {}; + \node[dot](d0) at (0.25,0) [] {}; + \end{tikzpicture} + \end{figure} + \begin{itemize} + \item We can label the node either with a $+$ or a $-$ sign, the choice being + irrelevant + \item $H = \mathbb{C}^n \otimes \mathbb{C}^{n\circ} \simeq + M_n(\mathbb{C})$ + \item $\gamma$ trivial grading ($+1$) + \item $J$ is a combination of complex conjugation and the flip + $n\otimes n^\circ$ ($\Rightarrow M_n(\mathbb{C})$ as matrix adjoint) + \item Because node label is $\pm$ there is no non-zero Dirac operator + \item $\Rightarrow (A = M_n(\mathbb{C}), H=M_n(\mathbb{C}) , D=0; J=(\cdot)^*, + \gamma = 1)$ + \end{itemize} +\end{example} +\section{Real Algebras and Krajewski Diagrams} + +\begin{definition} + A real Algebra is a Vector space $A$ over $\mathbb{R}$ with $A\times A + \rightarrow A$, $(a, b) \mapsto ab$ and $1a = a1 = a \;\; \forall a\in A$ +\end{definition} + +A real *-algebra is a real algebra with a bilinear map $*:A \rightarrow A$ +such that $(ab)^* = b^*a^*$ and $(a^*)^* \;\;\; \forall a,b\in A$ +\begin{example} + Real *-algebra of quaternions $\mathbb{H}$ subalgebra of $M_2(\mathbb{C})$. + \begin{align} + \mathbb{H} = \{ \begin{pmatrix}\alpha & \beta \\ -\bar{\beta} & + \bar{\alpha}\end{pmatrix} : \alpha, \beta \in + \mathbb{C}\} + \end{align} + $\mathbb{H}$ consists of matricies that commute in $M_2(\mathbb{C})$ with + the operator $I$ defined by: + \begin{align} + I\begin{pmatrix}v_1 \\ v_2\end{pmatrix} = \begin{pmatrix}-\bar{v}_2 \\ + \bar{v}_1\end{pmatrix} + \end{align} + The involution is the hermitian conjugation of $M_2(\mathbb{C})$. +\end{example} +\begin{MyExercise} + \textbf{ + \begin{enumerate} + \item Show that $\mathbb{H}$ is a real *-algebra which contains a + real subalgebra isomorphic to $\mathbb{C}$. + \item Show that $\mathbb{H} \otimes_\mathbb{R} \mathbb{C} \simeq + M_2(\mathbb{C})$ as complex *-algebras. + \item Show that $M_k(\mathbb{H})$ is areal *-algebra for any $k$ + \item Show that $M_k(\mathbb{H} \otimes_{\mathbb{R}} \mathbb{C} + \simeq M_{2k}(\mathbb{C})$ as complex *algebras. + \end{enumerate} + } +\end{MyExercise} +\begin{definition} + A representation of a finite-dimensional real * algebra $A$ is a pair $(\pi + , H$), $H$- Hilbertspace, $\pi : A \rightarrow L(H)$ +\end{definition} +\begin{MyExercise} + Show that there is a one-to-one correspondence between Hilbertspace + representations of real *-algebras $A$ and complex representations of its + complexification $A\otimes _\mathbb{R} \mathbb{C}$. Conclude that the + unique irreducible Hilbertspace representation of $M_k(\mathbb{H})$ is + $\mathbb{C}^{2k}$ +\end{MyExercise} +\begin{lemma} + Real *-algebra $A$ represented faithfully on a finite dimensional + Hilbertspace $H$ through a real linear *-algebra map $\pi: A \rightarrow + L(H)$ hen $A$ is a matrix algebra. + \begin{align} + A \simeq \bigoplus _{i=1}^N M_{n_i} (\mathbb{F}_i) + \end{align} + Where $\mathbb{F}_i = \mathbb{R}, \mathbb{C}, \mathbb{H}$ depending on $i$. +\end{lemma} +\begin{proof} + $\pi$ allows $A$ to be considered as a real *-subalgebra of + $M_{dim(H)}(\mathbb{C}) \Rightarrow A+iA$ complex *-subalgebra of + $M_{dim(H)}(\mathbb{C})$. Then $A+iA$ is a matrix algebra and $A+iA = + M_k(\mathbb{C})$ for $k \geq 1$. Thus we have + \begin{align} + A \cap iA = + \begin{cases} + \{0\} \;\;\;\; \text{if $A = M_k(\mathbb{C})$}\\ + A+iA = M_k(\mathbb{C}) + \end{cases} + \end{align} + Furthermore $A$ is a fixed point algebra of an anti-linear automorphism + $\alpha$ of $M_k(\mathbb{C})$ with $\alpha(a+ib) = a-ib$ for $a, b \in A$. + Implement $\alpha$ by an anti-linear isometry $I$ on $\mathbb{C}^n$ such + that $\alpha (x) = I\times I^{-1}\;\;\;\ \forall x\in M_k(\mathbb{C})$. + Now since $\alpha^2 = 1$, $I^2$ commutes with $M_k(\mathbb{C})$ and is + proportional to a complex scalar $I^2 = \pm 1 $ and A is the commutant of + $I$ + \begin{itemize} + \item if $I^2 = 1 \;\;\ \Rightarrow \;\; \exists \;\;\ \{e_i\}$ ONB of + $\mathbb{C}^k$ with $Ie_i = e_i$, then $A=M_k(\mathbb{R})$ + \item if $I^2 = -1 \;\;\ \Rightarrow \;\; \exists \;\;\ \{e_i,f_i\}$ ONB of + $\mathbb{C}^k$ with $Ie_i = f_i$, ($k$ even)\\ + Therefor $I$ must be a $k/2 \times k/2$ matrix because of commutation with + $M_k(\mathbb{C})$, then $A = M_{k/2} (\mathbb{H})$ + \end{itemize} +\end{proof} +The Krajewski diagrams can also classify real algebras, as long as we take +$\mathbb{F}_i$ for each $i$ into account. That is we enhance the set $\Lambda$ +to be +\begin{align} + \Lambda = \{ \textbf{n}_1 \mathbb{F}_1,\dots, \textbf{n}_N \mathbb{F}_N\} +\end{align} +Reducing in to the previous $\Lambda$ if all $\mathbb{F}_i = \mathbb{C}$. +\section{Classification of Irreducible Geometries} +Classify irreducible real spectral triples based on $M_N(\mathbb{C} \oplus +M_N(\mathbb{C})$ for some $N$ +\begin{definition} + A finite real spectral triple $(A, H, D; J, \gamma)$ is called irreducible + if the triple $(A, H, J)$ is irreducible, that is when + \begin{enumerate} + \item The representation of $A$ and $J$ on $H$ are irreducible + \item The action of $A$ on $H$ has a separating vector + \end{enumerate} +\end{definition} + +\begin{theorem} + Let $(A, H, D; J, \gamma)$ be an irreducible finite real spectral triple of + KO-dimension 6. Then exists a positive integer $N$ such that $A \simeq + M_N(\mathbb{C}) \oplus M_N(\mathbb{C})$. +\end{theorem} +\begin{proof} + Let $(A, H, D; J, \gamma)$ be an arbitrary finite real spectral triple, + corresponding to + \begin{align} + &A = \bigoplus_i^{N} M_{n_i}(\mathbb{C})\\ + &H = \bigoplus_{i,j=1}^N \mathbb{C}^{n_i} \otimes + \mathbb{C}^{n_j\circ} \otimes V_{ij} + \end{align} + Remember that each $\mathbb{C}^{n_i} \otimes \mathbb{C}^{n_j}$ is a + irreducible representation of $A$. In order for $H$ to support the real + structure $J$ we need both $\mathbb{C}^{n_i} \otimes \mathbb{C}^{n_j}$ + and $\mathbb{C}^{n_j} \otimes \mathbb{C}^{n_i}$. With Lemma \ref{lemma} + with $J^2 = 1$ with multiplicity $dim(V_{ij}) = 1$ we have such a + structure. Hence + \begin{align} + H = \mathbb{C}^{n_i} \otimes \mathbb{C}^{n_j} \oplus \mathbb{C}^{n_j} + \otimes \mathbb{C}^{n_i} + \end{align} + For $i,j \in \{1, \dots, N\}$ + \newline + + For the second condition (existence of the separating vector). The + representations of $A$ in $H$ are only faithful if $A = M_{n_i}(\mathbb{C}) + \oplus M_{n_j}(\mathbb{C})$. The stronger condition applies $n_i = n_j$ + then we have $A' \xi = H$ with the commutant of $A$ and $\xi \in H$ the + separating vector. Normally since $A' = M_{n_j}(\mathbb{C}) \oplus + M_{n_i}(\mathbb{C})$ with $dim(A') = n_i^2 + n_j^2$ and $dim(H) = 2n_i n_j$ + we have a equality $n_i = n_j$. +\end{proof} +\end{document} + + + + +