1: \documentclass{JHEP3}
2:
3: \usepackage{amsmath,amsfonts,amssymb,graphicx,subfigure,textcomp}
4: \usepackage[OT1,T1]{fontenc}
5:
6: \def\be#1#2\ee{\begin{equation}\label{#1}#2\end{equation}}
7: \def\bea#1#2\eea{\begin{eqnarray}\label{#1}#2\end{eqnarray}}
8: \def\note#1{\marginpar{\raggedright\if@twoside\ifodd\c@page\raggedleft\fi\fi\sf\scriptsize Note: #1}}
9: \newcommand{\bZ}{\mathbb{Z}}\newcommand{\bN}{\mathbb{N}}
10: \newcommand{\cN}{\mathcal{N}}\newcommand{\cR}{\mathcal{R}}
11: \newcommand{\A}{A}\newcommand{\B}{B}
12: \newcommand{\AAA}{AAA}\newcommand{\AAB}{AAB}
13: \newcommand{\ABA}{ABA}\newcommand{\ABB}{ABB}
14: \newcommand{\BBA}{BBA}\newcommand{\BBB}{BBB}
15: \newcommand{\ZZ}{\bZ_2\!\times\!\bZ_2}
16: \newcommand{\N}{\mathbf{N}}
17: \newcommand{\Sym}{\mathbf{Sym}}\newcommand{\Anti}{\mathbf{Anti}}
18: \newcommand{\rmap}{\stackrel{\cR}{\to}}
19: \newcommand{\lrmap}{\stackrel{\cR}{\leftrightarrow}}
20: \newcommand{\e}{\varepsilon}\newcommand{\te}{\tilde{\varepsilon}}
21: \newcommand{\const}{\mathrm{const}}
22: \newcommand{\rep}[1]{\mathbf{#1}}
23: \newcommand{\1}{\rep{1}}
24: \newcommand{\2}{\rep{2}}
25: \newcommand{\3}{\rep{3}}
26: \newcommand{\incclipfig}[1]{%
27: % \includegraphics[width=0.49\linewidth,trim=0mm 15mm 0mm 15mm,clip]{#1}}
28: \includegraphics[width=0.49\linewidth]{#1}}
29: \newcommand{\fig}[3]{\begin{figure}[t]\begin{center}%
30: \includegraphics[width=0.7\linewidth,trim=0mm 15mm 0mm 15mm,clip]{#1}%
31: \caption{#3}\label{#2}\end{center}\end{figure}}
32: \newcommand{\widefig}[3]{\begin{figure}[t]\begin{center}%
33: \includegraphics[width=\linewidth]{#1}\caption{#3}\label{#2}%
34: \end{center}\end{figure}}
35: \newcommand{\twofig}[4]{\begin{figure}[t]\begin{center}%
36: \subfigure[\label{#3_a}]{\incclipfig{#1}}\hfill%
37: \subfigure[\label{#3_b}]{\incclipfig{#2}}%
38: \caption{#4}\label{#3}\end{center}\end{figure}}
39: \newcommand{\fourfig}[6]{\begin{figure}[t]\begin{center}%
40: \subfigure[\label{#5_a}]{\incclipfig{#1}}\hfill%
41: \subfigure[\label{#5_b}]{\incclipfig{#2}}\\%
42: \subfigure[\label{#5_c}]{\incclipfig{#3}}\hfill%
43: \subfigure[\label{#5_d}]{\incclipfig{#4}}\\%
44: \caption{#6}\label{#5}\end{center}\end{figure}}
45: \hyphenation{models}
46: \renewcommand{\arraystretch}{1.3}
47:
48:
49: \preprint{\small{NIKHEF/2007-006}\\[-1ex]\small{MPP-2007-22}\\[-1ex]\small{LMU-ASC 14/07}\\[-1ex]\small{hep-th/0703011}}
50: \title{Statistics of intersecting D--brane models on $\mathbf{T^6/\bZ_6}$}
51: \author{Florian Gmeiner$^{1}$, Dieter L\"ust$^{2,3}$ and Maren Stein$^{3}$\\
52: ~\\
53: {}$^1$NIKHEF,\\
54: \phantom{$^1$}Kruislaan 409, 1098 SJ Amsterdam, The Netherlands\\
55: ~\vspace{-1ex}\\
56: {}$^2$Max--Planck--Institut f\"ur Physik,\\
57: \phantom{$^2$}F{\"o}hringer Ring 6, D--80805 M{\"u}nchen, Germany\\
58: ~\vspace{-1ex}\\
59: {}$^3$Arnold--Sommerfeld--Center for Theoretical Physics,\\
60: \phantom{$^3$}Department f{\"u}r Physik, Ludwig--Maximilians--Universit{\"a}t,\\
61: \phantom{$^3$}Theresienstra{\ss}e 37, D--80333 M{\"u}nchen, Germany\\
62: ~\vspace{-1ex}\\
63: E-mail: {\sf fgmeiner@nikhef.nl, luest@mppmu.mpg.de,
64: mstein@theorie.physik.uni-muenchen.de}
65: \bigskip}
66:
67: \abstract{%
68: We perform a statistical analysis of supersymmetric intersecting D--brane
69: models on the type II orientifold $T^6/\bZ_6$.
70: After providing an analytic proof of the finiteness of the number of possible
71: solutions in this setup we study the frequency distributions of properties
72: of the gauge group and the chiral matter content.
73: In particular we search for models with a standard model gauge group and
74: discuss their statistical suppression. The results are compared with the
75: recent studies on $T^6/(\ZZ)$.
76: The analysis is conducted using a statistical method, based on the choice of
77: random subsets of the full ensemble of solutions. This method allows to
78: calculate the total number of models with high precision to~$3\times 10^{28}$.}
79:
80: \keywords{string statistics, intersecting branes, orientifolds}
81:
82:
83: \begin{document}
84:
85: %
86: % INTRODUCTION
87: %
88: \section{Introduction}\label{intro}
89: In order to make contact with low energy physics, the quest to find a realistic
90: MSSM--like string vacuum is one of the most important tasks for string
91: phenomenology. In the context of type II orientifolds there has been a huge
92: amount of activity over the last years to obtain a model that resembles the
93: standard model as closely as possible\footnote{For reviews on this topic see
94: e.g.~\cite{lu04,bcls05,bkls06}.}.
95: Since it is believed that there exists a vast landscape of string
96: vacua containing a huge number of possible solutions~\cite{lls86,do03},
97: new methods have to be used to analyse this tremendous realm.
98:
99: Instead of studying individual solutions, it might be better to analyse an
100: ensemble of models using a statistical approach~\cite{do03}.
101: With statistical methods, one can try to answer questions about the
102: distribution of certain properties within the ensemble of solutions.
103: These distributions might give important insights into the overall shape
104: of the landscape. On the one hand, they could be a valuable guide for model
105: building, giving hints where to look for interesting
106: solutions\footnote{For recent reviews on distributions on the landscape and
107: counting of flux compactifications, see~\cite{ku06,doka06, ddk07}.}.
108: Moreover, the issue of correlations of properties within the ensemble of models
109: is of great importance. Finding correlations implies that it might be possible
110: to deduce general aspects of the landscape, independent of specific
111: models.
112:
113: Dealing with statistics, there are several caveats not to be overlooked.
114: One of them concerns the finiteness of solutions~\cite{acdo06}.
115: If the ensemble to be analysed is not finite, the possibility to make clear
116: statements is greatly diminished, since one has to rely on properties which
117: appear in a regular pattern.
118: The same applies for a random sample, which has to be chosen with great
119: care, in order to make it a representative subset of the full range of
120: solutions.
121:
122: In~\cite{bghlw04} and~\cite{gbhlw05} methods to analyse the open string
123: sector of intersecting brane models have been developed.
124: In the second paper a survey of models on a $T^6/(\ZZ)$
125: orbifold was carried out using a computer based approach. This
126: technique was also used to analyse the statistics of standard--model--like
127: as well as $SU(5)$ and flipped $SU(5)$ models on the same orbifold in greater
128: detail~\cite{gm05,gmst06} (for a summary of the results obtained for this
129: geometry see also~\cite{gm06}). In~\cite{kuwe05} a survey of standard model
130: vacua including fluxes has been accomplished for this background.
131: An analytic proof of the finiteness of solutions to the tadpole and
132: supersymmetry constraints in the case of an $T^6/(\ZZ)$ orbifold has been
133: given in~\cite{dota06}.
134: Moreover a statistical analysis of Gepner model orientifolds was performed
135: in~\cite{dhs04a,dhs04b,adks06}, and aspects of the heterotic landscape were
136: discussed e.g. in~\cite{di06,le06}.
137:
138: It is clear that the statistical analysis performed in the articles mentioned
139: above for the case of the $T^6/(\ZZ)$ orientifold should be repeated for other
140: background
141: geometries in order to see if these previous results are somehow generic and
142: persist, or if they are substantially different for other spaces.
143: In this article we use similar methods as in the works described above to
144: analyse a different intersecting brane setup, namely the IIA orientifold with
145: intersecting D--branes on the $T^6/\bZ_6$ orbifold.
146: This class of models is also interesting from a phenomenological perspective,
147: since it has already been shown that one can construct an intersecting brane
148: model with three generations of quarks and leptons on this space~\cite{hoot04}.
149:
150: There are many similarities to the $T^6/(\ZZ)$ case,
151: but we encounter some new aspects as well. In particular, this background
152: requires fractional branes, coming from the $\bZ_2$--twisted sector of the
153: orbifold~\cite{ddg97,digo99}. As it turns out, these fractional branes are
154: essential for the properties of the low energy theory, in particular for the
155: existence of chiral matter. Moreover, due to the existence of the fractional
156: branes the number of solutions to the constraining equations is largely
157: increased compared to the
158: $T^6/(\ZZ)$ case, and the statistical distributions are also different.
159:
160: In order to make statistical statements for the full parameter space,
161: we use a new method of analysis, based on the choice of random subsets of
162: solutions\footnote{If not further specified in the text, we will use the term
163: ``solution'' in this article to stand for a specific model that fulfills the
164: tadpole, supersymmetry and K--theory conditions, which will be given
165: explicitly in Section~\ref{constraints}.}.
166: As emphasized in~\cite{dile06}, this has to be done very
167: carefully, in order to obtain results that resemble the actual frequency
168: distributions as closely as possible, since ``floating correlations'' could
169: have the unwanted effect that certain observables are functions of the
170: considered examples.
171: Fortunately, as we will show in this article, the results
172: obtained in this way are indeed sufficiently close to the full results to be
173: trusted. We are confident that this method could also be applied to different
174: setups and, since it greatly reduces the amount of necessary computations,
175: might prove useful for subsequent surveys of the landscape.
176:
177:
178: \subsection{Outline}\label{outline}
179: This paper is organised as follows. In Section~\ref{geometry} we will recall
180: the geometric setup of $T^6/\bZ_6$, explain the orbifold and orientifold
181: projections and describe the space of three--cycles. Section~\ref{constraints}
182: contains a discussion of the constraining equations from tadpole cancellation,
183: supersymmetry and K--theory.
184: In Section~\ref{finite} we give an analytic proof of the finiteness of possible
185: solutions to the constraining equations. We explain our methods of statistical
186: analysis in Section~\ref{methods} and present the obtained results on the
187: distribution of gauge sector observables in Section~\ref{results}.
188: In particular, we look for the frequency distribution of models with a
189: standard model gauge group and their chiral matter content.
190: Finally we summarise our results and give an outlook to further directions of
191: research.
192:
193:
194: %
195: % BASIC SETUP
196: %
197: \section{Geometry}\label{geometry}
198: In this section we will review the geometric setup of the $T^6/\bZ_6$
199: orientifold and possible D--brane configurations. We will use the notation
200: and conventions of~\cite{hoot04}, to which we refer for more details on the
201: geometry and explicit derivations of some of the results we use in the
202: following.
203:
204: \subsection{Orbifold and orientifold projections}\label{projections}
205: We assume a factorisation of the $T^6$ into three two--tori, described by
206: complex coordinates $z^i$, $i=1,2,3$, on which the orbifold group $\bZ_6$ acts
207: as
208: \be{eqorbaction}\nonumber
209: \theta: z^i\mapsto e^{2\pi iv_i}z^i,
210: \ee
211: with the shift vector defined as $\vec{v}_i=\frac{1}{6}(1,1,-2)$. There
212: exists another possible action, often denoted by $\bZ'_6$, with a different
213: shift vector (for a recent model building approach on $\bZ'_6$
214: see~\cite{balo06}). We will not consider the $\bZ'_6$ orbifold in this article,
215: but plan to come back to it in the future~\cite{ghsXX}.
216:
217:
218: In addition to the orbifold group we introduce an orientifold projection,
219: consisting of the reflection of worldsheet parity $\Omega$ and an
220: antiholomorphic involution $\cR$, which we choose to be complex conjugation,
221: \be{eqraction}
222: \cR:z^i\mapsto\bar{z}^i.
223: \ee
224:
225: In order for the orbifold and orientifold projections to be compatible,
226: \eqref{eqraction} has to be an automorphism of the $\bZ_6$ lattice. This
227: allows for only two possible geometries of the three two--tori, denoted by
228: $\A$ and $\B$. In the case of an $\A$--geometry the torus lattice is given
229: by the root lattice of $SU(3)$, spanned by
230: $\{\sqrt{2}, (1+i\sqrt{3})/\sqrt{2})\}$.
231: The $\B$--geometry, which corresponds to a $D9$--brane with background--flux
232: in the dual type IIB picture, can be obtained from the $\A$--case by a
233: rotation of $e^{-i\pi/6}$.
234:
235: Choosing different geometries for the three two--tori and considering only
236: those combinations which cannot be obtained by trivially interchanging the
237: first and second torus, which transform in the same way under $\theta$, we
238: obtain six different possible setups, denoted in the following by
239: $\AAA,\AAB,\ABA,\ABB,\BBA$ and $\BBB$.
240:
241: \widefig{t6z6}{figt6z6}{The three two--tori of the $T^6/\bZ_6$ orbifold in the
242: $\AAA$--geometry. The fundamental cycles of the $T^6$ are denoted by $\pi_i$.
243: The fixed points of $\theta^3$ on the first two $T^2$s, which are relevant for
244: the definition of exceptional cycles, are marked by dots. The third torus is
245: invariant under $\theta^3$.}
246:
247:
248: \subsection{Three--cycles}\label{cycles}
249: To wrap O6-planes and D6-branes on this geometry, we are interested in the
250: number of three--cycles, given by the third Betti number $b_3=2(1+h_{2,1})$.
251: According to~\cite{klra00} we have $h_{2,1}=5$, all coming form the
252: orbifold--twisted sector. This leads in total to two bulk cycles inherited
253: from the six--torus and ten exceptional cycles, which wrap a combination of a
254: one--cycle on $T^3$ and a two--cycle around one of the $\bZ_3$ fixed points.
255: General three--cycles will be a combination of bulk and exceptional cycles, but
256: one has to keep in mind that only those combinations are possible in which the
257: bulk cycle passes through the fixed point in question.
258:
259: \subsubsection{Bulk cycles}\label{bulkcycles}
260: The factorisable bulk
261: cycles can be defined in terms of a basis of fundamental one--cycles on the
262: three two--tori. For these we use the notation $\pi_{2i-1}, \pi_{2i}$ for
263: $T^2_i$, $i=1,2,3$, as shown in Figure~\ref{figt6z6}. A basis for the bulk
264: cycles can be defined as
265: \bea{eqbasisbulkcycles}\nonumber
266: \rho_1=2\left[(1+\theta+\theta^2)\pi_1\otimes\pi_3
267: \otimes\pi_5\right],\\
268: \rho_2=2\left[(1+\theta+\theta^2)\pi_2\otimes\pi_3
269: \otimes\pi_5\right],
270: \eea
271: with intersection matrix given by
272: \be{eqbasisIab}
273: I^{(\rho)}_{ij}=\rho_i\circ\rho_j=\begin{pmatrix}0&-2\\2&0\end{pmatrix}.
274: \ee
275: Any bulk three--cycle can be expanded using the basis~\eqref{eqbasisbulkcycles}
276: as
277: \be{eqbulkcycles}
278: \Pi_a=Y_a\rho_1+Z_a\rho_2.
279: \ee
280: In terms of the wrapping numbers $n_i,m_i$ of the fundamental one--cycles
281: $\pi_{2i-1}$ and $\pi_{2i}$ the coefficients $Y_a$ and $Z_a$ read
282: \bea{eqzycoefficients}\nonumber
283: Y_a&=&n_{1,a}n_{2,a}n_{3,a}-m_{1,a}m_{2,a}m_{3,a}-
284: \sum_{i}m_{i,a}m_{j,a}n_{k,a},\\\nonumber
285: Z_a&=&\sum_{i}m_{i,a}m_{j,a}n_{k,a}+
286: \sum_{i}m_{i,a}n_{j,a}n_{k,a},\quad
287: i,j,k\in\{1,2,3\} \mathrm{cyclic}.
288: \eea
289: From~\eqref{eqbasisIab} and~\eqref{eqbulkcycles} one computes the intersections
290: between two bulk cycles to be
291: \be{eqbulkIab}\nonumber
292: I_{ab}:=\Pi_a\circ\Pi_b=2(Z_aY_b-Y_aZ_b).
293: \ee
294:
295: The action of the involution~\eqref{eqraction} on the fundamental one--cycles
296: of the two--tori for the two possible geometries $\A$ and $\B$ is given by
297: \be{eqcycleaction}\nonumber
298: \A:\left\{\begin{array}{rcl}\pi_{2i-1}&\rmap&\pi_{2i-1},\\\pi_{2i}&\rmap&
299: \pi_{2i-1}-\pi_{2i},\end{array}\right.\qquad
300: \B:\left\{\begin{array}{rcl}\pi_{2i-1}&\rmap&\pi_{2i},\\\pi_{2i}&\rmap&
301: \pi_{2i-1}.\end{array}\right.
302: \ee
303: This leads to the following transformations of the bulk
304: cycles~\eqref{eqbasisbulkcycles} for the six inequivalent geometries,
305: \be{eqronbulk}
306: \begin{array}{l@{\;\;\rho_1\rmap\;}l@{\quad\rho_2\rmap\;}l}
307: \AAA: & \rho_1, & \rho_1-\rho_2,\\
308: \AAB: & \rho_2, & \rho_1,\\
309: \ABA: & \rho_2, & \rho_1,\\
310: \ABB: & \rho_2-\rho_1, & \rho_2,\\
311: \BBA: & \rho_2-\rho_1, & \rho_2,\\
312: \BBB: & -\rho_1, & \rho_2-\rho_1.
313: \end{array}
314: \ee
315: To obtain the cycles wrapped by O6-planes we have to combine two orbits,
316: invariant under $\Omega\cR\theta^{2k}$ and $\Omega\cR\theta^{2k+1}$,
317: respectively. For the different geometries we obtain
318: \be{eqoplanes}
319: \Pi_{O6} = \left\{\begin{array}{lll}
320: \AAA: 4\rho_1,&\qquad&\ABB: 6\rho_2,\\
321: \AAB: 4(\rho_1+\rho_2),&\qquad&\BBA: 4\rho_2,\\
322: \ABA: 2(\rho_1+\rho_2),&\qquad&\BBB:4(-\rho_1+2\rho_2).
323: \end{array}\right.
324: \ee
325:
326:
327: \subsubsection{Exceptional cycles}\label{excycles}
328: In addition to the three--cycles inherited from the
329: six--torus we obtain additional, so--called exceptional cycles, which wrap a
330: product of cycles around the $\theta^3$--orbifold fixed points
331: (denoted by 1,2,3,4 in Figure~\ref{figt6z6}) and a one--cycle on $T_3$. This
332: situation is similar to the one that has been encountered in the case of
333: compactifications on $T^6/\bZ_4$ in~\cite{bgo02}.
334:
335: We can choose the following basis of exceptional cycles, invariant under the
336: orbifold projection,
337: \bea{eqbasisexceptionalcycles}\nonumber
338: \e_1 &=& (e_{21}-e_{41})\otimes\pi_5+(e_{41}-e_{31})\otimes\pi_6, \\\nonumber
339: \te_1 &=& (e_{31}-e_{41})\otimes\pi_5+(e_{21}-e_{31})\otimes\pi_6, \\\nonumber
340: \e_2 &=& (e_{12}-e_{14})\otimes\pi_5+(e_{14}-e_{13})\otimes\pi_6, \\\nonumber
341: \te_2 &=& (e_{13}-e_{14})\otimes\pi_5+(e_{12}-e_{13})\otimes\pi_6, \\\nonumber
342: \e_3 &=& (e_{22}-e_{44})\otimes\pi_5+(e_{44}-e_{33})\otimes\pi_6, \\\nonumber
343: \te_3 &=& (e_{33}-e_{44})\otimes\pi_5+(e_{22}-e_{33})\otimes\pi_6, \\\nonumber
344: \e_4 &=& (e_{23}-e_{42})\otimes\pi_5+(e_{42}-e_{34})\otimes\pi_6, \\\nonumber
345: \te_4 &=& (e_{34}-e_{42})\otimes\pi_5+(e_{23}-e_{34})\otimes\pi_6, \\\nonumber
346: \e_5 &=& (e_{24}-e_{43})\otimes\pi_5+(e_{43}-e_{32})\otimes\pi_6, \\
347: \te_5 &=& (e_{32}-e_{43})\otimes\pi_5+(e_{24}-e_{32})\otimes\pi_6,
348: \eea
349: where we denoted the two--cycles stuck at the fixed points on $T_1$ and $T_2$
350: by $e_{ij}$, $i,j=1,\ldots,4$ as shown in Figure~\ref{figt6z6}.
351: The intersection matrix of these exceptional cycles is given by
352: \be{eqexceptionalIab}\nonumber
353: I^{(\e)}_{ij} := \bigoplus_{k=1}^5\begin{pmatrix}
354: \te_k\circ\te_k&\te_k\circ\e_k\\
355: \e_k\circ\te_k&\e_k\circ\e_k\end{pmatrix}
356: = \bigoplus_{k=1}^5\begin{pmatrix}0&-2\\2&0\end{pmatrix}.
357: \ee
358: Under the action~\eqref{eqraction} of $\cR$ the fixed points on the first two
359: $T^2$s are mapped into each other as follows,
360: \be{eqexrmap}\nonumber
361: \A:\left\{\begin{array}{rcl}1&\rmap&1,\\2&\rmap&2,\\3&\lrmap&4
362: \end{array}\right.\qquad
363: \B:\left\{\begin{array}{rcl}1&\rmap&1,\\2&\lrmap&3,\\4&\rmap&4
364: \end{array}\right.
365: \ee
366:
367: It is possible to combine the bulk cycles~\eqref{eqbasisbulkcycles} and
368: exceptional cycles~\eqref{eqbasisexceptionalcycles}, including their images
369: under the orbifold projection, into an unimodular lattice of basic
370: three--cycles~\cite{hoot04}, which is an important consistency check for
371: completeness of the symplectic basis.
372: Since this particular basis is not very convenient for computations, we will
373: not use it in the following.
374:
375:
376: \section{Model building constraints}\label{constraints}
377: In addition to the O6--planes described by~\eqref{eqoplanes}, we introduce
378: $k$ stacks of D6--branes, wrapping fractional cycles. However, we would like
379: to obtain supersymmetric models which are stable and free of anomalies.
380: Therefore the brane configuration has to fulfil several constraining
381: equations, which we will describe in the following.
382:
383: \subsection{Tadpole cancellation}\label{tadpoles}
384: In order to obtain consistent models we have to make sure that the total
385: charge of the RR seven--forms in the compact space cancels.
386: This imposes a condition on the cohomology classes of these forms, which can
387: be reformulated in homology. Denoting the orientifold image of a cycle $\Pi_a$
388: wrapped by some brane $a$ by $\Pi_{a'}$ it reads
389: \be{eqRRtadgen}
390: \sum_a N_a\left(\Pi_a+\Pi_{a'}\right)-4\Pi_{O6}=0.
391: \ee
392: We can split the tadpole condition into two parts containing contributions from
393: the bulk and exceptional cycles, respectively.
394: Since the orientifold planes wrap only bulk cycles according
395: to~\eqref{eqoplanes}, the contributions from D--branes wrapping exceptional
396: cycles have to cancel among themselves.
397:
398: Using the basis~\eqref{eqbulkcycles} and the transformation
399: rules~\eqref{eqronbulk},
400: we find for the six different geometries the following
401: conditions\footnote{These conditions can also be derived explicitly by
402: computing open string amplitudes, see~\cite{hoot04}.} for $k$ bulk branes with
403: stack sizes $N_a$,
404: \be{eqtadbulk}\begin{array}{rlcrl}
405: \AAA:&\sum_{a=1}^k N_a(2Y_a+Z_a)=16,&\qquad&
406: \ABB:&\sum_{a=1}^k N_a(Y_a+2Z_a)=24,\\
407: \AAB:&\sum_{a=1}^k N_a(Y_a+Z_a)=16,&\qquad&
408: \BBA:&\sum_{a=1}^k N_a(Y_a+2Z_a)=16,\\
409: \ABA:&\sum_{a=1}^k N_a(Y_a+Z_a)=8,&\qquad&
410: \BBB:&\sum_{a=1}^k N_aZ_a=16.
411: \end{array}\ee
412:
413:
414: \subsection{Supersymmetry conditions}\label{susy}
415: In order to preserve $\cN=1$ supersymmetry, the bulk cycles have to be
416: calibrated with respect to the same calibration form as the orientifold
417: planes. In our case of three--cycles, this is the holomorphic three--form and
418: this means that the cycles have to be special Lagrangian.
419: Expressed in terms of the expansion coefficients~\eqref{eqbulkcycles} the
420: conditions are given by
421: \be{eqsusybulk}\begin{array}{rlcrl}
422: \AAA:&Z_a=0,&\qquad&\ABB:&Y_a=0,\\
423: \AAB:&Y_a=Z_a,&\qquad&\BBA:&Y_a=0,\\
424: \ABA:&Y_a=Z_a,&\qquad&\BBB:&2Y_a=-Z_a.
425: \end{array}\ee
426: Since these conditions boil down to the fact that the bulk
427: branes have to wrap the same cycles as the O6--planes, we obtain the result
428: that the intersection number between these branes and the orientifold planes
429: always vanishes,
430: \be{eqisbbop}
431: I_{aO6}=\Pi_a\circ\Pi_{O6}=0.
432: \ee
433: To exclude anti--branes from the spectrum, we have to impose one further
434: condition,
435: \be{eqantibranes}
436: \AAA,\AAB,\ABA:\; Y_a>0,\qquad \ABB,\BBA,\BBB:\; Z_a>0.
437: \ee
438: Fractional branes, being a combination of bulk and exceptional cycles,
439: preserve half of the supersymmetry, if the bulk part
440: obeys~\eqref{eqisbbop} and~\eqref{eqantibranes}, and the exceptional part
441: comes from fixed points that are traversed by the bulk cycle.
442: In total there are 128 different possible combinations of exceptional cycles
443: for a given bulk cycle. All possible combinations can be found in Tables 23
444: and 24 of~\cite{hoot04}.
445:
446:
447: \subsection{K--theory constraints}\label{ktheory}
448: In addition to the constraints from tadpole cancellation and supersymmetry, we
449: have to demand that the four--dimensional models are anomaly--free.
450: Cancellation of local
451: gauge anomalies is guaranteed by a generalised Green--Schwarz mechanism,
452: yet there exists the possibility to obtain a global gauge anomaly~\cite{wi82},
453: which can be deduced from a K--theory analysis.
454: In the case of our models, this condition requires an even amount of
455: chiral matter from $Sp(2)$ probe branes, inserted in the geometric
456: setup~\cite{ur00}.
457: $Sp(2)$ gauge groups are carried by branes that are invariant under the
458: orientifold action. Unfortunately this is not the only possible gauge group
459: for these branes, they can equally well support an $SO(2)$ group.
460: To differentiate between these two, one has to go beyond the algebraic approach
461: that suffices to calculate the tadpole, the susy constraints and the chiral
462: matter content. It is necessary to analyse the open string M\"obius amplitude
463: for each possible brane.
464:
465: Fortunately the geometrical setup of the $\bZ_6$ orientifold is such that we
466: do not have to worry about this issue. In fact, it can be generally proved
467: that all possible solutions that fulfil the tadpole and susy constraints will
468: automatically satisfy the stronger condition where all possible
469: orientifold--invariant probe branes are used.
470: In this case we obtain the following condition for a model with $k$ stacks of
471: branes,
472: \be{eqktheory}
473: \sum_{a=1}^kN_a\Pi_a\circ\Pi_{p} \equiv 0 \mod 2,
474: \ee
475: and this equation should hold for any probe brane $p$ invariant under the
476: orientifold map.
477:
478: Because of this property and the fact that the bulk part of the probe branes
479: does not intersect with the bulk part of all other branes, several of the
480: terms in~\eqref{eqktheory} vanish and we can rewrite it as
481: \be{eqkt2}
482: \sum_{i=1}^5\left(\sum_{a=1}^kN_as_a^i\right)r_p^i \equiv 0 \mod 2,
483: \ee
484: where the values $s_a^i$ are the coefficients of the cycles of brane $a$ which
485: are odd under the orientifold projection and the $r_p^i$ parametrise the
486: cycles of the probe branes which are even under the orientifold map. Note that
487: we are summing over exactly half of the dimension of the basis of exceptional
488: three--cycles.
489: However, not all of the $r_p^i$ are independent, since the probe branes are
490: bound to be on top of the orientifold planes. An explicit calculation shows
491: that there exist only eight different possibilities and that the coefficients
492: $r_p^i$ are always even. Therefore~\eqref{eqkt2} is always fulfilled.
493:
494:
495:
496: \subsection{Open string spectrum}\label{spectrum}
497: The massless chiral states arising at the intersection of different stacks
498: of D--branes and at the intersection of branes with their orientifold
499: mirrors and the orientifold planes, can be computed from the intersection
500: numbers. In general a stack of $N$ branes supports a $U(N)$ gauge group, unless
501: the three--cycle wrapped by this stack lies on top of the orientifold plane.
502: In this case we are dealing with an $SO(N)$ or $Sp(N)$ group.
503:
504: To compute the non--chiral spectrum, one has to analyse the open string
505: amplitudes. In our statistical analysis we will not do so, but concentrate on
506: the chiral spectrum only. As shown in Table~\ref{tabspec}, we obtain chiral
507: matter in a bi--fundamental representation at the intersection of two stacks
508: $a$ and $b$ with $N_a$ and $N_b$ branes, respectively.
509: In addition there is the possibility for each stack to contribute matter in the
510: symmetric and antisymmetric representations of the gauge group.
511: From the discussion in Section~\ref{susy} it follows that the amount of
512: symmetric and antisymmetric representations will always be the same, since
513: there can be no contribution from the intersection with the orientifold planes.
514: Moreover, it is crucial to work with fractional cycles, since all bulk cycles that occur
515: lie on top of the orientifold plane and do hence not intersect which each other.
516:
517: \begin{table}[ht]
518: \begin{center}%
519: \begin{tabular}{|c|c|}\hline
520: representations & multiplicity \\\hline
521: $(\N_a,\overline{\N}_b)$&$\Pi_a\circ\Pi_b$\\
522: $(\N_a,\N_b)$&$\Pi_a\circ\Pi'_b$ \\
523: $\Sym_a$&$\frac{1}{2}\left(\Pi_a\circ\Pi_{a'}-\Pi_a\circ\Pi_{O6}\right)$\\
524: $\Anti_a$&$\frac{1}{2}\left(\Pi_a\circ\Pi_{a'}+\Pi_a\circ\Pi_{O6}\right)$\\
525: \hline\end{tabular}%
526: \caption{Multiplicities of the chiral spectrum.}
527: \label{tabspec}
528: \end{center}
529: \end{table}
530:
531:
532: \subsection{Embedding of the standard model}\label{secsmembed}
533: Since our final goal is to quantify the number of standard model--like vacua
534: that can be found in this type of compactifications, we have to chose a way to
535: realise the gauge group and chiral matter content of the MSSM in terms of
536: intersecting branes.
537: In the present work we will consider only one type of embedding, mainly for
538: two reasons. One is given by external constraints on computational power and
539: feasibility. The second one lies in the special properties of the orbifold we
540: are investigating. Since we saw in the previous section that the amount of
541: symmetric and antisymmetric representations is always equal, several possible
542: constructions of standard model spectra that use antisymmetric representations
543: of $SU(2)$ cannot be realised, unless one also allows for chiral matter in the
544: symmetric representation, which is not desirable from a phenomenological point
545: of view.
546:
547: \begin{table}[ht]
548: \begin{center}%
549: \begin{tabular}{|c|c|c|c|}\hline
550: particle & representation & multiplicity \\\hline
551: $Q_L$ & $(\3,\overline{\2})_{0,0}+(\3,\2)_{0,0}$
552: & $I_{ab}+I_{ab'}$\\
553: $U_R$ & $(\overline{\3},\1)_{-1,0}+(\overline{\3},\1)_{0,-1}$
554: & $I_{a'c}+I_{a'd}$\\
555: $D_R$ & $(\overline{\3},\1)_{1,0}+(\overline{\3},\1)_{0,1}$
556: & $I_{a'c'}+I_{a'd'}$\\
557: $L$ & $(\1,\overline{\2})_{-1,0}+(1,\overline{\2})_{0,-1}
558: +(\1,\2)_{-1,0}+(\1,\2)_{0,-1}$
559: & $I_{bc}+I_{bd}+I_{b'c}+I_{b'd}$\\
560: $E_R$ & $(\1,\1)_{1,1}$ & $I_{cd}$ \\
561: $N_R$ & $(\1,\1)_{1,-1}$ & $I_{cd'}$ \\\hline
562: \end{tabular}%
563: \caption{Realisation of standard model particles with four stacks of branes.
564: The notation in the second column gives the representation under $SU(3)$ and
565: $SU(2)$ in brackets with the charges under the $U(1)$s of the third and fourth
566: stack as subscripts.}
567: \label{tabsmspec}
568: \end{center}
569: \end{table}
570:
571: The construction we will use in Section~\ref{secsm} for the analysis of the
572: frequency distribution of standard models is well--known and has been used
573: in many model building approaches of intersecting branes. It consists of two
574: stacks of branes ($a$ and $b$) with gauge groups $U(3)$ and $U(2)$, and two
575: branes ($c$ and $d$) with a $U(1)$ group.
576: The standard model spectrum is realised through chiral matter
577: transforming in bifundamental representations of the gauge groups.
578: The complete spectrum and the assignment to particles is given in
579: Table~\ref{tabsmspec}.
580:
581: The hypercharge $Q_Y$ is realised in this construction as a combination of the
582: $U(1)$ charges $Q_i$, with $i=\{a,b,c,d\}$ of the four branes.
583: Explicitly it is given by
584: \be{eqhyper}\nonumber
585: Q_Y=\frac{1}{6}Q_a+\frac{1}{2}Q_b+\frac{1}{2}Q_c.
586: \ee
587:
588:
589:
590: %
591: % FINITENESS
592: %
593: \section{Finiteness of solutions}\label{finite}
594: An important question that we would like to answer before analysing the
595: four--dimensional models in detail concerns the finiteness of possible
596: solutions to the constraining equations outlined in Section~\ref{constraints}.
597: To answer this question, it is sufficient to analyse the solution space of
598: the system of equations~\eqref{eqtadbulk} and~\eqref{eqsusybulk}.
599: We do not have to take the analogue expressions for the exceptional cycles into
600: account, although the set of solutions is greatly enhanced by models containing
601: exceptional cycles, because the number of possible combinations of these
602: cycles is always finite (cf. Section~\ref{susy}).
603: The K--theory constraints will play no r{\^o}le anyway, as has been argued
604: above.
605:
606: One important drawback of our approach has to be mentioned here. We cannot
607: make any statement about the dependence of the number of solutions on the
608: complex structure moduli\footnote{Concerning this point the present case
609: differs from the $\ZZ$--case considered in~\cite{dota06}.
610: On the one hand this is an advantage, because it makes the
611: proof of finiteness in the $\bZ_6$--case less involved since no free parameters
612: besides the brane wrapping numbers appear in the constraining equations. On the
613: other hand we lose a great deal of generality that can only be regained by a
614: proper analysis of the open string moduli space of the exceptional cycles --
615: an issue that is beyond the scope of this work.}.
616: The complex structures of the three two--tori are fixed by the requirement
617: to be compatible with the orbifold projection. Since $h_{2,1}=5$, we find
618: ten complex structure moduli in the twisted sector. We do not analyse the
619: blow up of the orbifold singularities and can therefore not make any statements
620: about the behaviour of our models away from the orbifold point.
621: Having said this, we will continue to prove that there is only a finite number
622: of models at this point in moduli space.
623:
624: After the susy conditions are fulfilled, we are left with one tadpole
625: condition for each possible geometry, according to~\eqref{eqtadbulk}.
626: It will contain one unknown wrapping number ($Y$ or $Z$, depending on
627: the geometry), which is always positive according to~\eqref{eqantibranes}.
628: Therefore it follows trivially that the remaining unknown in the tadpole
629: equations is bounded from above by the orientifold charge, which also depends
630: on the geometry, but will never be greater than 24.
631: To proof the finiteness of the number of models, it remains to be shown that
632: the possible combinations of wrapping numbers $\{n_i,m_i\}$, which make up
633: $Y$ and $Z$ according to~\eqref{eqzycoefficients}, are always finite.
634:
635: In the following we will give an explicit proof for the $\AAA$--geometry, the
636: other five possibilities can be treated analogously.
637: In order to simplify the discussion and reflect the symmetries of the
638: problem, we define new variables for the wrapping numbers on the first two
639: tori, while keeping the wrapping numbers on the third torus explicit.
640: \bea{eqfinvar}\nonumber
641: \alpha &:=& m_1m_2 + n_1m_2 + n_2m_1,\\
642: \qquad\beta &:=& n_1n_2+n_1m_2+n_2m_1.
643: \eea
644: Exchanging the first two tori, which is a symmetry of the geometric setup,
645: will leave $\alpha$ and $\beta$ invariant.
646: In terms of $\alpha,\beta,n_3,m_3$ \eqref{eqsusybulk} reads
647: \be{eqfin1}
648: Z = n_3\alpha + m_3\beta = 0.
649: \ee
650: Since we know from~\eqref{eqantibranes} that $Y$ has to be positive, one stack
651: of branes has to contribute a finite value $0<T<16$ to the tadpole constraint.
652: This amounts to a second equation,
653: \be{eqfin2}
654: Y = n_3(\beta-\alpha)-m_3\alpha = T.
655: \ee
656: To analyse the possibility of an infinite set of solutions to~\eqref{eqfin1}
657: and~\eqref{eqfin2}, we have to distinguish between the cases
658: $n_3=0$ and $n_3\neq 0$.%\\[2ex]
659:
660: \paragraph{$\mathbf{n_3=0}$:}
661: Since $n_3$ and $m_3$ cannot vanish simultaneously, we get from~\eqref{eqfin1}
662: that
663: \be{eqfin3}
664: \beta=n_1(n_2+m_2)+n_2m_1=0.
665: \ee
666: and from~\eqref{eqfin2} we obtain
667: \be{eqfin4}
668: -m_3\alpha=-m_3\left(m_2(m_1+n_1)+n_2m_1\right)=T.
669: \ee
670: An infinite number of solutions can only exist, if there is an infinite series
671: of solutions to $\beta=0$ or $\alpha=\const$. Both cases can be treated
672: analogously, so let us pick one of them and examine $\beta=0$.
673: Again we analyse two cases, depending on the value of $n_1$. If $n_1=0$,
674: we get from~\eqref{eqfin3} that $n_2=0$ and~\eqref{eqfin4} reads
675: $-m_1m_2m_3=T$, which puts bounds on $\{m_i\}$.
676: If $n_1\neq 0$, we can rewrite~\eqref{eqfin3} as
677: \be{eqfin6}\nonumber
678: m_2=-\frac{n_2}{n_1}\left(n_1+m_1\right).
679: \ee
680: Substituting this into~\eqref{eqfin4} leads to
681: \be{eqfin7}\nonumber
682: \frac{m_3n_2}{n_1}\left(n_1^2+m_1^2+n_1m_1\right)=T.
683: \ee
684: To obtain an infinite series, the expression in brackets would have to have an
685: infinite number of solutions. This is not possible, since the term always
686: defines an ellipse, which supports only a finite number of integer--valued
687: points.%\\[2ex]
688:
689: \paragraph{$\mathbf{n_3\neq 0}$:}
690: In this case we can rewrite~\eqref{eqfin1} as
691: \be{eqfin8}
692: \alpha=-\frac{m_3}{n_3}\beta.
693: \ee
694: Substitution in~\eqref{eqfin2} leads to
695: \be{eqfin9}
696: \left(n_3+m_3+\frac{m_3^2}{n_3}\right)\beta=T.
697: \ee
698: The expression in brackets defines again an ellipse and can have only a finite
699: number of solutions. The remaining possibility would be that there exists an
700: infinite series to $\beta=n_1n_2+n_1m_2+n_2m_1=\const$.
701:
702: Let us analyse the different possibilities
703: for $\{n_1,m_1,n_2,m_2\}$. If $n_1$ or $n_2$ are zero, we can see
704: immediately that an infinite series is impossible. If $m_1$ vanishes instead,
705: we obtain $n_1(n_2+m_2)=\const$. If there should exist an infinite series,
706: $n_2$ and $m_2$ have to be unbounded. Using the definition~\eqref{eqfinvar}
707: we deduce that $\alpha=n_1m_2$ has would be unbounded as well.
708: This is only consistent with~\eqref{eqfin8} if $m_3$ would grow beyond all
709: bounds, which is in contradiction to~\eqref{eqfin9}. The argument can be
710: repeated analogously for $m_2$ vanishing.
711:
712: So we are left with the case of $n_1,n_2,m_1,m_2$ all non--vanishing. In this
713: situation we can rewrite $\beta$ as
714: \be{eqfin10}\nonumber
715: n_1n_2\left(1+\frac{m_2}{n_2}+\frac{m_1}{n_1}\right)=\const,
716: \ee
717: which can have only an infinite number of solutions, if $m_2n_1=-m_1n_2$ with
718: $m_1,m_2$ unbounded. But in this case we find $\alpha$ to be unbounded, which
719: is not possible with $\beta$ bounded at the same time.
720: This completes the proof that there can only be a finite number of solutions
721: to the tadpole and supersymmetry conditions.
722:
723:
724:
725: %
726: % STATISTICAL METHODS
727: %
728: \section{Methods of analysis}\label{methods}
729: In the following we describe the computational methods we used to obtain
730: an ensemble of solutions to the tadpole, supersymmetry and K--theory
731: conditions. The results of the
732: statistical analysis are based on this explicitly calculated ensemble.
733:
734: \subsection{Choice of basis}
735: It turns out that it is convenient to use a different basis of three--cycles
736: for the computational analysis, because it makes the tadpole
737: conditions~\eqref{eqRRtadgen} for the bulk cycles and exceptional cycles
738: more uniform.
739: The basis consists of $\mathcal{R}$ even cycles $\eta_i$ and $\mathcal{R}$ odd
740: cycles $\lambda_i$, $i=0,\ldots,5$, which are given in terms of the basis of
741: bulk cycles~\eqref{eqbasisbulkcycles} and
742: exceptional cycles~\eqref{eqbasisexceptionalcycles} for the different
743: geometries as\footnote{Note that in comparison to tables 6 and 7
744: in~\cite{hoot04} we use a slightly different notation. Due to a sign error the
745: cycles $\eta_I$ and $\chi_I$ in the notation of that article have to be
746: exchanged for $I=1,\ldots,5$ (cf. the erratum on p. 33). This we take into
747: account and moreover we will use $\lambda$ instead of $\chi$.}
748: \bea{eqrefbaseetalambda}\nonumber
749: \vec{\eta}&=&\frac{1}{2}\left\{\begin{array}{rl}
750: \AAA: & (\rho_1, -\e_1+2\te_1,
751: -\e_2+2\te_2, -\e_3+2\te_3,
752: \te_4-\e_5+\te_5, \e_4-\e_5),\\
753: \AAB: & (\rho_1+\rho_2, -\e_1+\te_1,
754: -\e_2+\te_2, -\e_3+\te_3,
755: \e_4-\te_5, -\te_4+\e_5),\\
756: \ABA: & (\rho_1+\rho_2, -\e_1+2\te_1,
757: 2\e_2-\te_2, -\e_5+2\te_5,
758: \te_3-\e_4+\te+4, \e_3-\e_4),\\
759: \ABB: & (\rho_2, -\e_1+\te_1,
760: \e_2, -\e_5+\te_5,
761: \e_3-\te_4, -\te_3+\e_4),\\
762: \BBA: & (\rho_2, 2\e_1-\te_1,
763: 2\e_2-\te_2, 2\e_3-\te_3,
764: \te_4-\te_5, \e_4-\te_4+\e_5),\\
765: \BBB: & (-\rho_1+2\rho_2, \e_1,
766: \e_2, \e_3,
767: -\e_4+\te_4-\te_5, \e_4+\e_5),
768: \end{array}\right.\\\nonumber
769: \mbox{and}&&\\\nonumber
770: \vec{\lambda}&=&\frac{1}{2}\left\{\begin{array}{rl}
771: \AAA: & (-\rho_1+2\rho_2, \e_1,
772: \e_2, \e_3,
773: \e_4+\e_5, -\te_4-\e_5+\te_5),\\
774: \AAB: & (-\rho_1+\rho_2, \e_1+\te_1,
775: \e_2+\te_2, \e_3+\te_3,
776: -\te_4-\e_5, -\e_4-\te_5),\\
777: \ABA: & (-\rho_1+\rho_2, \e_1,
778: -\te_2, \e_5,
779: \e_3+\e_4, -\te_3-\e_4+\te_4),\\
780: \ABB: & (-2\rho_1+\rho_2, \e_1+\te_1,
781: \e_2-2\te_2, \e_5+\te_5,
782: -\te_3+\e_4, -\e_3-\te_4),\\
783: \BBA: & (-2\rho_1+\rho_2, -\te_1,
784: -\te_2, -\te_3,
785: \e_4-\e_5+\te_5, -\te_4-\te_5),\\
786: \BBB: & (-\rho_1, \e_1-2\te_1,
787: \e_2-2\te_2, \e_3-2\te_3,
788: \e_4-\e_5, -\te_4+\e_5-\te_5).
789: \end{array}\right.
790: \eea
791: The expansion of a three--cycle in terms of this basis reads
792: \be{eqrefcycle}\nonumber
793: \Pi_a=\vec{r}\cdot\vec{\eta}+\vec{s}\cdot\vec{\lambda}
794: =\sum_{i=0}^5 \left(r_a^i \eta_i + s_a^i\lambda_i\right),
795: \ee
796: with expansion coefficients $r^i, s^i$, $i=0\ldots 5$. The tadpole equations
797: are given by
798: \be{eqreftad}
799: \sum_a N_a \vec{r}_a =4 \vec{r}_{O6},
800: \ee
801: The zeroth entry of $\vec{r}_{O6}$ can be read off from \eqref{eqtadbulk},
802: while all others have to vanish, since the orientifold planes do not contribute
803: to the tadpole equations of the exceptional cycles.
804:
805: In terms of this new basis the intersection between two stacks of branes $a$
806: and $b$ defined by cycles $\Pi_a$ and $\Pi_b$ reads
807: \be{eqisrs}
808: I_{ab}=\Pi_a\circ\Pi_b
809: =\frac{1}{2}\left( \vec{s}_a\cdot\vec{r}_b
810: -\vec{r}_a\cdot\vec{s}_b\right).
811: \ee
812:
813:
814: \subsection{Algorithm}\label{secalg}
815: To obtain a large number of models that fulfil the constraining equations,
816: we used several computers to generate the solutions, which were subsequently
817: stored in a database for later analysis. A priori no constraints have been
818: imposed on the models besides being consistent solutions to the
819: tadpole and supersymmetry conditions.
820:
821: As mentioned before, the model building constraints described in
822: Section~\ref{constraints} can be treated separately for bulk and exceptional
823: cycles.
824: The first part of the computer program we use, which searches for pure bulk
825: configurations, employs the partition algorithm used in~\cite{gbhlw05} to find
826: all possible realisations of the left hand side of equation~\eqref{eqreftad}.
827: Subsequently it runs through a certain range of pairwise coprime wrapping
828: numbers searching for groups $(n_i,m_i),\medspace i=1,2,3$ that yield the
829: desired $r^0$ values. Care has to be taken to avoid multiple counting
830: of cycles which are identified under the orbifold or orientifold action.
831: Explicitly, two of the wrapping numbers are restricted to be always $> 0$ and
832: the wrapping numbers on the third torus, $(n_3,m_3)$ have been chosen to be
833: both odd. In this way no double counting of solutions which are related by
834: a geometric symmetry of the problem will occur.
835: Subsequently the program checks the bulk supersymmetry
836: conditions~\eqref{eqsusybulk} and~\eqref{eqantibranes}, which amount to
837: $r_a>0,\medspace s_a^0=0$ in the notation introduced above.
838: Finally one finds configurations of bulk cycles, which fulfil all consistency
839: conditions, by combining the results of the previous steps.
840:
841: According to tables 23 and 24 in~\cite{hoot04} $128$ exceptional cycles, which
842: already satisfy the supersymmetry conditions, arise for one single bulk cycle.
843: The second part of our program runs through all $128^k$ possible combinations
844: of exceptional cycles for a bulk configuration with $k$ stacks and checks the
845: exceptional tadpole conditions explicitly.
846: Unfortunately there is no way to exclude part of these $128^k$ combinations a
847: priori and we have no choice but to compute every single one of them in order
848: to perform a complete analysis. As a consequence the time necessary for the
849: computation scales exponentially with the number of stacks and could reach the
850: realm of years or even decades. In the following we will thus only present full
851: statistics for models with a low number of stacks. For
852: configurations with a higher number of stacks we randomly select a
853: fraction of the $128^k$ possible combinations.
854: As we will argue in the next section, these randomly chosen subsets can
855: be trusted to resemble the full statistical distributions and are therefore
856: sufficient to make statements about frequency distributions of gauge group
857: properties and chiral matter content.
858:
859:
860:
861: %
862: % RESULTS
863: %
864: \section{Results}\label{results}
865:
866: In the following we present the results of a statistical analysis of the
867: ensemble of solutions to the tadpole, supersymmetry and K--theory constraints,
868: which have been computed as outlined in the last section.
869:
870: As already mentioned before, a full analysis of all possible models is as yet
871: impossible. This comes from the simple fact that the total number of solutions
872: is of the order $10^{28}$, as we are going to show in the following, and an
873: explicit computation of every single solution is beyond reach of contemporary
874: computer technology.
875: Therefore we used the technique of choosing random subsets of possible
876: solutions which in turn were analysed in detail.
877: As it turns out this method is perfectly sufficient for a statistical analysis.
878:
879: After a more detailed explanation of this random method, we
880: discuss the total number of solutions. Then we turn to discuss
881: frequency distributions of various properties of the models, in particular the
882: gauge group factors, the total rank and the chiral matter content.
883: Finally we look for solutions that realise the gauge group of
884: the standard model, discuss their suppression within the set of all solutions
885: and the properties of the hidden sector gauge group.
886:
887: Along the way we compare the results with an analysis of $\ZZ$ models.
888: We only cite the relevant results here, a summary of the statistical analysis
889: that has been done in that case can be found in~\cite{gm06}.
890:
891:
892: \subsection{Choosing random subsets}\label{secrand}
893: The most time--consuming part of computing full solutions is
894: given by adding exceptional cycles to configurations of bulk cycles that
895: already fulfil the tadpole condition.
896: As explained in Section~\ref{secalg}, the bulk solutions are obtained
897: using a fast partition algorithm, while for the exceptional part there is no
898: other way then to run through all $128^k$ possible combinations and check if
899: they fulfil the constrains. This algorithm clearly scales exponential with the
900: number of stacks $k$, such that a complete survey of models with more then
901: three stacks is not feasible.
902:
903: \fig{num_lego}{figlego}{The number of solutions for different numbers of
904: stacks and sizes of the random sample.}
905:
906: Nevertheless, as we will explain shortly, we are able to derive quite
907: robust statistical statements about the full set of solutions.
908: To do so, we apply a procedure to obtain random subsets of the
909: $128^k$ possible ways to add exceptional cycles to a bulk solution. If the
910: total number of solutions is large enough,
911: it is possible to assume a linear dependence between the size of the sample
912: $s$ and the number of solutions $n(s)$.
913: Moreover, the gradient can be used to compute the total number of solutions
914: $n_{tot}(k)$ for a given stack size, if we assume that this number scales with
915: $128^k$.
916:
917: To summarise, we assume that the following equation holds approximately,
918: \be{eqfit}
919: n_{tot}\approx\frac{n(s)}{s}128^k.
920: \ee
921: In Figure~\ref{figlego} the number of solutions for different numbers of stacks
922: and sizes of random samples is shown. Although not clearly visible in this
923: three dimensional plot, the number of solutions grows indeed linearly with the
924: size of the random sample.
925: The accuracy of the linear fit increases with the number of stacks.
926: According to~\eqref{eqfit}, the slope of the logarithmic plot gives the
927: average number of full solutions per
928: bulk configuration, which varies between $10^2$ for the two--stack models and
929: $4.3\times 10^4$ in the case of models with eight stacks.
930:
931: \begin{table}[ht]
932: \begin{center}%
933: \begin{tabular}{|r|r|r|r|}\hline
934: stacks & exact & estimate & error\\\hline
935: $2$ & $1.7068\times 10^6$ & $1.7079\times 10^6$ & $<7\times 10^{-4}$ \\
936: $3$ & $3.9816\times 10^8$ & $3.9818\times 10^8$ & $<6\times 10^{-5}$ \\\hline
937: \end{tabular}%
938: \caption{Exact number of solutions and estimated values for models with two
939: and three stacks of branes and the relative error of the estimate.}
940: \label{tabnumsol}
941: \end{center}
942: \end{table}
943:
944: Using the exact results in the two-- and three--stack case we can compare the
945: total number of solutions with the estimated result from the random procedure.
946: The results of this comparison are shown in Table~\ref{tabnumsol}.
947: It turns out that the estimate
948: is correct up to an error smaller then 0.7\textperthousand~in the case of
949: models with two stacks and even an order of magnitude less in the case of three
950: stacks.
951: Although we cannot completely rule out that something dramatically different
952: happens for models with a larger number of stacks, this seems very unlikely.
953: Our results rather suggest that on the contrary one might conjecture that
954: the estimate gets better for larger stack size $k$.
955: This can be justified given that the deviation from linearity in the scaling
956: gets smaller for larger $k$.
957:
958: It can therefore be expected that the results obtained using the random method
959: are sufficient for a statistical analysis and that we are allowed to
960: extrapolate the frequency
961: distributions obtained for a random sample to the full set of solutions
962: using the relation~\eqref{eqfit}.
963: However, it should be emphasised that a good approximation of the number of
964: solutions is not enough to obtain an accurate description of the properties of
965: the models.
966: Therefore we always perform a check for each distribution against the models
967: with two and three stacks to see if the frequency distributions of the complete
968: solution and the extrapolated distributions from the random samples do agree.
969: In particular for properties of the gauge group we expect the method to work
970: very well, since the gauge group factors depend on the bulk configuration
971: only.
972:
973:
974:
975: \subsection{Total number of solutions}\label{sectotnum}
976: In order to make statistical statements about the probability of certain
977: properties of solutions, it is certainly important to know about how many
978: solutions we are talking.
979: In Figure~\ref{fignumsol} the number of solutions depending on the number
980: of stacks is given. The left figure shows the number of
981: solutions to the bulk equations alone, not including exceptional cycles,
982: while the right figure contains the full result of consistent models.
983: The minimum number of stacks is two in both cases, while the maximum is twelve,
984: which can be deduced immediately from the tadpole equations~\eqref{eqreftad}.
985: Remember that all variables are positive and the maximum value of
986: the right hand side is $24$, while the wrapping number on the left hand side
987: is always a multiple of two.
988:
989: \twofig{num_bulk}{num_total}{fignumsol}{Logarithmic plot of the total number
990: of solutions to the tadpole equations. The left plot (a) shows only bulk
991: solutions, while the right one (b) show the full set of solutions, including
992: exceptional cycles.}
993:
994: Let us begin with an analysis of supersymmetric solutions to the bulk part of
995: the tadpole conditions~\eqref{eqtadbulk} alone. Note that these configurations
996: are just an intermediate step to a full solution, since we need to include
997: exceptional cycles to obtain consistent models.
998: Nevertheless it is an interesting question to ask how many solutions of the
999: bulk equations exist, since this gives an overview of the number of candidate
1000: solutions to the full tadpole and susy constraints. As explained above, we
1001: will have to consider $128^k$ possibilities of configurations of exceptional
1002: cycles for each bulk solution.
1003:
1004: As one can deduce from Figure~\ref{fignumsol_a}, the maximum number of
1005: solutions of possible bulk cycles is obtained for models with $8$ stacks.
1006: In principle one would assume that the number of possible configurations grows
1007: dramatically with the number of stacks, since naively the number of models
1008: with $k$ stacks should be proportional to the number of factorisations of
1009: integer partitions of length $k$.
1010: However, the negative contribution of the orientifold planes to the tadpole
1011: equation is different for the six possible geometries. For the $\AAA$, $\AAB$,
1012: $\BBA$ and $\BBB$ cases we get a total contribution of $16$, while in the
1013: $\AAB$ and $\ABB$ cases we obtain $8$ and $24$, respectively. Keeping in mind
1014: that there is a factor of two on the left hand side of $\eqref{eqtadbulk}$
1015: and that all brane contributions are positive, one finds that the condition
1016: for models with $\AAB$ geometry can only be fulfilled if the number of stacks
1017: is smaller then five. In the case of $\AAA$, $\AAB$, $\BBA$ and $\BBB$ models
1018: with a maximum of eight stacks are possible. This explains the relatively
1019: small contributions for models with more then eight stacks.
1020:
1021: After completing the models with exceptional cycles, the picture changes quite
1022: a bit. This is due to the aforementioned fact that there are in principle
1023: $128^k$ possible configurations of exceptional cycles for each bulk configuration.
1024: Not all of them are consistent, in the sense that they fulfil the full
1025: tadpole equations~\eqref{eqreftad}, but as we have shown in
1026: Section~\ref{secrand} the total number of solutions scales precisely with this
1027: number, multiplied by a coefficient of order $10^2$ to $10^4$. This explains
1028: the domination of models with twelve stacks, that can be seen in
1029: Figure~\ref{fignumsol_b}.
1030: As we will see in the following, this dominance of models with large stack
1031: numbers has a large impact on the statistical distributions.
1032:
1033: Using the randomly generated solutions for all possible numbers of stacks we
1034: can compute the total number of models to be $3.43\times 10^{28} \pm 1\%$.
1035: Since the linearity of the growth of solutions increases with large numbers
1036: of stacks, we can estimate the error in this calculation to be smaller then
1037: the relative error calculated explicitly for the two stack models in the last
1038: section.
1039:
1040: \twofig{numgeo3full}{numgeo3diff}{fignumgeo}{Relative contributions of the
1041: different geometries to the full set of solutions for models with three
1042: stacks. The right figure~(b) shows the relative error between the
1043: random solutions for different stack sizes and the full set of solutions.
1044: The stacks sizes are 4, 16, 64, 256, 1024, 4096 and 16384
1045: (from left to right).}
1046:
1047: To complete the picture, we analyse the individual contributions of the
1048: different geometries. This will also serve as a test of the random method
1049: that we used to obtain the statistical distributions. As can be seen in
1050: Figure~\ref{fignumgeo_a}, the largest contribution comes from the $\BBB$
1051: geometry. Concerning the relative error we make using the random method, it
1052: is found to be sufficiently small. As shown in Figure~\ref{fignumgeo_b},
1053: already at a random sample size of $64$ out of $2^{21}$ combinations of
1054: exceptional cycles, we obtain an error smaller then 1\%.
1055:
1056:
1057:
1058: \subsection{Gauge groups}
1059: We consider two properties of the gauge group of the models, which
1060: consists of a product of $U(N)$, $SO(2N)$ and $Sp(2N)$ groups.
1061: Firstly we analyse the distribution of the total rank, defined as
1062: \be{eqtotrk}
1063: r := \sum_{a=1}^k N_a.
1064: \ee
1065: In a second step we discuss the probability to find one brane with a gauge
1066: group of rank $N$. Both properties are obviously important to
1067: classify models which resemble the standard model.
1068:
1069:
1070: \subsubsection{Rank distribution}\label{sectotrank}
1071: The frequency distribution of the total rank, see Figure~\ref{figgauge_a},
1072: grows exponentially and reaches a maximum at rank $12$.
1073: This behaviour can be explained by the dominance of models with twelve stacks
1074: of branes. The exponential scaling of the total rank is directly related to
1075: the exponential scaling of the total number of solutions, because these are
1076: dominated by models with an $U(1)$ gauge group. This follows from the
1077: solutions to the tadpole equations, which are given as factorisations of
1078: partitions of the orientifold charge. The factor one is not only the number
1079: with the highest abundance in integer partitions, but it is in fact the only
1080: possible gauge group in models with twelve stacks of branes, as follows
1081: directly from the positivity of all variables in the tadpole equation.
1082:
1083: This rank distribution has to be considered with some caution however, since it
1084: includes all possible $U(1)$ factors of the spectrum. It is well known that
1085: some of the $U(1)$s will acquire a mass in the effective theory. This could
1086: happen through a generalised Green--Schwarz mechanism that compensates a
1087: mixed gauge anomaly involving the $U(1)$ in question, but more general cases
1088: are possible. It would be of course very interesting to study the rank
1089: distribution that one obtains after subtracting the massive $U(1)$ factors,
1090: but for the full set of models the necessary computations are not feasible.
1091: In the analysis of models that contain the gauge group of the standard model
1092: in Section~\ref{secsm}, we will make sure that at least one massless $U(1)_Y$
1093: (the hypercharge) exists, but a quantitative statement about the number of
1094: massive $U(1)$s in the general case is beyond the means of our approach.
1095:
1096: \twofig{totrank}{totun}{figgauge}{Frequency distributions of (a)~the total
1097: rank~$r$ and (b)~the probability to find a gauge group of rank~$N$.}
1098:
1099: One striking fact of the rank distribution still has to be explained:
1100: There are only solutions with even rank. This is a consequence of the specific
1101: $\bZ_6$ geometry and different from other orbifold models, as for example the
1102: $\ZZ$ models we already mentioned.
1103: To show why this is always the case, we have to take a closer look at the
1104: tadpole equation~\eqref{eqreftad}. The right hand side is always a multiple
1105: of $4$, depending on the geometry. Therefore we have to have
1106: \be{eqevrk1}
1107: \sum_{a\in A}S_a\equiv 0 \mod 4,\qquad\mbox{with}\quad
1108: S_a:=N_aY_a,\quad A:=\{1,\ldots,k\}.
1109: \ee
1110: We split the sum over $S_a$ into two parts, consisting of only even
1111: and only odd values:
1112: \be{eqevrk2}\nonumber
1113: \sum_{a\in A}S_a=\sum_{a\in O}S_a^{(odd)}+\sum_{a\in E}S_a^{(even)},\qquad
1114: \mbox{with}\quad O\cup E=A,\quad O\cap E=\emptyset.
1115: \ee
1116: The equivalence~\eqref{eqevrk1} can only be fulfilled, if there is an even
1117: number of $S_a^{odd}$. Writing the total rank~\eqref{eqtotrk} as
1118: \be{eqevrk3}
1119: r=\sum_{a\in A}N_a=\sum_{a\in O}N_a+\sum_{a\in E}N_a,
1120: \ee
1121: we get that the first part of this sum is even. Here we used that all branes
1122: in the set $O$ have to obey $N_a\equiv 1\mod 2$, for
1123: $S_a^{(odd)}\equiv 1\mod 2$. For the second part of~\eqref{eqevrk3} to be even,
1124: it is enough to show that $Y_a$ is always odd. This can be done by writing
1125: the wrapping number in terms of the fundamental torus wrapping numbers, similar
1126: to what we did in Section~\ref{finite}. From
1127: \be{eqevrk4}
1128: Y=n_2(\alpha-\beta)-m_2(\beta)\quad,\quad Z=n_2\beta+m_2\alpha=0
1129: \ee
1130: and the constraints $(n_2,m_2)\equiv(1,1)\mod 2$, explained in
1131: Section~\ref{methods}, we obtain from the second equation in~\eqref{eqevrk4}
1132: that $\alpha\equiv\beta\mod 2$ and therefore $Y\equiv1\mod 2$.
1133: This completes the proof.
1134:
1135:
1136: \subsubsection{Gauge group factors}
1137: In Figure~\ref{figgauge_b} the probability to find a gauge group of rank~$N$ is
1138: shown. For the reason explained in the last paragraph, namely the abundance of
1139: $U(1)$ gauge factors, the probability to find one brane with gauge group of
1140: rank one is almost~100\%. The distribution falls off exponentially for larger
1141: $N$, which is again due to the exponential scaling of the number of solutions
1142: with the number of stacks.
1143:
1144: \twofig{rank3comp}{un3comp}{figgaugecomp}{Comparison of the results for the
1145: distribution of~(a) the total rank $r$ and (b)~the probability to find a
1146: rank~$N$ gauge factor for models with three stacks of branes. The full result
1147: is given by the red bars on the left, while the solutions obtained using a
1148: random set of $2^{14}$ exceptional cycles are shown as blue bars on the right.
1149: An upper bound on the relative error is given by the value above each bar.}
1150:
1151: As in the case of the total number of solutions, we have obtained the
1152: distributions using an extrapolation of results from random subsets. To check
1153: the validity of this approach, we compare with the full set of models in
1154: the case of three stacks of branes.
1155: The result is shown in Figure~\ref{figgaugecomp}. For both cases, the total
1156: rank distribution as well as the probability distribution of single gauge
1157: factors, we obtain very accurate results. The relative error is
1158: always smaller then~1\textperthousand in both cases.
1159:
1160:
1161: \subsection{Mean chirality}
1162: To understand on a qualitative level how many of the solutions are chiral, we
1163: analyse the ``mean chirality'' of the set of solutions. To do so, we define
1164: the mean chirality to be the average of chiral representations in each model.
1165: This definition is identical to the one used in the statistical analysis of
1166: $\ZZ$ orbifold models (cf. Section 3.2.2 of~\cite{gm06}).
1167: For a model with $k$ stacks we define the mean chirality $\chi$ as
1168: \be{eqmeanchi}
1169: \chi := \frac{2}{k(k+1)}\sum_{\substack{a,b=0\\a<b}}^k
1170: \left|I_{a'b}-I_{ab}\right|
1171: = \frac{2}{k(k+1)}\sum_{\substack{a,b=0\\a<b}}^k
1172: \left|\vec{s}_a\cdot\vec{r}_b\right|,
1173: \ee
1174: where we used the definition of the intersection $I_{ab}$ in terms of the
1175: $\eta,\lambda$--basis~\eqref{eqisrs}.
1176:
1177: \fourfig{chi3_full}{chi3_64}{chi3_1024}{chi3_16384}{figchi3}{Frequency
1178: distribution of the mean chirality $\chi$ for models with three stacks of
1179: branes. Shown are the full set of solutions (a) and three sets generated using
1180: randomly chosen subsets of 64~(b), 1024~(c) and 16384~(d) out of all
1181: 2097152 possible combinations of exceptional cycles.}
1182:
1183: Before considering all random subsets, we have to make sure that
1184: the method can also be trusted in this case, since we are asking a different
1185: question then in the case of gauge factors or rank distributions.
1186: The definition of the mean chirality~\eqref{eqmeanchi} involves a summation
1187: over intersection numbers. These depend very much on the choice of exceptional
1188: cycles, in contrast to the properties of the gauge group, which depend only on
1189: the configuration of bulk cycles.
1190:
1191: In Figure~\ref{figchi3} we compare the distribution obtained from the full set
1192: of solutions for models with three stacks of branes, including all
1193: $128^3=2^{21}$ possible choices of exceptional cycles, shown in
1194: Figure~\ref{figchi3_a}, with different random subset--models, shown in
1195: Figures~\ref{figchi3_b},~\ref{figchi3_c} and~\ref{figchi3_d}, which take
1196: 64, 1024 and 16384 randomly chosen combinations of exceptional cycles into
1197: account.
1198: Keeping in mind that the plots are logarithmic, one can see that the
1199: qualitative behaviour of the full solution is already captured by the sample
1200: with only 64 randomly chosen exceptional cycles, although we are losing a
1201: good deal of information about models with chirality above $6$.
1202: To obtain a quantitatively satisfying result, it is therefore necessary to
1203: include a bigger subset of cycles. For the highest value of $2^{14}$ random
1204: sets, we get a distribution which differs from the complete result by an
1205: overall error smaller then 1~\textperthousand, comparable to the errors we
1206: found for frequency distributions of gauge group properties.
1207:
1208: \twofig{chi_z6}{chi_z2}{figchi}{Frequency distribution of models with mean
1209: chirality~$\chi$, as defined in~\protect\eqref{eqmeanchi}, for the present
1210: $\bZ_6$ case~(a) to be compared with the result for $\ZZ$~(b).}
1211:
1212: The inclusion of random samples of all possible stack sizes, weighted according
1213: to~\eqref{eqfit}, leads to a frequency distribution as displayed in
1214: Figure~\ref{figchi_a}. Until a value of $\chi\approx 2.8$ the contribution from
1215: models with more than eight stacks dominates. For these models the chirality is
1216: smaller on average, since the $\AAB$ geometry, which allows for solutions with
1217: high chirality is no longer possible.
1218:
1219: The distribution is quite different from what has been found for the
1220: $\ZZ$ orbifold. In that case a general scaling behaviour was discovered, that
1221: has been conjectured in~\cite{bghlw04} based on a saddle point
1222: approximation\footnote{An analysis of the mean chirality distribution based on
1223: explicit, computer--generated data can be found in~\cite{gm06}. We will use
1224: this data, which is more accurate then the estimate of~\cite{bghlw04}, to
1225: compare with the present case.}.
1226: The chirality distribution, displayed in Figure~\ref{figchi_b}, scales to a
1227: quite good approximation as $P(\chi) \sim e^{-3\sqrt{\chi}}$.
1228: In the present case the behaviour is different, especially because the
1229: distribution has two parts that scale differently. The first part goes roughly
1230: like $e^{-(\chi^2)}$, while the second part scales like $e^{-\chi}$.
1231:
1232:
1233: \subsection{Standard model configurations}\label{secsm}
1234: In the following we are going to focus our analysis on a special subclass of
1235: models, namely those which contain the gauge group and the chiral matter
1236: content of the standard model. To be precise, we should speak about the
1237: MSSM here, since all our models are $\cN=1$ supersymmetric.
1238:
1239: In order to simplify the analysis we use the term ``standard
1240: model'' in a very broad sense. In this section a standard model refers to
1241: a consistent solution which contains at least the gauge group and the chiral
1242: matter content of the MSSM.
1243: This means that there always exists a hidden sector, containing additional
1244: gauge groups and chiral matter. This is actually not necessarily bad for
1245: phenomenology, since in the end we need a mechanism to break supersymmetry,
1246: which can be nicely accomplished using a mediation through hidden sector
1247: fields.
1248:
1249: Concerning so--called ``chiral exotics'', i.e. matter that transforms non
1250: trivially under one of the gauge groups of the standard model, we will
1251: distinguish three cases to make our results comparable with the literature.
1252: Case (i) will have no restrictions on exotic matter at all. In case (ii) we
1253: forbid all exotic matter with the exception of bifundamental representations
1254: of the $SU(2)$ group of the standard model and an additional $U(1)$.
1255: These models are those that have been considered in~\cite{hoot04} and might be
1256: of phenomenological relevance, since the bifundamentals can be interpreted as
1257: supersymmetric Higgs particles. However, since they do not transform under the same
1258: $U(1)$ as the weak doublets, it should be expected that the Yukawa couplings
1259: will be non--standard. In the most restrictive case (iii) we do not allow for
1260: any exotic matter at all.
1261:
1262: As has been shown in~\cite{hoot04}, standard model configurations can only
1263: occur if the number of stacks is five or greater. The maximum number of stacks
1264: is nine, since models with more stacks do not support an $SU(3)$ gauge group.
1265: To simplify the analysis we will restrict ourselves to a special type of
1266: embedding of the standard model gauge group and chiral matter, namely the
1267: one introduced in Section~\ref{secsmembed}.
1268:
1269: \twofig{smgen}{smgenex}{figgen}{Distributions of the probability to find
1270: models with the gauge group of the standard model and $g$ generations of
1271: the chiral matter content. Figure~(a) shows the results without any
1272: restrictions on chiral exotics, while in~(b) the amount of bifundamental
1273: matter has been restricted either to maximally one pair transforming in the
1274: $SU(2)$ of the standard model (red bars on the left) or to no additional chiral
1275: matter at all (blue bars on the right).}
1276:
1277:
1278: \subsubsection{Number of generations}
1279: At this point we leave the number of generations of quarks and leptons
1280: as a free parameter. In Figure~\ref{figgen} the frequency distribution of
1281: standard models with different numbers of families is shown.
1282: In Figure~\ref{figgen_a} we allowed all solutions with the gauge group and the
1283: chiral matter content of the standard model, while in Figure~\ref{figgen_b} we
1284: imposed additional constraints to exclude models with chiral exotics, as
1285: outlined above.
1286:
1287: Models with more then two generations have only been found in the cases~(i)
1288: and~(ii), which allow for some amount of non--standard matter. In particular,
1289: there are~$\approx 5.7\times 10^6$ solutions with three generations.
1290: These models all contain five stacks of branes and are of
1291: type~(ii), containing one pair of bifundamental matter that transforms in
1292: the $SU(2)$ of the standard model gauge group and the $U(1)$ coming from the
1293: additional fifth brane. The models are of a type similar to
1294: the ones described in~\cite{hoot04} and contain those as special cases.
1295: We also confirm the statement of that work, that such models can only exist
1296: for the $\AAA$ geometry.
1297:
1298:
1299: \subsubsection{Spectra}
1300: As already mentioned above, we find the spectrum described in~\cite{hoot04}.
1301: It can be described by the following cycles for the five brane stacks,
1302: \bea{eqspec1}\nonumber
1303: \pi_a &=& \frac{1}{2}\left(\rho_1+\rho_2 + \e_1-2\te_1-2\e_2+\te_2+\e_5-2\te_5 \right),\\\nonumber
1304: \pi_b &=& \frac{1}{2}\left(\rho_1+\rho_2-\e_1+2\te_1-2\e_2+\te_2-\e_5+2\te_5\right),\\\nonumber
1305: \pi_c &=& \frac{1}{2}\left(\rho_1+\rho_2+3\e_2-3\te_2-\e_4-\te_4+\e_5+\te_5\right),\\\nonumber
1306: \pi_d &=& \frac{1}{2}\left(\rho_1+\rho_2-\e_1+2\te_1+2\e_2-\te_2-\e_5+2\te_5\right),\\\nonumber
1307: \pi_e &=& \frac{1}{2}\left(\rho_1+\rho_2+3\e_2-3\te_2+\e_4+\te_4-\e_5-\te_5\right).
1308: \eea
1309: The chiral spectrum is given in Table~\ref{tab_smspec1}.
1310:
1311: \begin{table}[ht]
1312: \begin{center}%
1313: \begin{equation*}
1314: \begin{array}{|l|l|c|r|r|r|r|r|r|} \hline
1315: \text{matter}&\text{intersection}&
1316: SU(3)_a \times SU(2)_b & Q_a & Q_b & Q_c & Q_d & Q_e & Q_Y \\\hline
1317: Q_L & I_{ab'}=-3 & (\overline{\3},\2) & -1 & -1 & 0 & 0 & 0 & \frac{1}{6} \\
1318: U_R& I_{ac}=3 & (\3,\1) & 1 & 0 & -1 & 0 & 0 & -\frac{2}{3} \\
1319: D_R & I_{ac'}=3 & (\3,\1) & 1 & 0 & 1 & 0 & 0 & \frac{1}{3} \\
1320: L & I_{bd'}=3 & (\1,\2) & 0 & 1 & 0 & 1 & 0 & -\frac{1}{2} \\
1321: E_R & I_{cd}=3 & (\1,\1) & 0 & 0 & 1 & -1 & 0 & 1\\
1322: N_R & I_{cd'}=-3 & (\1,\1) & 0 & 0 & -1 & -1 & 0 &0 \\\hline
1323: & I_{be}=3 & (\1,\2) & 0 & 1 & 0 & 0 & -1 & 0\\
1324: & I_{be'}=3 & (\1,\2) & 0 & 1 & 0 & 0 & 1 & 0\\
1325: \hline
1326: \end{array}
1327: \end{equation*}
1328: \caption{Chiral spectrum of one of the models with gauge group $SU(3)\times SU(2)\times U(1)^5$ that contains the chiral matter spectrum of the standard model. $Q_Y$ is the hypercharge, realised as a combination of all $U(1)$ factors. Intersection numbers that are not listed are zero. Note the explicit appearance of right--handed neutrinos ($N_R$).}
1329: \label{tab_smspec1}
1330: \end{center}
1331: \end{table}
1332:
1333: In addition we find several variations of this configuration, all very similar
1334: in structure. The difference between all of these models is only given by the
1335: explicit realisation in terms of wrapping numbers. Furthermore the left--handed
1336: quarks might be realised through $I_{ab}=-3$ and $I_{ab'}$ vanishing.
1337: However, this is only a technical detail and does not change the general setup
1338: of the model, which can therefore be seen as the unique construction to obtain
1339: a standard model spectrum on this particular orbifold.
1340:
1341:
1342: \subsubsection{Hidden sector}
1343: The hidden sector of the standard models is generically very small.
1344: This is in sharp contrast to the models on the $\ZZ$ orbifold, where we found
1345: quite large hidden sectors with a distribution of gauge groups that turned out
1346: to be almost identical to the distribution in the full set of
1347: models~\cite{gbhlw05}.
1348: The reason for this is that the number of stacks in the
1349: present case is restricted to a maximum of nine, and the tadpole equations limit
1350: the total rank to be lower than or equal to twelve
1351: (cf. Section~\ref{sectotrank}).
1352:
1353: If we restrict our attention to the group of models which are most interesting
1354: from a phenomenological point of view, namely the three generation models, we
1355: find that they only occur in configurations with five stacks of branes. In
1356: this case the ``hidden sector'' consists only of one $U(1)$ gauge factor and
1357: in addition we always have chiral matter transforming under this $U(1)$ and
1358: the $SU(2)$ group of the standard model.
1359:
1360:
1361:
1362: %
1363: % CONCLUSIONS
1364: %
1365: \section{Conclusions and outlook}\label{conclusions}
1366: In this work we have performed a complete analysis of type II intersecting
1367: D--brane models on the $T^6/\bZ_6$ orientifold. We found that there
1368: exist~$3.4\times 10^{28}$ solutions in total, out of which~$5.7\times 10^6$
1369: contain the gauge group and chiral matter content of the standard model.
1370: We therefore obtained a probability of~$1.7\times 10^{-22}$ to find an
1371: MSSM--like vacuum, a number considerably lower then the value of~$10^{-9}$
1372: that has been calculated in the case of $\ZZ$ orientifolds in~\cite{gbhlw05}.
1373:
1374: The distribution of gauge groups and chiral matter in the full set of solutions
1375: has been analysed and we compared the results with those from a similar study
1376: of $\ZZ$ models. Similar frequency distributions of single gauge group factors
1377: have been found, but the distribution of the total rank of the gauge group and
1378: of the chiral matter content are quite different. This has been explained by
1379: the fact that the branes considered in this work are actually fractional branes
1380: that wrap not only torus cycles, but generically also exceptional cycles around
1381: fixed points of the orbifold. Since there exists a large number of
1382: possibilities to combine these cycles, the number of solutions is considerably
1383: increased and the statistical distributions are altered significantly compared
1384: to the $\ZZ$--case, in which fractional branes have not been considered.
1385:
1386: To obtain the full statistics, a method based on the choice of randomly chosen
1387: subsets of the full solution space has been used. Therefore our results are
1388: not exact, but come with a statistical error, which is however very small and
1389: always below 1\%.
1390:
1391: Concerning future directions, it would certainly be very interesting to
1392: compare our results with other string compactifications that use different
1393: setups.
1394: In particular a better comparison with the heterotic
1395: landscape~\cite{di06,le06} and the statistics of M--theory vacua~\cite{acde05}
1396: would be desirable.
1397: Comparing our results with the extensive analysis of Gepner
1398: models~\cite{dhs04a,dhs04b,adks06} would also be very interesting, although
1399: in this case the analysis is complicated by the fact that we are
1400: considering only one particular geometry over a wide range of (untwisted)
1401: moduli here, whereas the analysis in the
1402: works cited above has been done for a very large set of different geometries
1403: at a particular point in moduli space.
1404: Moreover the Gepner model statistic considers only models which resemble the
1405: standard model gauge group. Nevertheless we hope to come back to this issue
1406: in the future.
1407:
1408: On a more technical level, our analysis of solutions that resemble properties
1409: of the standard model could be improved. Since we discussed only one possible
1410: embedding there might be more possible realisations with interesting
1411: phenomenological features, although most of the embeddings used in different
1412: contexts will not work due to the fact that the number of symmetric and
1413: antisymmetric representations has to be equal.
1414:
1415: Another extension of this work concerns the inclusion of fluxes. In a naive
1416: way this can be done easily by considering a lowered orientifold charge, as
1417: this would generically be the effect of switching on three--form flux.
1418: However, to incorporate the most general NSNS-- and RR--fluxes into an
1419: orientifold setup, it seems very likely that the simple mathematical formalism
1420: used in this article has to be considerably extended.
1421:
1422:
1423: \acknowledgments
1424: We would like to thank Gabriele Honecker for many valuable discussions and
1425: explanations. We acknowledge interesting conversations with Ralph
1426: Blumenhagen, Tim Dijkstra, Bert Schellekens and Chris White.
1427: The work of F. G. is supported by the Foundation for Fundamental Research of
1428: Matter (FOM) and the National Organisation for Scientific Research (NWO).
1429:
1430:
1431: %
1432: % BIBLIOGRAPHY
1433: %
1434: \bibliographystyle{JHEP}
1435: \bibliography{refs_z6}
1436:
1437:
1438: \end{document}
1439: