%alfaro 25/04/03
%urrutia 241102
%urrutia 271102
%marat 190303
%urrutia250303
%urrutia310303
%urrutia020403
%Alfaro, morales y urrutia 040403
\documentclass[12pt] {iopart}
%\documentclass[12pt]{article}
%\hoffset=-1.6cm
%\textwidth=16cm
%\documentstyle[preprint,aps]{revtex}
%\usepackage{iopams}  
\usepackage[dvips]{graphicx}


\newcommand{\be}{\begin{equation}}
\newcommand{\ee}{\end{equation}}
\newcommand{\ba}{\begin{eqnarray}}
\newcommand{\ea}{\end{eqnarray}}
\newcommand{\ua}{{\underbar A}}
\newcommand{\uh}{{\under bar h}}
\newcommand{\ux}{{\underbar X}}
\newcommand{\mb}{\mathbf}


%\documentstyle[aps]{revtex}


\begin{document}
\title{On Non-Abelian Holonomies }
\author{J. Alfaro \dag \footnote[3]{jalfaro@puc.cl}
,H.A. Morales-T\'ecotl\ddag,M. Reyes \dag \ and  L.F. Urrutia\P}

\address{\dag Facultad de F\'\i sica\\
Pontifica Universidad Cat\'olica de Chile\\
Casilla 306, Santiago 22, Chile }
\address{\ddag Departamento de F\'\i sica\\
Universidad  Aut\'onoma Metropolitana-Iztapalapa\\
A. Postal 55-534, M\'exico D.F. 09340, M\'exico\\
}
\address{\P Departamento de F\'\i sica de Altas Energ\'\i as\\
Instituto de Ciencias Nucleares\\
Universidad Nacional Aut\'onoma de M\'exico\\
A. Postal 70-543, M\'exico D.F. 04510, M\'exico \\}
%\maketitle



\begin{abstract}
We provide a method and the results  for the calculation of  the holonomy of a Yang-Mills connection in 
an  arbitrary triangular path, in an expansion (developed here to fifth order) in powers of the corresponding 
segments. The results might have applications in generalizing to Yang-Mills fields previous calculations  of  
the corrections to particle dynamics induced by loop quantum gravity,  as well as in the field of random lattices.
\end{abstract}

\pacs{04.60.Pp, 11.55.Fv, 12.38.Gc}

\maketitle
%\draft

%\clearpage

\section{Introduction}


Constraints in  Lorentz covariance violations have been experimentally studied since a
long time ago
\cite{HUDR} by obtaining observational bounds upon the violating parameters. Recent experiments have shown an impressive increase in their sensitivities  thus producing even more stringent bounds \cite{Exper2}.

In order to correlate such experimental results, Kostelecky and collaborators
have proposed a phenomenological extension of the standard model,
which incorporates the most general Lorentz-violating interactions compatible with power counting renormalizability together with the particle structure of the standard model \cite{kostelecky1}. An impressive  number of applications to very different processes have been already considered,
as can be seen in Ref. \cite{CPTPROC}, for example.

Different models accounting for minute Lorentz violations have recently arisen in
the context
of quantum gravity induced corrections to the propagation and interactions of
particles \cite{Gambini,amufot,Us,Ellis,QuantFluct}. This amounts to  realize the generic violating
parameters appearing in the standard model
extension in terms of specific quantities involving the Planck length $\ell_P$ together with additional  physically relevant objects. Moreover, the high precision  obtained in the
determination of the experimental bounds has brought  quantum gravity
induced effects to the level of observable phenomena \cite{BILLER,SUV,ap,observ}.

On the other hand, in the early 80's, the growing importance of computer simulations of gauge theories
required a short distance cutoff of geometrical origin such as a lattice. However regular lattices
break essential symmetries of continuum theories such as translational and rotational invariance.
Motivated by the need of maintaining these symmetries, the field theory on a random lattice was suggested
\cite{cfl}. Further on, the connection of random lattices with quantum gravity and strings
was studied and low dimensional systems on random lattices were solved using
matrix model techniques \cite{jan}.

In this work we concentrate on some aspects arising in the process of generalizing
the loop quantum gravity inspired model described
in \cite{amufot,Us}  to Yang-Mills fields, in order to obtain  the non-Abelian generalization of the
corrections previously found for the dynamics of
photons. Namely, corrections to standard matter dynamics are obtained by means of calculating non-Abelian holonomies, either of gravitational or Yang-Mills type, around triangular paths. To this end
we have to revise  and extend the procedure of Ref.\cite{FGK} that was applied to the case of rectangular cells.
It turns out that the method we present in this paper contains as a particular case
the result of Ref.\cite{FGK}, though it is applicable to arbitrary cells made up of triangles.
The basic building block in our analysis is the holonomy along a straight line
segment which path order property we consistently keep to all orders in our expansion.

The problem we deal with here is closely related to the non Abelian Stokes theorem which has been repeatedly
discussed in the literature \cite{FGK,STOKES}.


This paper is organized as follows: in Section 2 we state the problem to be dealt with and introduce some 
notation. Section 3 summarizes  the results for the Abelian case which we intend to generalize here. 
The non-Abelian case is  subsequently discussed in Section 4 which contains our main results. 
Using the procedure  of Ref. \cite{FGK} the corresponding calculation  is performed in Section 5, 
which allows us to show some discrepancies  that arise  between the two methods. Finally we close with  
a summary and discussion in Section 6. For the sake of completeness we have included some intermediate 
steps of the
calculations in  Appendices A and B.

Part of the computations involved in deriving the results of section 4 and 5 were done using FORM\cite{form}.

\section{Statement of the problem}

The proposed method to obtain the quantum gravity induced corrections to the Yang-Mills Lagrangian
requires the calculation of the object
\begin{equation}
\label{TGHOL}
T_\rho=tr\left({\mathbf G}_\rho\,{\mathbf h}_{\alpha_{IJ}} \right),
\end{equation}
where ${\mathbf G}_\rho$ are the generators of the corresponding Lie algebra and ${\mathbf h}_{\alpha_{ij}(\Delta)}$ is the holonomy of the Yang-Mills connection ${\mathbf  A}_a=A_a^\rho\,{\mathbf G}_\rho$ in the triangle $\alpha_{IJ}$, with vertex  $v$, defined
by the vectors ${\vec s}_I$ and ${\vec s}_J$, arising from the vertex $v$ in the way described below.

 Our  main task will be to construct an expansion of $T_\rho$ in powers of the segments $s^a_I, \,
s^b_J$.

To be more precise, we have
\be
{\mathbf h}_{\alpha_{IJ}}=P\,exp\left({\oint_{\alpha_{IJ}} {\mathbf A}_a({\vec x}(s))\,\frac{dx^a}{ds}\,ds}\right),
\ee
where $P$ is a path-ordered product to be specified later.
As shown in Fig.1, the closed path $\alpha_{IJ}$, parameterized by ${\vec x}(s)$, is defined in the following  way: we start from the vertex $v$ following a straight line with the direction and length of ${\vec s}_I$, then follow another straight line in the direction and length of ${\vec s}_J-{\vec s}_I$, and finally return to $v$ following $-{\vec s}_J$. From the definition of the holonomy, we have the transformation property
\be
{\mathbf h}_{\alpha_{IJ}}\rightarrow  {\mathbf U}(v)\,{\mathbf h}_{\alpha_{IJ}}\,{\mathbf U}(v)^{-1},
\ee
under a gauge transformation of the connection, where ${\mathbf U}(v)$ is a group element evaluated at the vertex $v$. In other words, ${\mathbf h}_{\alpha_{IJ}}$  transforms covariantly under the group.



\section{The Abelian case}

The corresponding calculation was performed in Ref.\cite{amufot} and 
here we summarize the results in order to have the
correct expressions to which the non-Abelian result must reduce when taking the commuting limit. In this case 
Eq.(\ref{TGHOL}) reduces to
\be
T=exp(\Phi_{IJ})-1 ,
\ee
where $\Phi_{IJ}$ is the magnetic flux through the area of the triangle, given by
\begin{eqnarray}
\Phi ^{B}(F_{IJ})&=&\oint_{\alpha _{IJ}}dt\;\dot{s}^{a}(t)A_{a}(t) \nonumber \\
&=&\int_{{\vec{v}}}^{{\vec{v}}+{\vec{s}}_{I}}{{A}}%
_{a}\,dx^{a}+\,\int_{{\vec{v}}+{\vec{s}}_{I}}^{{\vec{v}}+{\vec{s}}_{J}}{%
{A}}_{a}\,dx^{a}+\int_{{\vec{v}}+{\vec{s}}_{J}}^{{\vec{v}}}{%
{A}}_{a}\,dx^{a}, \label{STOKES}
\end{eqnarray}
where the connection $A_a({\vec x}(s))$ is now a commuting object.

The basic building block in (\ref{STOKES}) is
\begin{eqnarray}
\int_{{\vec{v}}_{1}}^{{\vec{v}}_{2}}{{A}}_{a}({\vec{x}})\,dx^{a}
&=&\int_{0}^{1}{{A}}_{a}({\vec{v}}_{1}+t\,({\vec{v}}_{2}-{\vec{v}}%
_{1}))\,({\vec{v}}_{2}-{\vec{v}}_{1})^{a}\,dt  \nonumber \\
&=&\int_{0}^{1}{{A}}_{a}({\vec{v}}_{1}+t\,\vec{\Delta})\,\Delta
^{a}\,dt  \nonumber \\
&=&\left( 1+\frac{1}{2!}\Delta ^{b}\partial _{b}+\frac{1}{3!}(\Delta ^{b}\partial
_{b})^{2}+\dots \right) \Delta ^{a}{A}_{a}(v),
\end{eqnarray}
with$\;\Delta ^{a}=({\vec{v}}_{2}-%
{\vec{v}}_{1})^{a}$. The infinite series in parenthesis is
\begin{equation}
F(x)=1+\frac{1}{2!}x+\frac{1}{3!}x^{2}+\frac{1}{4!}x^{3}+\dots =\frac{e^{x}-1%
}{x},  \label{EFE}
\end{equation}%
yielding
\be
\int_{{\vec{v}}_{1}}^{{\vec{v}}_{2}}{{A}}_{a}({\vec{x}}%
)\,dx^{a}=F(\Delta ^{a}\,\partial _{a})\left( \Delta ^{a}\,{{A}}%
_{a}({\vec{v}}_{1})\right) .\quad
\label{PATHI}
 \ee
In the following we employ the notation $\Delta
^{a}\,V_{a}={\vec{\Delta}}\cdot {\vec{V}}$. Using the above result in the three
integrals appearing in (\ref{STOKES}) and after some algebra, we obtain
\begin{eqnarray}
\Phi ^{B}(F_{IJ}) &=&F_{1}({\vec{s}}_{I}\cdot {\nabla},{\vec{s}}%
_{J}\cdot {\nabla})\,\,s_{J}^{a}\,s_{I}^{b}\,\left( \partial _{a}\,{%
{A}}_{b}({\vec{v}})-\partial _{b}\,{{A}}_{a}({\vec{v}}%
)\right) \nonumber \\
&=&F_{1}({\vec{s}}_{I}\cdot {\nabla},{\vec{s}}_{J}\cdot {\nabla}%
)\,\,s_{J}^{a}\,s_{I}^{b}\,\epsilon _{abc}B^{c}(v),
\end{eqnarray}%
where the gradient acts upon the coordinates of ${\vec{v}}$. The function $%
F_{1}$ is \be
F_{1}(x,y)=\frac{y({\rm e}^{x}-1)-x({\rm e}^{y}-1)}{x\,y\,(y-x)}%
=-\sum_{n=1}^{\infty }\frac{1}{(n+1)!}\,\frac{x^{n}-y^{n}}{x-y}. \ee
Let us emphazise
that $F_{1}(x,y)$ is just a power series in the variables $x$ and $y$. Expanding in powers of the
 segments $s_{I}^{a}$ we obtain
\begin{eqnarray}
\Phi ^{B}(F_{IJ}) &=&\left( 1\,\,+\frac{1}{3}(s_{I}^{c}+s_{J}^{c})\,\partial
_{c}\,+\frac{1}{12}(s_{I}^{c}\,s_{I}^{d}+s_{I}^{c}\,s_{J}^{d}+s_{J}^{c}%
\,s_{J}^{d})\,\,\partial _{c}\,\partial _{d}+...\right) \times  \nonumber
 \\
&&\times \,\frac{1}{2}s_{I}^{a}s_{J}^{b}\epsilon _{abc}B^{c}(v).
\label{FLUX}
\end{eqnarray}%
Notice that the combination%
\be \frac{1}{2}s_{I}^{a}s_{J}^{b}\epsilon _{abc}={\cal A}\;n_{c},
\ee
is just the oriented area of the triangle with vertex $v$ and sides $%
s_{I}^{c},\;s_{J}^{c}$, joining at this vertex, having value ${\cal A}$ and unit
normal$\;$vector$\;n_{c}$.

To conclude we have to calculate
\ba
 \label{WLC}
\left(
e^{{{\Phi}}^{B}(F_{IJ}(\Delta))}-1\right) &=& \sum_{n=2}^{\infty}
\frac{1}{n!} ({{\Phi}}^{B}(F_{IJ}))^n= \sum_{n=2}^{\infty}
M_{nIJ(\Delta)},
\ea
where the subindex $n$ labels the corresponding power in the vectors $s^a$. 
The results are
\ba \label{MJKD2}
M_{2IJ(\Delta)}&:=&  s^a_I s^b_J \frac{1}{2!}  {F}_{ab}, \\
\label{MJKD3}
M_{3IJ(\Delta)}&:=& s^a_I s^b_J \frac{1}{3!} (x_I+x_J)  {F}_{ab}, \\
\label{MJKD4}
M_{4IJ(\Delta)}&:=&s^a_I s^b_J  \frac{1}{4!} (x_I^2+x_Ix_J+x_J^2)  {F}_{ab}
+ s^a_I s^b_Js^c_Is^d_J \frac{1}{8}  {F}_{ab}                                 {F}_{cd},\\
\label{MIJD5}
M_{5IJ(\Delta)}:&=&\frac{1}{5!}(x_I^3+x_J^3+x_I^2x_J+x_Ix_J^2)s_I^as_J^b F_{ab}
\nonumber\\ 
&&+\frac{ s^a_I s^b_Js^c_I s^d_J}{4!}\left[(x_I+x_J){F}_{ab}{F}_{cd}+{F}_{ab}(x_I+x_J){F}_{cd}\right]
\ea
up to fifth order. We are using the notation $x_I={\vec
s}_I\cdot\nabla=s_I^a\,\partial_a$.

We expect that the non-Abelian generalization of the  quantities (\ref{MJKD2}), (\ref{MJKD3}), (\ref{MJKD4}), (\ref{MIJD5}) is produced by the replacement
\ba
&& A_a \rightarrow {\mathbf A}_a=A^\rho_a\,G_\rho, \quad \partial_a \rightarrow {\mathbf D}_a=
\partial_a-[{\mathbf A}_a , {\ } ] \\
&& F_{ab} \rightarrow {\mathbf F}_{ab}=
 \partial_a{\mathbf A}_b-\partial_b{\mathbf A}_a-[{\mathbf A}_a, {\mathbf A}_b]
\ea
Nevertheless, at this level there are potential ordering ambiguities which will be resolved in the next sections.

\section{The non-Abelian case}

In a similar way to the Abelian case we separate the calculation of the holonomy ${\mb{h}}_{\alpha_{IJ}}$ in three basic pieces through the straight lines along the sides of the triangle ${\alpha_{IJ}}$. We have
\be
{\mb{h}}_{\alpha_{IJ}}=P(e^{L_3})P(e^{L_2})P(e^{L_1})\equiv \,U_3\,U_2\,U_1,
\label{RHOL}
\ee 
where
\ba
&&L_1=\int_0^ 1 dt\, {\mb{A}}_a({\vec v} + t \, {\vec s}_I)\, s^a_I\\
&&L_2=\int_0^ 1 dt\, {\mb{A}}_a({\vec v}+ {\vec s}_I + t \, ({\vec s}_J-{\vec s}_I))\,(s^a_J-s^a_I)\\
&&L_3=\int_0^ 1 dt\, {\mb{A}}_a({\vec v}+ {\vec s}_J - t \, {\vec s}_J)\,(- s^a_J)
\ea

Here we have parameterized each segment with $0 \leq t \leq 1$.

\subsection{The basic building block}

Let us consider in detail the contribution
\be
U_1=P(e^{L_1}), \qquad L_1=\int_0^ 1 dt\, {\mb{A}}_a({\vec v} + t \, {\vec s_I})\, s_I^a,
\ee
with ${\vec s_I}= \{ s_I^a \}$.

Using the  definition
\begin{eqnarray}
U_1&=&1+\int_{0}^{1}dt {\mathbf A}_{a}({\vec v}+t{\vec s_I})s_I^{a}+\int_{0}^{1}dt\int_{0}^{t}dt'{\mathbf A}_{a}({\vec v}+t{\vec s_I}){\mathbf A}_{b}({\vec v}+t'\vec s_I)s_I^{a}s_I^{b}{}
\nonumber\\
&&{}+\int_{0}^{1}dt\int_{0}^{t}dt'\int_{0}^{t'}dt''{\mathbf A}_{a}{\mathbf A}_{b}{\mathbf A}_{c}s_I^{a}s_I^{b}s_I^{c}+ \dots,
\end{eqnarray}
for the path ordering, we arrive at the following expression
\begin{eqnarray}
\label{U1}
U_1&=&1+I_{1}(x){\mathbf A}_{a}(v)s_I^{a}+I_{2}(x,\bar{x}){\mathbf A}_{a}(v)\bar{{\mathbf A}}_{b}(v)s_I^{a}s_I^{b}\nonumber \\
&&+I_{3}(x,\bar{x},\bar{\bar{x}}){\mathbf A}_{a}(v)\bar{{\mathbf A}}_{b}(v){\bar{\bar{{\mathbf A}}}}_c(v)s_I^{a}s_I^{b}s_I^{c}+ \dots
\end{eqnarray}
Here we are using  the conventions
\begin{eqnarray}
&&x=s^{c}_{I}\partial_{c}\qquad \bar{x}=s^{c}_{I}\bar{\partial}_{c} \qquad \bar{\bar{x}}=s^{c}_{I} \bar{\bar{\partial}}_{c}\\
&&I_{1}=F(x), \qquad
I_{2}=\frac{F(x+\bar{x})-F(x)}{\bar{x}}\\
&&I_{3}=\frac{1}{\bar{\bar{x}}}\left[\frac{1}{\bar{x}+\bar{\bar{x}}}( F(x+\bar{x}+\bar{\bar{x}})-F(x))-\frac{1}{\bar{x}}(F(x+\bar{x})-F(x))  \right]
\end{eqnarray}
with  $F(x)$ given by Eq.(\ref{EFE}).
The notation in Eq. (\ref{U1}) is that  each operator
 $x,\bar{x},\bar{\bar{x}}$ acts only in the corresponding field $A,\bar{A},\bar{\bar{A}}$ respectively.
We write
\be
U_1=\sum_N\, U_{1}^{(N)},
\ee
where the superindex  $N$ indicates the powers of $s_I^a$ contained in each term. A detailed calculation produces
\begin{equation}
U_{1}^{(1)}=s_{I}^{a}{\mb A}_{a}, \qquad U_{1}^{(2)}=\frac{1}{2}(x s_{I}^{a}{\mb A}_{a}+ s_{I}^{a}s_{I}^{b}{\mb A}_{a}{\mb A}_{b})
\end{equation}
\begin{equation}
U_{1}^{(3)}=\frac{1}{3!}(x^{2} s_{I}^{a}{\mb A}_{a}+(\bar{x}+2x)s_{I}^{a}s_{I}^{b}{\mb A}_{a}\bar{{\mb A}_{b}}+s_{I}^{a}s_{I}^{b}s_{I}^{c}{\mb A}_{a}{\mb A}_{b}{\mb A}_{c})
\end{equation}
\begin{eqnarray}
U_{1}^{(4)}&=&\frac{1}{4!}\left[x^{3}s_{I}^{a}{\mb A}_{a}+(3x^{2}+3x\bar{x}+\bar{x}^{2})s_{I}^{a}s_{I}^{b}{\mb A}_{a}\bar{{\mb A}_{b}}+(3x+2\bar{x}
+\bar{\bar{x}})s_{I}^{a}s_{I}^{b}s_{I}^{c}{\mb A}_{a}\bar{{\mb A}_{b}}\bar{\bar{{\mb A}_{c}}}\right.\nonumber \\
&&\qquad  \left.+ s_{I}^{a}s_{I}^{b}s_{I}^{c}s_{I}^{d}{\mb A}_{a}{\mb A}_{b}{\mb A}_{c}{\mb A}_{d}\right]
\end{eqnarray}
Specializing to the case ${\vec s}_I=(a,0,0)$ and to third order in $a$ we obtain
\be
\label{COMP}
U_{1}=1+a{\mb A}_{1}+\frac{1}{2}a^{2}(\partial_{1}{\mb A}_{1}+{\mb A}_{1}^{2})+\frac{1}{3!}a^{3}(\partial^{2}_{1}{\mb A}_{1}+{\mb A}_{1}\partial_{1}{\mb A}_{1}
+2(\partial_{1}{\mb A}_{1}){\mb A}_{1}+ {\mb A}_1^3).
\ee
Next we compare our result (\ref{COMP}) with the calculation according to the method of Ref.\cite{FGK}. Using the equation (3.15a) of such reference  for $L_1$   we obtain
\begin{equation}
e^{L_{1}}=1+a{\mb A}_{1}+\frac{1}{2}a^{2}(\partial_{1}{\mb A}_{1}+{\mb A}_{1}^{2})+\frac{1}{3!}a^{3}\left(\partial^{2}_{1}{\mb A}_{1}+\frac{3}{2}
\left(\frac{}{} {\mb A}_{1}\partial_{1}{\mb A}_{1}+(\partial_{1}{\mb A}_{1}){\mb A}_{1}\right)+{\mb A}_1^3\right)
\label{U1FGK}
\end{equation}
which does not agree with the expression (\ref{COMP}). This basic discrepancy propagates when combining more segments and ultimately it is the source of  the differences between our results and those obtained via the application of the methods in Ref.\cite{FGK}.

It is interesting to remark that in  the Abelian limit both (\ref{COMP}) and (\ref{U1FGK}) reduce to
\be
\label{COMPA}
U_{1}=1+a{ A}_{1}+\frac{1}{2}a^{2}(\partial_{1}{ A}_{1}+{ A}_{1}^{2})+\frac{1}{3!}a^{3}(\partial^{2}_{1}{ A}_{1}+ 3{ A}_{1}\partial_{1}{ A}_{1} + A_1^3)
\ee
which is obtained from a direct calculation using the expression (\ref{PATHI}).



\subsection{The holonomy ${\mb{h}}_{\alpha_{IJ}}$ }

Now we put the remaining pieces together in order to calculate  ${\mb{h}}_{\alpha_{IJ}}=U_3U_2U_1$. Using the notation
\be
y=s^a_J\,\partial_a
\ee
and starting from the basic structure (\ref{U1}) we obtain, mutatis mutandis,


\begin{eqnarray}
U_{2}^{(1)}&=&(s_{J}^{a}-s_{I}^{a}){\mb A}_{a}, \\
U_{2}^{(2)}&=&\frac{1}{2}[(x+y) (s_{J}^{a}-s_{I}^{a}){\mb A}_{a}+ (s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b}){\mb A}_{a}{\mb A}_{b}], \\
U_{2}^{(3)}&=&\frac{1}{3!}[(x^{2}+y^{2}+xy) (s_{J}^{a}-s_{I}^{a}){\mb A}_{a}+(x+2y+\bar{y}+2\bar{x})(s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b}){\mb A}_{a}\bar{{\mb A}_{b}}{}
\nonumber\\
&&{}+ (s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})(s_{J}^{c}-s_{I}^{c}){\mb A}_{a}{\mb A}_{b}{\mb A}_{c})], \\
U_{2}^{(4)}&=&\frac{1}{4!}[(x^{3}+y^{3}+x^{2}y+xy^{2})(s_{J}^{a}-s_{I}^{a}){\mb A}_{a}+(x\bar{y}+3x\bar{x}+x^{2}+2xy+2\bar{x}\bar{y}+3\bar{x}^{2}{}
\nonumber\\
&&{}+3y^{2}+\bar{y}^{2}+3y\bar{y}+5\bar{x}y)(s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b}){\mb A}_{a}\bar{{\mb A}_{b}}+(x+2\bar{x}+3\bar{\bar{x}}+2\bar{y}+\bar{\bar{y}}+3y){}
\nonumber\\
&&{}\times (s_{J}^{a}-s_{I}^{a})   (s_{J}^{b}-s_{I}^{b}) (s_{J}^{c}-s_{I}^{c}){\mb A}_{a}\bar{{\mb A}_{b}}\bar{\bar{{\mb A}_{c}}}+(s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})\nonumber\\
&&(s_{J}^{c}-s_{I}^{c})(s_{J}^{d}-s_{I}^{d}){\mb A}_{a}{\mb A}_{b}{\mb A}_{c}{\mb A}_{d}]
\end{eqnarray}
for $U_2$, together with
\begin{eqnarray}
U_{3}^{(1)}&=&-s_{J}^{a}{\mb A}_{a}, \\
U_{3}^{(2)}&=&\frac{1}{2}(-y s_{J}^{a}{\mb A}_{a}+ s_{J}^{a}s_{J}^{b}{\mb A}_{a}{\mb A}_{b}), \\
U_{3}^{(3)}&=&\frac{1}{3!}(-y^{2} s_{J}^{a}{\mb A}_{a}+(2\bar{y}+y)s_{J}^{a}s_{J}^{b}{\mb A}_{a}\bar{{\mb A}_{b}}-s_{J}^{a}s_{J}^{b}s_{J}^{c}{\mb A}_{a}{\mb A}_{b}{\mb A}_{c}), \\
U_{3}^{(4)}&=&\frac{1}{4!}
\left[-y^{3}s_{J}^{a}{\mb A}_{a}+(3\bar{y}^{2}+3y\bar{y}+y^{2})s_{J}^{a}s_{J}^{b}{\mb A}_{a}\bar{{\mb A}_{b}}-(3\bar{\bar{y}}+2\bar{y}+y)s_{J}^{a}s_{J}^{b}s_{J}^{c}{\mb A}_{a}\bar{{\mb A}_{b}}\bar{\bar{{\mb A}_{c}}}\right. \nonumber \\
&& \qquad \qquad \left. +s_{J}^{a}s_{J}^{b}s_{J}^{c}s_{J}^{d}{\mb A}_{a}{\mb A}_{b}{\mb A}_{c}{\mb A}_{d}\right],
\end{eqnarray}
for $U_3$. Let us emphasize that in all the expressions above for $U_1$, $U_2$ and $U_3$, the connection is evaluated at the vertex $v$. The bars only serve to indicate the position in which the corresponding  derivative acts.

Next we write the contributions to  the holonomy in powers of the segments. Some further details are presented in the Appendix A. We obtain
\be
{\mb{h}}_{\alpha_{IJ}}^{(2)}=\frac{1}{2}s_{I}^{a}s_{J}^{b}\, {\mb F}_{ab},
\label{2ORDHOL}
\ee
\begin{equation}
{\mb{h}}_{\alpha_{IJ}}^{(3)}=\frac{1}{3!}(s^{c}_I+s^{c}_J)s_{I}^{a}s^{b}_{J}{\mb D}_{c}{\mb F}_{ab},
\label{3ORDHOL}
\end{equation}

\begin{equation}
{\mb{h}}_{\alpha_{IJ}}^{(4)}=\frac{1}{4!}(s^{c}_Is^{d}_I+s_{I}^{c}s^{d}_{J}+s_{J}^{c}s^{d}_{J})s_{I}^{a}s^{b}_{J}{\mb D}_{c}{\mb D}_{d}{\mb F}_{ab}
+\frac{1}{8}s_{I}^{a}s_{J}^{b}s_{I}^{c}s_{J}^{d}{\mb F}_{ab}{\mb F}_{cd}.
\label{4ORDHOL}
\end{equation}


\begin{eqnarray}
{\mb{h}}_{\alpha_{IJ}}^{(5)}&=&\frac{1}{5!}(s^{c}_Is^{d}_Is^{e}_I+s_{J}^{c}s^{d}_{J} s^{e}_J+2( s_{J}^{c}s^{d}_{I}s^{e}_I+s_{I}^{c}s^{d}_{J}s^{e}_J)-s_{I}^{c}s^{d}_{J}s^{e}_I-s_{J}^{c}s^{d}_{I}s^{e}_J)s^a_Is^b_J {\mb D}_c{\mb D}_d{\mb D}_e    \mb{F}_{ab}\nonumber\\&&
+ \frac{1}{4!}s^a_Is^b_Js^d_Is^e_J ( \mb{ F}_{ab} ( s_{I}^{c}+s^{c}_{J} ){\mb D}_c {\mb F}_{de}+(s_{I}^{c}+s^{c}_{J} )({\mb D}_c \mb{ F}_{de})\mb{ F}_{ab})
\label{5ORDHOL}
\end{eqnarray}


Eq. (\ref{4ORDHOL}) resolves the ordering ambiguity which apparently arised in covariantizing the first term in the RHS of Eq.(\ref{MJKD4}). Nevertheless, as we subsequently show there is really no such ambiguity at this order. Let us consider the combination
\ba
s_{I}^{a}s^{b}_{J}s_{I}^{c}s^{d}_{J} \left(  {\mb D}_{c}{\mb D}_{d}-  {\mb D}_{d}{\mb D}_{c}\right){\mb F}_{ab}&=&s_{I}^{a}s^{b}_{J}s_{I}^{c}s^{d}_{J}\, \left[ {\mb D}_{c},  {\mb D}_{d}\right]{\mb F}_{ab}\nonumber \\
&=&-s_{I}^{a}s^{b}_{J}s_{I}^{c}s^{d}_{J}\, \left[ {\mb F}_{cd},{\mb F}_{ab} \right]=\left[{\mb {F}},{\mb F} \right]=0
\ea
where we have used the notation ${\mb F}=s_{I}^{a}s^{b}_{J}{\mb F}_{ab}$ together with the property
\ba
\left[ {\mb D}_{c},  {\mb D}_{d}\right]{\mb G}=-\left[{\mb F}_{cd},{\mb G} \right]
\ea
valid for any object ${\mb G}$ in the adjoint representation.

The results (\ref{2ORDHOL}), (\ref{3ORDHOL}) ,  (\ref{4ORDHOL}) and (\ref{5ORDHOL}), which we have obtained by direct calculation, constitute in fact the unique gauge covariant generalization of the corresponding Abelian expressions
(\ref{MJKD2}), (\ref{MJKD3}),  (\ref{MJKD4}) and (\ref{MIJD5}). This provides a strong support to our method of calculation.

\section{ The holonomy according to the method of Ref.[16]}


In this section we calculate the holonomy ${\mb{h}}_{\alpha_{IJ}}$ using the method of Ref. \cite{FGK} adapted for our case of three segments: $L_1, L_2, L_3$. From Eqs. (3.1),  (3.3), (3.4), (3.12) and (3.14), with $\lambda=1$,  of that reference it follows that
 \begin{equation}
 {\mb{h}}_{\alpha_{IJ}}=e^{L_3}e^{L_2}e^{L_1}=e^{ \Sigma+\sum_{n=2}F_{n}} =e^{H}
\label{EQH}
\end{equation}
The basic building blocks are
\begin{equation}
L_{1}=s^{a}_{I}F(x){\mb A}_{a}(v), \quad
L_{2}=(s^{a}_{J}-s^{a}_{I})F(y-x)e^{x}{\mb A}_{a}(v), \quad
L_{3}=-s^{a}_{J}F(y){\mb A}_{a}(v).
\label{BBLOCK}
\end{equation}
where the vertex $v$ generalizes de point $(x_1, x_2)$ in the notation of
\cite{FGK}. The calculational method indicated in Eqs.(\ref{EQH}) and (\ref{BBLOCK}) is clearly not equivalent to the correct prescription (\ref{RHOL}): ${\mb h}_{\alpha_{IJ}}=U_3\,U_2\,U_1$ with
the $U$'s given by Eq.(\ref{U1}) together with the corresponding extensions that take into account the change of the starting point in the corresponding path. 

Let us write here those expressions arising from the method in Ref. \cite{FGK} that we will use in our calculation
\ba
\Sigma&=&L_{1}+L_{2}+L_{3}, \label{SIGMA}\\
2!F_{2}&=&[(L_{3}+L_{2}), L_{1}]+[L_{3},L_{2}], \label{F2}\\
3!F_{3}&=&-[\Sigma,F_{2}]+[(L_{3}+L_{2})^{2} L_{1}]+[L_{3}^{2}L_{2}], \label{F3}\\
4!F_{4}&=&-3![\Sigma,F_{3}]-[\Sigma^{2}F_{2}]+[(L_{3}+L_{2})^{3} L_{1}]+[L_{3}^{3}L_{2}]
\label{F4}\ea
with the notation  $[A^{2}B]=[A,[A,B]]$ and so on.

Expanding each segment in powers of the vectors $s^a$, which number is denoted by the superindex, leads to (up to fourth order)
\begin{equation}
L^{(1)}_{1}=s^{c}_{I}{\mb A}_{c}, \qquad
L^{(2)}_{1}=\frac{1}{2!}s^{a}_{I}s^{b}_{I}\partial_{b}{\mb A}_{a}, \qquad\label{L1}
\ee
\ba
L^{(3)}_{1}&=&\frac{1}{3!}s^{c}_{I}s^{b}_{I}s^{a}_{I}\partial_{b}\partial_{c}{\mb A}_{a},\qquad
L_1^{(4)}= \frac{1}{4!}s_{I}^{d} s_{I}^{c}s_{I}^{b} s_{I}^{a}\partial_d\partial_c\partial_bA_a
\ea
\begin{equation}
L^{(1)}_{2}=(s^{a}_{J}-s^{a}_{I}){\mb A}_{a}, \qquad
L^{(2)}_{2}=\frac{1}{2!}(s^{b}_{I}+s^{b}_{J})(s^{a}_{J}-s^{a}_{I})\partial_{b}{\mb A}_{a}, \quad
\end{equation}
\ba
L^{(3)}_{2}&=&\frac{1}{3!}[(s^{b}_{J}s^{c}_{J}+s^{b}_{J}s^{c}_{I}+s^{b}_{I}s^{c}_{I})(s^{a}_{J}-s^{a}_{I})]\partial_{b}\partial_{c}{\mb A}_{a}\\
L_2^{(4)}&=& \frac{1}{4!}(s_{I}^{d}s_{I}^{b}s_{I}^{c}+s_{J}^{d}s_{J}^{b}s_{J}^{c}+s_{I}^{d}s_{I}^{b}s_{J}^{c}+s_{I}^{d}s_{J}^{b}s_{J}^{c}) (s_{J}^{a}-s_{I}^{a})       \partial_d\partial_c\partial_bA_a
\ea
\begin{equation}
L^{(1)}_{3}=-s^{c}_{J}{\mb A}_{c}, \qquad
L^{(2)}_{3}=-\frac{1}{2!}s^{a}_{J}s^{b}_{J}\partial_{b}{\mb A}_{a},\qquad
\end{equation}
\ba
L^{(3)}_{3}&=&-\frac{1}{3!}s^{c}_{J}s^{b}_{J}s^{a}_{J}\partial_{b}\partial_{c}{\mb A}_{a}, \qquad
L_3^{(4)}= -\frac{1}{4!}s_{J}^{d} s_{J}^{c}s_{J}^{b} s_{J}^{a}\partial_d\partial_c\partial_bA_a
\label{L3}
\ea
Next we write the contributions to $H$ in powers of the segments. The details are shown in the Appendix B. The first order contribution $H^{(1)}$ vanishes and the remaining contributions are
\begin{equation}
H^{(2)}=\frac{1}{2}s^{a}_{I}s^{b}_{J}{\mb F}_{ab},
\label{H2FGK}
\end{equation}
\begin{equation}
\label{H3FGK}
H^{(3)}=\frac{1}{3!}(s^{c}_{I}+s^{c}_{J})s^{a}_{I}s^{b}_{J}{\mb D}_{c}{\mb F}_{ab}
+{\mb \Sigma}^{(3)},
\end{equation}
with
\begin{equation}
{\mb \Sigma}^{(3)}=\frac{1}{12}(-s^{b}_{I}s^{a}_{I}s^{c}_{J}+s^{b}_{I}s^{a}_{J}s^{c}_{J}+s^{b}_{J}s^{a}_{I}s^{c}_{J}+s^{b}_{J}s^{a}_{J}s^{c}_{I}-s^{b}_{J}s^{a}_{I}s^{c}_{I}-s^{b}_{I}s^{a}_{J}s^{c}_{I})[\partial_{b}{\mb A}_{a},{\mb A}_{c}]
\label{NCFGK}
\end{equation}
and finally
\begin{equation}
H^{(4)}=\frac{1}{4!}\left(s^c_I s^d_J+s^c_J s^d_J+s^c_I s^d_I\right)\,s^a_I s^b_J\,
{\mb D}_c{\mb D}_d {\mb F}_{ab}+{\mb \Sigma_2}^{(4)}+{\mb \Sigma_3}^{(4)},
\label{H4FGK}
\end{equation}
with
\begin{eqnarray}
{\mb \Sigma}_{2}^{(4)}&=&\frac{1}{4!}\Big\{[{\mb A}_{d},\partial_{b}\partial_{c}{\mb A}_{a}](s_{I}^{d} s_{J}^{c} s_{J}^{b} s_{I}^{a}+s_{I}^{d}s_{I}^{c}s_{I}^{b}s_{J}^{a}-s_{I}^{d} s_{J}^{c} s_{J}^{b} s_{J}^{a}+s_{J}^{d} s_{I}^{c} s_{I}^{b} s_{I}^{a}-s_{J}^{d} s_{J}^{c} s_{J}^{b} s_{I}^{a}\nonumber\\
&&-s_{J}^{d} s_{I}^{c} s_{I}^{b} s_{J}^{a})+[\partial_c {\mb A}_{d},\partial_{b}{\mb A}_{a}](2s_{I}^{d} s_{I}^{c} s_{J}^{b} s_{I}^{a}-2s_{I}^{d}s_{I}^{c}s_{J}^{b}s_{J}^{a}-2s_{J}^{d} s_{I}^{c} s_{J}^{b} s_{I}^{a}\nonumber\\
&&+2s_{J}^{d} s_{I}^{c} s_{J}^{b} s_{J}^{a}) \Big\},
\label{NCFGK42}
\end{eqnarray}

\begin{eqnarray}
{\mb \Sigma}_{3}^{(4)}&=& \frac{\Big[ {\mb A}_d,[{\mb A}_c,\partial_{b}{\mb A}_a] \Big]}{4!}\Big\{-s_{I}^{d} s_{I}^{c} s_{J}^{b} s_{I}^{a}-s_{I}^{d}s_{I}^{c}s_{I}^{b}s_{J}^{a}+s_{I}^{d} s_{I}^{c} s_{J}^{b} s_{J}^{a}-s_{I}^{d} s_{J}^{c} s_{I}^{b} s_{I}^{a}  +s_{I}^{d} s_{J}^{c} s_{J}^{b} s_{I}^{a}       \nonumber\\
&&+s_{I}^{d} s_{J}^{c} s_{I}^{b} s_{J}^{a}-s_{I}^{d} s_{J}^{c} s_{J}^{b} s_{J}^{a}+s_{J}^{d}s_{I}^{c}s_{I}^{b}s_{I}^{a}-s_{J}^{d} s_{I}^{c} s_{J}^{b} s_{I}^{a}-s_{J}^{d} s_{I}^{c} s_{I}^{b} s_{J}^{a}+s_{J}^{d} s_{I}^{c} s_{J}^{b} s_{J}\nonumber\\
&&-s_{J}^{d} s_{J}^{c} s_{I}^{b} s_{I}+s_{J}^{d} s_{J}^{c} s_{J}^{b} s_{I}+s_{J}^{d} s_{J}^{c} s_{I}^{b} s_{J} \Big\},
\label{NCFGK43}
\end{eqnarray}

The reader can verify that the terms ${\mb \Sigma}_{2}^{(3)}$, ${\mb \Sigma}_{2}^{(4)}$    ${\mb \Sigma}_{3}^{(4)}$  are not covariant under the gauge group, as it should be expected from the definition of the path-ordering. This property has to be preserved at each order in the expansion in powers of the segments.

\section{Summary and discussion}

We have calculated the holonomy ${\mb h}_{\alpha_{IJ}}$ of the Yang-Mills connection ${\mb A}_a$ in the triangle $\alpha_{IJ}$ with vertex $v$ and sides $s^a_I$, $s^b_J$ joining at that vertex, as shown in Fig. (1). Our results, to fifth order in the segments, are given in Eqs.
(\ref{2ORDHOL}) (\ref{3ORDHOL}) (\ref{4ORDHOL}) and (\ref{5ORDHOL}) of section 4. The direct calculation shows that, to fourth order, the results are directly given by the replacement $\partial_a \rightarrow {\mb D}_a, \,\,  F_{ab}
\rightarrow {\mb F}_{ab}$ in the corresponding formulas for the Abelian case. This is so because, as explained at the end of section 4, the potential order ambiguity in the fourth order term is absent. From the fifth order on such ordering ambiguities arise, so that  is not possible to guess the correct answer from the Abelian case. Clearly then, one has to perform the full calculation in order  to obtain the correct result. 

In section 5 the same  calculation was  performed using the method of Ref.\cite{FGK}.
The results for the triangle coincide to second order and start to differ from the third order on. This difference can be traced back to that arising in the calculation of the holonomy for a straight-line segment according to Eqs. (\ref{COMP}) and (\ref{U1FGK}), which the reader can easily verify. Contrary to what is expected, the calculation according to the method of Ref.\cite{FGK} produces non-covariant contributions to third and fourth order, given in Eqs. (\ref{NCFGK}) and (\ref{NCFGK42}), (\ref{NCFGK43}), respectively.

The specific results presented in Ref.\cite{FGK} correspond to the  calculation of the holonomy for a rectangle of sides  $a$ and $b$, respectively. Using the method described in section 4 we have verified that, up to fourth order, the result given in Eq. (3.19) of that reference is correct.

Nevertheless, since the calculational method of Ref. \cite{FGK} does not properly take into account the path ordering, it is not possible to guarantee the multiplicative composition law of holonomies. In fact, one might think of  obtaining the holonomy for the rectangle by composing the results of properly chosen triangles. Even though, as commented above, one should not expect to obtain the correct result we have explored this possibility.
To this end let us consider a quadrilateral $ABCD$  with vertex $v$ at the point $A$, as shown in Fig. (2). The sides $AB$ and $CD$ are parallel, but $DA$ and $BC$ are not. We are interested in calculating the holonomy for the path $ABCDA$ as indicated in Fig. (2). This can be done by composing, via matrix multiplication, the holonomies corresponding to the triangles $ABC$ (spanned by the vectors ${\vec s}_I$, ${\vec s}_J$) and $ACD$ (spanned by the vectors ${\vec{ \bar{s}}}_I{}$, ${\vec{ \bar{s}}}_J{}$), each of which is calculated according to
expressions (\ref{H2FGK}),(\ref{H3FGK}) and (\ref{H4FGK}), together with their corresponding ${\mb \Sigma}$'s. In other words,


\begin{figure}
\begin{center}
\includegraphics[height=5cm]{vectores}% Here is how to import EPS art
\caption{Triangle $\alpha_{IJ}$ with vertex $v$ }
\end{center}
\end{figure}






\begin{figure}
\begin{center}
\includegraphics[height=5cm]{cuadrilatero}% Here is how to import EPS art
\caption{Quadrilateral $ABCD$  with vertex $v$ at the point $A$ }
\end{center}
\end{figure}


%\begin{figure*}{h}
%\includegraphics[width=5cm]{cuadrilatero.eps}
%\end{figure}


\be
{\mb h}_{ABCDA}={\mb h}_{ACDA}{\mb h}_{ABCA}.
\ee
We introduce the further notation in the plane of the quadrilateral
\ba
&&{\vec {AB}}={\vec s}_I= (a,0), \qquad  {\vec {AC}}={\vec s}_J= (a',b),\\
&&{\vec {AC}}={\vec{ \bar{s}}}_I{\,}= (a',b), \qquad  {\vec {AD}}={\vec{\bar{ s}}}_J{}= (a'',b),
\ea
We only pay attention to the non-covariant contributions. To third order they just add up and we obtain
\begin{eqnarray}
\label{NCFGK3}
{\mb \Sigma}_{ABCA}^{(3)}+{\mb \Sigma}_{ACDA}^{(3)}&=&\frac {(a''-a'+a)}{12}
\Big\{3a'(a''-a) [\partial_1{\mb A}_1,{\mb A}_1] \nonumber\\
&+& b (a''+a'-a) \Big([\partial_1{\mb A}_1,{\mb A}_2]+[\partial_1{\mb A}_2,{\mb A}_1]+[\partial_2{\mb A}_1,{\mb A}_1] \Big)\nonumber \\
&&  +b^2\Big[[\partial_2{\mb A}_1,{\mb A}_2]+[\partial_2{\mb A}_2,{\mb A}_1]+[\partial_1{\mb A}_2,{\mb A}_2]\Big]\Big\}.
\end{eqnarray}
We see that the above is not zero in general. Nevertheless, in the  symmetrical case of  a
parallelepiped, characterized by the condition
\be
a''-a'+a=0,
\ee
the non-covariant piece (\ref{NCFGK3}) vanishes. Of course, the above condition includes that of a rectangle which is  $a=a'$ y   $ a''=0 $.

Nevertheless, the failure of the composition property of the calculational method in Ref. \cite{FGK} is illustrated at the fourth order where one should have a cancelation among each of the corresponding pieces $({\mb \Sigma}_{ABCA})_p^{(4)}$ and $ ({\mb \Sigma}_{ACDA})_p^{(4)}, \,\, p=2,3.$ The simplest symmetrical case corresponds to  ${\mb \Sigma}_2^{(4)}$ and leads to

\begin{eqnarray}
&&({\mb \Sigma}_{ABCA})_{2}^{(4)}+ ({\mb \Sigma}_{ACDA})_{2}^{(4)}=\frac{1}{4!}([A_J,(2x_I^2+x_J^2-2x_Ix_J)(A_I-A_J)+x_J^2A_I]\nonumber\\
&&+2[A_I,(x_I^2-x_Ix_J)A_J]+2[x_I(A_I-A_J),x_J(A_I-A_J)]-2[x_JA_I,x_IA_I] )\nonumber\\
&&\neq 0.
\end{eqnarray}

Where:
\begin{eqnarray}
A_I=s_I^aA_a, \qquad A_J=s_J^aA_a 
\end{eqnarray}





\section*{Acknowledgements}

LFU acknowledges partial support from Fondecyt 7010967 as well as
from the project CONACYT 40745-F. 
MR acknowledges support from project (Apoyo de Tesis Doctoral) CONICYT (Chile), and wants 
to thanks professors Luis Fernando Urrutia and Hugo Morales-T\'ecotl for their hospitality at ICN-UNAM and UAM.
HMT acknowledges partial support from  CONICYT-CONACYT 2001502170 as well as
from the project CONACYT 40745-F.  
The work of JA has been partially supported by Fondecyt 1010967.


\section*{Appendix A}

Here we include some intermediate steps in the calculation  of the expansion of the  non-Abelian holonomy ${\mb {h}}_{\alpha_{IJ}}$ in powers of the segments as described in section 4.2. The order refers to the corresponding power.

\subsection{Quadratic order}
We have
\be
{\mb{h}}_{\alpha_{IJ}}^{(2)}= U_1^{(2)} + U_2^{(2)} + U_3^{(2)}+ U_2^{(1)}U_1^{(1)}+
U_3^{(1)}U_1^{(1)}+U_3^{(1)}U_2^{(1)}.
\ee
The separate contributions  are
\ba
&&U_1^{(2)} + U_2^{(2)} + U_3^{(2)}=\frac {1}{2}[(-ys_{I}^{a}+xs_{J}^{a}){\mb A}_{a}+(2s_{J}^{a}s_{J}^{b}+2s_{I}^{a}s_{I}^{b}-s_{J}^{a}s_{I}^{b}-s_{I}^{a}s_{J}^{b})\nonumber\\
&&{\mb A}_{a}{\mb A}_{b}],\\
&&U_2^{(1)}U_1^{(1)}+
U_3^{(1)}U_1^{(1)}+U_3^{(1)}U_2^{(1)}
=(-s_{J}^{a}s_{J}^{b}-s_{I}^{a}s_{I}^{b}+s_{J}^{a}s_{I}^{b}){\mb A}_{a}{\mb A}_{b}
\ea

\subsection{ Third order}

Here
\ba
{\mb{h}}_{\alpha_{IJ}}^{(3)}&=&U_1^{(3)} + U_2^{(3)} + U_3^{(3)}
+  U_3^{(2)}U_1^{(1)}+ U_3^{(2)}U_2^{(1)}\nonumber \\
&&+U_2^{(2)}U_1^{(1)}
+  U_3^{(1)}U_1^{(2)}+ U_3^{(1)}U_2^{(2)}+U_2^{(1)}U_1^{(2)}
\ea
The separate contributions are
\begin{eqnarray}
&&U_1^{(3)} + U_2^{(3)} + U_3^{(3)}=\frac{1}{3!}\left[x^2s_I^a-y^2s_I^a+xy(s_J^a-s_I^a)\right]{\mb A}_{a}+\left[\bar{x}\left(2(s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})+s_{I}^{a}s_{I}^{b}\right){}\right.
\nonumber\\&&
{}\left.+\bar{y}\left(2s_{J}^{a}s_{J}^{b}+(s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})\right)+x\left((s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})+2s_{I}^{a}s_{I}^{b}\right){}\right.
\nonumber\\&&
{}\left.+y\left(s_{J}^{a}s_{J}^{b}+2(s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})\right)\right]{\mb A}_{a}\bar{ \mb {A}}_{b}
+[-s_{J}^{a}s_{J}^{b}s_{J}^{c}+(s_{J}^a-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})(s_{J}^{c}-s_{I}^c)
\nonumber\\&&
+s_{I}^{a}s_{I}^{b}s_{I}^{c}]{\mb A}_a {\mb A}_b {\mb A}_c
\end{eqnarray}
together with
\begin{eqnarray}
&&U_3^{(2)}U_1^{(1)}+ U_3^{(2)}U_2^{(1)}+U_2^{(2)}U_1^{(1)}
+  U_3^{(1)}U_1^{(2)}+ U_3^{(1)}U_2^{(2)}+U_2^{(1)}U_1^{(2)}
\nonumber \\
&&= \frac{1}{2}\left[\bar{x}\left(-s^{a}_Is^{b}_I-s_{J}^{a}(s_{J}^{b}-s_{I}^{b})\right)-\bar{y}s_{J}^{a}(s_{J}^{b}-s_{I}^{b}){}\right.
\nonumber\\
&&{}\left. +x(s_{J}^{a}-s_{I}^{a})s_{J}^{b}+y\left(-s^{a}_Is^{b}_I-s_{J}^{a}(s_{J}^{b}-s_{I}^{b}) \right)\right]{\mb A}_{a}\bar{{\mb A}}_{b}
\end{eqnarray}

\subsection{ Fourth order}


In the same way we have
\ba
{\mb{h}}_{\alpha_{IJ}}^{(4)}=U_1^{(4)} + U_2^{(4)} + U_3^{(4)}
+(U_3^{(3)}+U_2^{(3)}+ U_3^{(1)}U_2^{(2)}+U_3^{(2)}U_2^{(1)})U_1^{(1)}+ (U_3^{(2)}+U_2^{(2)}
\nonumber\\
+ U_3^{(1)}U_2^{(1)})U_1^{(2)}+(U_3^{(1)}+U_2^{(1)})U_1^{(3)}+U_3^{(3)}U_2^{(1)}+U_3^{(2)}U_2^{(2)}+U_3^{(1)}U_2^{(3)}
\ea
The separate contributions are
\ba
&&(U_3^{(3)}+U_2^{(3)}+U_3^{(1)}U_2^{(2)}+U_3^{(2)}U_2^{(1)})U_1^{(1)}=\frac{1}{3!}\Big\{ \left[(x^2+xy)(s_J^a-s_I^a)-y^2s_I^a\right]s_I^b {\mb A}_{a}\bar{\mb A}_{b}  \nonumber \\
&&+\left[(x+2y+\bar{y}+2\bar{x})(s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})+(2\bar{y}+y) s_{J}^{a}s_{J}^{b}-3(\bar{x}+\bar{y})s_{J}^{a}(s_{J}^{b}-s_{I}^{b})\right. \nonumber \\
&&\left.-3ys_{J}^{a}(s_{J}^{b}-s_{I}^{b})\right]s_I^c{\mb A}_a\bar{\mb A}_b\bar{\bar{\mb A}}_c+\left[(s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})(s_{J}^{c}-s_{I}^{c})-s_{J}^{a}s_{J}^bs_{J}^{c}-3s_{J}^{a}(s_{J}^{b}-s_{I}^{b}){}\right.  
\nonumber\\&&
{}\left. (s_{J}^{c}-s_{I}^{c})   +3s_{J}^{a}s_{J}^{b}(s_{J}^{c}-s_{I}^{c})\right]s_I^d{\mb A}_a{\mb A}_b{\mb A}_c{\mb A}_d\Big\}
\ea

\ba
(U_3^{(2)}+U_2^{(2)}&+&U_3^{(1)}U_2^{(1)})U_1^{(2)}=\frac{1}{2!2!}\Big\{[x\bar{x}(s_J^a-s_I^a)-y\bar{x}s_I^a]s_I^b{\mb A}_a{\bar{\mb A}}_b+\left[x(s_{J}^{a}-s_{I}^{a})s_{I}^{b}\right. \nonumber \\
&&\left.-y s_{I}^{a}s_{I}^{b}   +\bar{\bar{x}}(s_{J}^{a}-s_{I}^{a})(s_{J}^{b}-s_{I}^{b})\right]s_I^c{\mb A}_a\bar{{\mb A}}_b\bar{{\bar{\mb A}}}_c\nonumber\\
&&+\left[(s_{J}^{a}-s_{I}^{a})(s_{J}^b-s_{I}^b)s_I^c\right]s_I^d{\mb A}_a{\mb A}_b{\mb A}_c {\mb A}_d\Big\}  
\ea

\ba
\left(U_3^{(1)}+U_2^{(1)}\right)U_1^{(3)}&=&\frac{1}{3!}\left[-\bar{x}^2s_I^as_I^b{\mb A}_a\bar{\mb A}_b-(\bar{\bar{x}}+2\bar{x})s_{I}^{a}s_{I}^{b}s_{I}^{c}\right. \nonumber\\
&&\left.{\mb A}_a\bar{\mb A}_b\bar{\bar{\mb A_c}}-s_{I}^{a}s_{I}^{b}s_{I}^cs_{I}^d{\mb A}_a{\mb A}_b{\mb A}_c{\mb A}_d\right] 
\ea

\ba
U_3^{(3)}U_2^{(1)}&=&\frac{1}{3!}\left[-y^2 s_J^a (s_J^{b}-s_I^{b}){\mb A}_a\bar{\mb A}_b+(2\bar{y}+y) s_J^{a} s_J^{b} (s_J^{c}-s_I^c){\mb A}_a\bar{\mb A}_b\bar{\bar{\mb A}_c} {}\right.
\nonumber\\&&
{}\left.-s_J^{a}s_J^{b}s_{J}^c(s_J^d-s_I^d){\mb A}_a{\mb A}_b{\mb A}_c{\mb A}_d\right]
\ea

\ba
&&U_3^{(2)}U_2^{(2)}=\frac{1}{2!2!}\left[-y(\bar{x}+\bar{y})s_J^a (s_J^{b}-s_I^{b}){\mb A}_a\bar{\mb A}_b-y  s_J^{a} (s_J^{b}-s_I^b)(s_J^{c}-s_I^c) {\mb A}_a\bar{\mb A}_b\bar{\bar{\mb A}_c} \right.
\nonumber\\&&
{}\left. +(\bar{\bar{x}}+\bar{\bar{y}})  s_J^{a} s_J^{b}(s_J^{c}-s_I^c){\mb A}_a\bar{\mb A}_b\bar{\bar{\mb A_c}}{}+s_J^as_J^b(s_J^c-s_I^c)(s_J^d-s_I^d){\mb A}_a{\mb A}_b{\mb A}_c{\mb A}_d\right] 
\ea

\ba
U_3^{(1)}U_2^{(3)}&=&\frac{1}{3!}\left[-(\bar{x^2}+\bar{y^2}+\bar{x}\bar{y}) s_J^a (s_J^b-s_I^b){\mb A}_a\bar{\mb A}_b -s_J^a(s_J^b-s_I^b)(s_J^c-s_I^c)(s_J^d-s_I^d){}\right.
\nonumber\\&&\left.{\mb A}_a{\mb A}_b{\mb A}_c{\mb A}_d-(\bar{x}+2\bar{y}+\bar{\bar{y}}+2\bar{\bar{y}})  s_J^{a} (s_J^{b}-s_I^b)(s_J^{c}-s_I^c){\mb A}_a\bar{\mb A}_b\bar{\bar{\mb A_c}}
{}\right]
\ea

\section*{Appendix B}

Here we include some intermediate steps in the calculation  of the expansion of  the non-Abelian holonomy ${\mb {h}}_{\alpha_{IJ}}$ in powers of the segments as described in section 5. The order refers to the corresponding power




\subsection{Third order }

We  consider the third order contribution to the expressions (\ref{SIGMA}),  (\ref{F2}), (\ref{F3}) and (\ref{F4}) together with the basic expressions given in Eqs. (\ref{L1}-\ref{L3}). The contributing pieces are

(i)
\begin{eqnarray}
\label{THIRDO}
\Sigma^{(3)}&=&L^{(3)}_{1}+L^{(3)}_{2}+L^{(3)}_{3}
=\frac{1}{3!}(s^c_I+ s^c_J)s^a_Is^b_
J\,\partial_c\,(\partial_a {\mb A}_b-\partial_b {\mb A}_a),
\ea


(ii)
\ba
2!F^{(3)}_{2}&=&[L_{3},L_{1}]^{(3)}+[L_{3},L_{2}]^{(3)}+[L_{2},L_{1}]^{(3)}, 
\label{THIRDO1}
\ea
with
\begin{equation}
[L^{(2)}_{3},L^{(1)}_{1}]=-\frac{1}{2}s^{b}_{J}s^{a}_{J}s^{c}_{I}[\partial_{b}{\mb A}_{a},{\mb A}_{c}],
\qquad
[L^{(1)}_{3},L^{(2)}_{1}]=-\frac{1}{2}s^{b}_{I}s^{a}_{I}s^{c}_{J}[{\mb A}_{c},\partial_{b}{\mb A}_{a}],
\end{equation}
\begin{equation}
[L^{(2)}_{2},L^{(1)}_{1}]=\frac{1}{2}(s^{b}_{I}+s^{b}_{J}) (s^{a}_{J}-s^{a}_{I})s^{c}_{I}[\partial_{b}{\mb A}_{a},{\mb A}_{c}], \quad
[L^{(1)}_{2},L^{(2)}_{1}]=\frac{1}{2}(s^{c}_{J}-s^{c}_{I}) s^{b}_{I}s^{a}_{I}[{\mb A}_{c},\partial_{b}{\mb A}_{a}],
\end{equation}
\begin{equation}
[L^{(2)}_{3},L^{(1)}_{2}]=-\frac{1}{2}s^{b}_{J}s^{a}_{J}  (s^{c}_{J}-s^{c}_{I}) [\partial_{b}A_{a},A_{c}], \quad
[L^{(1)}_{3},L^{(2)}_{2}]=\frac{1}{2}(s^{b}_{I}+s^{b}_{J})(s^{a}_{J}-s^{a}_{I})s^{c}_{J}[\partial_{b}{\mb A}_{a},{\mb A}_{c}].
\end{equation}
leading to
\begin{equation}
2!F^{(3)}_{2}=\frac{1}{2}(s^{b}_{I}s^{a}_{J}s^{c}_{I}-s^{b}_{J}s^{a}_{I}s^{c}_{I}+s^{b}_{J}s^{a}_{J}s^{c}_{I}+s^{b}_{I}s^{a}_{J}s^{c}_{J}-s^{b}_{I}s^{a}_{I}s^{c}_{J}-s^{b}_{J}s^{a}_{I}s^{c}_{J})[\partial_{b}{\mb A}_{a},{\mb A}_{c}].
\end{equation}

(iii)
\ba
3!F_{3}^{(3)}
&=&[L_{3}^2L_1]^{(3)}+[L_{2}^2L_1]^{(3)}+[L_{3}^{2}L_{2}]^{(3)}+2[L_3L_2L_1]^{(3)},
\end{eqnarray}
 with
\begin{eqnarray}
\left[L_{3},[L_{3},L_{1}]\right]^{(3)}&=&-s^{b}_{J}s^{a}_{I}s^{c}_{J}\left[{\mb A}_{c},[{\mb A}_{a},{\mb A}_{b}]\right],
\\
\left[L_{2},[L_{2},L_{1}]\right]^{(3)}&=&-(s^{b}_{J}-s^{b}_{I})s^{a}_{I}(s^{c}_{J}-s^{c}_{I})\left[{\mb A}_{c},[{\mb A}_{a},{\mb A}_{b}]\right],
\\
\left[L_{3},[L_{3},L_{2}]\right]^{(3)}&=&-s^{b}_{J}(s^{a}_{J}-s^{a}_{I})s^{c}_{J}\left[{\mb A}_{c},[{\mb A}_{a},{\mb A}_{b}]\right],
\\
2\left[L_{3},[L_{2},L_{1}]\right]^{(3)}&=&2(s^{b}_{J}-s^{b}_{I})s^{a}_{I}s^{c}_{J}\left[{\mb A}_{c},[{\mb A}_{a},{\mb A}_{b}]\right],
\end{eqnarray}
leading to
\begin{equation}
F^{(3)}_{3}=\frac{1}{3!}(s^{c}_{I}+s^{c}_{J})s^{a}_{I}s^{b}_{J}\partial_{c}\left[{\mb A}_{c},[{\mb A}_{a},{\mb A}_{b}]\right]
\end{equation}
Adding up the above results we
finally obtain Eqs. (\ref{H3FGK}) and (\ref{NCFGK}).

\subsection{Fourth order }

The contributions are

(i)
\begin{equation}
\Sigma^{(4)}=L_1^{(4)}+L_2^{(4)}+L_3^{(4)}=  \frac{1}{4!}(s_{I}^{c} s_{I}^{d}+ s_{I}^{c} s_{J}^{d}+s_{J}^{c}s_{J}^{d})s_{I}^{a}s_{J}^{b}     \partial_c\partial_d\,(\partial_a {\mb A}_b-\partial_b {\mb A}_a)
\end{equation}

(ii)
\begin{eqnarray}
2!F_{2}^{(4)}&=&[L_2,L_1]^{(4)}+[L_3,L_2]^{(4)}+[L_3,L_1]^{(4)} \nonumber \\
&=& [L_2,L_1-L_3]^{(4)}+[L_3,L_1]^{(4)}
\label{F2(4)}
\end{eqnarray}
The first term in the RHS of Eq. (\ref{F2(4)}) contributes with
\begin{equation}
[L_2^{(1)},L_1^{(3)}-L_3^{(3)}]= \frac{1}{3!}( s_{J}^{d}- s_{I}^{d})  ( s_{I}^{c}s_{I}^{b}s_{I}^{a}+s_{J}^{c} s_{J}^{b} s_{J}^{a})[ {\mb A}_d,\partial_b\partial_c{\mb A}_a]
\end{equation}
\begin{equation}
[L_2^{(2)},L_1^{(2)}-L_3^{(2)}]= \frac{1}{2!2!}( s_{J}^{c}+ s_{I}^{c}) ( s_{J}^{d}- s_{I}^{d})  ( s_{I}^{a}s_{I}^{b}+s_{J}^{a} s_{J}^{b} )[ \partial_c{\mb A}_d,\partial_b{\mb A}_a]
\end{equation}
\begin{equation}
[L_2^{(3)},L_1^{(1)}-L_3^{(1)}]= \frac{1}{3!}( s_{J}^{b}s_{J}^{c}+s_{J}^{b}s_{I}^{c}+ s_{I}^{b} s_{I}^{a})   ( s_{J}^{a}- s_{I}^{a})( s_{J}^{d}+ s_{I}^{d})  [ \partial_b\partial_c{\mb A}_a,{\mb A}_d]
\end{equation}
The second term in the RHS of Eq. (\ref{F2(4)}) contributes with
\begin{equation}
[L_3^{(1)},L_1^{(3)}]=- \frac{1}{3!}s_{J}^{d}s_{I}^{c}  s_{I}^{b}s_{I}^{a}[ {\mb A}_d,\partial_b\partial_c{\mb A}_a]
\end{equation}
\begin{equation}
[L_3^{(2)},L_1^{(2)}]=- \frac{1}{2!2!} s_{J}^{a}s_{J}^{b} s_{J}^{c}s_{J}^{d}[\partial_b{\mb A}_a,\partial_c{\mb A}_d]
\end{equation}
\begin{equation}
[L_3^{(1)},L_1^{(3)}]=- \frac{1}{3!}    s_{I}^{d}s_{J}^{c}  s_{J}^{b}s_{J}^{a}       [\partial_b\partial_c{\mb A}_a, {\mb A}_d]
\end{equation}

(iii)
\begin{equation}
3!F_{3}^{(4)}=[L_3^2L_1]^{(4)}+[L_2^2L_1]^{(4)}+[L_3^2L_2]^{(4)}
+2[L_3L_2L_1]^{(4)}-[\Sigma F_2]^{(4)}
\end{equation}
The individual contributions are
\begin{equation}
[L_3^{(1)},[L_3^{(1)},L_1^{(2)}]+[L_3^{(2)},L_1^{(1)}]]=- \frac{1}{2!}s_{J}^{d} ( s_{I}^{c}  s_{J}^{b}s_{J}^{a}- s_{J}^{c}  s_{I}^{b}s_{I}^{a})[{\mb A}_d,[{\mb A}_c,\partial_b {\mb A}_a]]
\end{equation}
\begin{eqnarray}
[L_3^{(2)},[L_3^{(1)},L_1^{(1)}]+2[L_2^{(1)},L_1^{(1)}]]&=&-\frac{1}{2!}(s_I^ds_{J}^{c}-2 s_{I}^{d}s_{I}^{c})s_{J}^{b}s_J^a \nonumber\\
&& \times [\partial_b {\mb A}_a,[{\mb A}_c,{\mb A}_d]]
\end{eqnarray}

\begin{eqnarray}
[L_2^{(1)},[L_2^{(1)},L_1^{(2)}]+[L_2^{(2)},L_1^{(1)}]]&=&\frac{1}{2!}(s_{J}^{d}-s_I^d)\left((s_{J}^{c}-s_I^c)s_{I}^{b}s_{I}^{a}-s_{I}^{c}( s_{I}^{b}+s_{J}^{b})(s_{J}^{a}-s_I^a)\right)     \nonumber \\
&&  \times [{\mb A}_d,[{\mb A}_c,\partial_b {\mb A}_a]]
\end{eqnarray}
\begin{eqnarray}
[L_2^{(2)},[L_2^{(1)},L_1^{(1)}]]=\frac{1}{2!}s_I^d (s_{J}^{c}-s_{I}^{c})( s_{I}^{b}+s_{J}^{b}) ( s_J^a-s_I^a) [\partial_b {\mb A}_a,[{\mb A}_c,{\mb A}_d]]
\end{eqnarray}

\begin{eqnarray}
[L_3^{(1)},[L_3^{(1)},L_2^{(2)}]+[L_3^{(2)},L_2^{(1)}]]&=&-s_J^d\frac{1}{2!} \left( (s_{J}^{c}( s_{I}^{b}+s_{J}^{b})(s_{J}^{a}-s_I^a)+ (s_{J}^{c}-s_I^c)s_J^bs_J^a   \right)     \nonumber \\
&&  \times [{\mb A}_d,[{\mb A}_c,\partial_b {\mb A}_a]]
\end{eqnarray}
\begin{eqnarray}
[L_3^{(2)},[L_3^{(1)},L_2^{(1)}]]=\frac{1}{2!}(s_J^d-s_I^d) s_{J}^{c} s_{J}^{b}s_J^a [\partial_b {\mb A}_a,[{\mb A}_c,{\mb A}_d]]
\end{eqnarray}





\begin{eqnarray}
2[L_3^{(1)},[L_2^{(1)},L_1^{(2)}]+[L_2^{(3)},L_1^{(1)}]]&=&-s_J^d \left((s_{I}^{c}( s_{I}^{b}+s_{J}^{b})(s_{J}^{a}-s_I^a)+ (s_{J}^{c}-s_I^c)s_I^bs_I^a   \right)     \nonumber \\
&& \times  [{\mb A}_d,[{\mb A}_c,\partial_b {\mb A}_a]]
\end{eqnarray}


\begin{eqnarray}
-[\Sigma^{(2)},F_2^{(2)}]=\frac{1}{2!2!}s_J^ds_{I}^{c}( s_{I}^{b}s_J^a-s_J^bs_I^a) [\partial_b {\mb A}_a,[{\mb A}_c,{\mb A}_d]]
\end{eqnarray}


(iv)
\begin{eqnarray}
4!F_{4}^{(4)}&=&[L_3^2L_2]^{(4)} +[L_3^3L_1]^{(4)}+[L_2^3L_1]^{(4)}+3[L_3^2L_2L_1]^{(4)}+3[L_3L_2^2L_1]^{(4)}\nonumber \\
&&-3![\Sigma,F_3]^{(4)}-[\Sigma^2,F_2]^{(4)}  
\end{eqnarray}
The contributions are

\begin{equation}
[L_3^3L_2]^{(4)}=-s_{J}^{c} s_{J}^{d}  s_{J}^{a} (s_{J}^{b}-s_I^b) [{\mb A}_c,[{\mb A}_d,[{\mb A}_a,{\mb A}_b]]]
\end{equation}


\begin{equation}
[L_3^3L_1]^{(4)}=-s_{J}^{c} s_{J}^{d}  s_{J}^{a} s_I^b [{\mb A}_c,[{\mb A}_d,[{\mb A}_a,{\mb A}_b]]]
\end{equation}


\begin{equation}
[L_2^3L_1]^{(4)}=(s_{J}^{c}-s_{I}^{c})(s_{J}^{d}-s_{I}^{d}) (s_J^a-s_I^a)s_I^b [{\mb A}_c,[{\mb A}_d,[{\mb A}_a,{\mb A}_b]]]
\end{equation}


\begin{equation}
3[L_3^2L_2L_1]^{(4)}=3s_{J}^{c} s_{J}^{d} ( s_{J}^{a}-s_I^a) s_I^b [{\mb A}_c,[{\mb A}_d,[{\mb A}_a,{\mb A}_b]]]
\end{equation}


\begin{equation}
3[L_3L_2^2L_1]^{(4)}=-3s_{J}^{c} (s_{J}^{d}-s_I^d) ( s_{J}^{a}-s_I^a) s_I^b [{\mb A}_c,[{\mb A}_d,[{\mb A}_a,{\mb A}_b]]]
\end{equation}

The terms $[\Sigma,F_3]^{(4)}$ and $[\Sigma^2,F_2]^{(4)}$ do not contribute to fourth order
because they start at the fifth and sixth power in the segments, respectively.




\section*{References}
\begin{thebibliography}{99}

\bibitem{HUDR} V.W. Hughes, H.G. Robinson and V. Beltr\'an-L\'opez, Phys.
Rev. Letts. {\bf 4}, 342 (1960); R.W.P. Drever, Philos. Mag. {\bf
6}, 683 (1961); J.D. Prestage {\it et al.}, Phys. Rev. Letts. {\bf
54}, 2387 (1985); S.K. Lamoreaux {\it et al.}, Phys. Rev. Letts.
{\bf 57}, 3125 (1986); S.K. Lamoreaux {\it et al.}, Phys. Rev.
{\bf A 39}, 1082 (1989); C. J. Berglund, {\it et al.}, Phys. Rev.
Lett. {\bf 75} 1879 (1995); D. F. Phillips {\it et al.}, Phys. Rev
D {\bf 63} 111101 (2001).


\bibitem{Exper2} T.E. Chupp {\it et al.}, Phys. Rev. Letts. {\bf 63},
1541 (1989); D. Bear {\it et al.},
Phys. Rev. Letts. {\bf 85} 5038 (2000).


\bibitem{kostelecky1} V.A. Kostelecky and S. Samuel, Phys. Rev. {\bf D39},
683 (1989), Phys. Rev. {\bf D40}, 1886 (1989); V.A. Kostelecky and
R. Potting, Nucl. Phys. {\bf B 359},545 (1991); Phys. Lett. {\bf B
381}, 89
(1996); D. Colladay and V.A. Kostelecky {\ Phys. Rev.} {\bf %
D55}, 6760 (1997); {\ Phys. Rev.} {\bf %
D58}, 116002 (1998);  V.A. Kostelecky, [arXiv: hep-ph/0104227] and
references therein;  R. Bluhm,[ arXiv: hep-ph/0111323] and
references therein.

\bibitem{CPTPROC} Proceedings of the Meeting on CPT and Lorentz Symmetry, V. Alan Kosteleck\'y, ed.,
World Scientific, Singapore (1999).

\bibitem{Gambini}R. Gambini and J. Pullin, {\ Phys. Rev.} {\bf D59},
124021 (1999).
\bibitem{amufot} J. Alfaro, H. Morales-T\'ecotl and L. Urrutia, Phys. Rev.
{\bf D65}, 103509 (2002).

\bibitem{Us} J. Alfaro, H.A. Morales-Tecotl, L.F. Urrutia, {\ Phys.
Rev.
Lett.} {\bf 84}. 2318 (2000); {\ JHEP Proceedings, Cartagena de
Indias 2000, High Energy Physics}; L.F. Urrutia, Mod. Phys. Letts. {\bf A17},943 (2002);J. Alfaro, H. Morales-T\'ecotl and L. Urrutia, Phys. Rev.
{\bf D66}, 124006 (2002); J. Alfaro, H. Morales-T\'ecotl and L.
Urrutia, in {\ Proceedings of the Ninth Marcel Grossmann Meeting
on General Relativity}, eds. R.T Jantzen, V. Gurzadyan and R.
Rufini (World Scientific, 2002); L. F. Urrutia, {\it Loop quantum gravity induced modification to
particle dynamics}, hep-ph/0303189.

\bibitem{Ellis} J. Ellis, N.E. Mavromatos and D.V. Nanopoulos,
in Tegernsee 1999, Beyond the desert, {\ Gen.
Rel. Grav.} {\bf 32}, 127 (2000), {\it Phys. Rev. D} {\bf 61},
027503 (2000); J. Ellis {\it et al.}, {\ Astrophysical Jour.} {\bf 535%
}, 139 (2000); J. Ellis {\it et al.}, {\ Gen. Rel. Grav.} {\bf
32}, 1777 (2000).



\bibitem{QuantFluct} G. Amelino-Camelia {\it et al.} {\ Nature}
(London) {\bf 393}, 763 (1998); {Nature}
(London) {\bf 400}, 849 (1999); D.V. Ahluwalia, Nature (London)
{\bf 398}, 199 (1999); G. Amelino-Camelia, Lect.\ Notes Phys.\
{\bf 541}, 1 (2000).




\bibitem{BILLER} S.D. Biller {\it et. al.}, {\ Phys. Rev. Lett.} {\bf 83},
2108 (1999).

\bibitem{SUV}
D. Sudarsky, L. Urrutia, and H. Vucetich, Phys. Rev. Letts. {\bf 89}, 231301 (2002); {\it Bounds on stringy quantum gravity from low energy existing data}, gr-qc/0211101.

\bibitem{ap}J. Alfaro
and G. Palma, Phys. Rev. {\bf D65}:103516 (2002);{\it Loop Quantum Gravity and Ultra High Energy
Cosmic Rays}, hep-th/0208193.

\bibitem{observ} C. L\"{a}mmerzhal and C. Bord\'e, in
Lecture Notes in Physics 562, Springer 2001; R.J. Gleisser and C.
N. Kozamech, Phys. Rev. {\bf D64}:083007 (2001); R. Brustein, D.
Eichler and S. Foffa, Phys. Rev. {\bf D65}:105006 (2002); G.
Lambiase, Gen. Rel.Grav. {\bf 33}, 2151 (2001);  G.
Amelino-Camelia, Phys. Letts. {\bf B528}, 181 (2002);   T. Jacobson, S.
Liberati and D. Mattingly, [arXiv: hep-ph/0112207]; T. Konopka
and S. Major, New J. Phys. {\bf 4}:57 (2002).

\bibitem{cfl} N. H. Christ, R. Friedberg and T.D. Lee, Nucl. Phys. {\bf B202}, 89(1982);
{\bf B210(FS6)} 310, 337(1982).

\bibitem{jan} F. David, Nucl. Phys. {\bf B257(FS14)}, 543(1985); V.A. Kazakov, I.K. Kostov and
A.A. Migdal, Phys. Lett. {\bf 157B}, 295(1985); J. Ambjorn, B. Durhuus and J. Frohlich,
Nucl. Phys.{\bf B257(FS14)}, 433(1985).

\bibitem{FGK} P. Fishbane, S. Gasiorowicz and P. Kaus, Phys. Rev. {\bf D24}, 2324 (1981).

\bibitem{STOKES} See for example: N.E. Bralic, Phys. Rev. {\bf D22}, 3090(1980); M. Hirayama and S. Matsubara, Prog. Theor. Phys. {\bf 99}, 691(1998); D. Diakonov and V. Petrov, {\it On the non-Abelian Stokes theorem}, hep-lat/0008004; B. Broda, {\it Non-Abelian Stokes theorem in action}, math-ph/0012035.

\bibitem{form}J.A.M.Vermaseren "New features of FORM" math-ph/0010025.
%%%%%%%%%%%%%%%%%%%%%%%%

\end{thebibliography}






\end{document}
