% $Id: title-LHCb-PAPER.tex 122889 2018-08-17 17:59:55Z pkoppenb $
% ===============================================================================
% Purpose: LHCb-PAPER journal paper title page template
% Author: 
% Created on: 2010-09-25
% ===============================================================================

%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%  TITLE PAGE  %%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{titlepage}
\pagenumbering{roman}

% Header ---------------------------------------------------
\vspace*{-1.5cm}

\vspace*{1.5cm}
\noindent
\begin{tabular*}{\linewidth}{lc@{\extracolsep{\fill}}r@{\extracolsep{0pt}}}

\vspace*{-1.2cm}\mbox{\!\!\!\includegraphics[width=.25\textwidth]{images/uzh_logo_e_pos.eps}}

 & & \today \\ % Date - Can also hardwire e.g.: 23 March 2010
 & & \\
% not in paper \hline
\end{tabular*}

\vspace*{4.0cm}

% Title --------------------------------------------------
{\normalfont\bfseries\boldmath\huge
\begin{center}
% DO NOT EDIT HERE. Instead edit macro in main.tex to keep metadata correct
  \papertitle 
\end{center}
}

\vspace*{2.0cm}

% Authors -------------------------------------------------
\begin{center}
	\Large{\textit{Master thesis of}}
	
	\LARGE{Sascha Liechti}
	
\end{center}

\begin{center}
	\Large{\textit{Supervised by}}
	
	\Large{Prof. Nicola Serra
	
	Dr. Patrick Owen}
	
\end{center}

%\vspace{\fill}
\newpage
% Abstract -----------------------------------------------
\begin{abstract}
  \noindent
Fitting a model to data is an essential part in most High Energy Physics 
analyses. Several 
frameworks to perform this action exist in C++, but no powerful enough 
counterpart exists in Python, a language 
recently becoming 
more and more popular for analyses. With the recent success of deep learning, 
frameworks in Python such as TensorFlow came up, offering a high level 
interface for 
efficient, parallelised computing on modern architectures. 
In this thesis, \zfit{}, a library for model fitting in HEP 
implemented in pure Python and based on top of TensorFlow, is presented. It 
offers a well 
structured model fitting workflow allowing to build composite models from a 
variety of shapes. A high level of customisation is possible 
due to well specified interfaces and convenient base classes allowing to easily 
replace any part in the workflow with a custom implementation. Together with 
the flexibility and scalability of TensorFlow, \zfit{} extends its 
functionality well beyond 
what current model fitting libraries offer.
An overview over the current status of model fitting libraries and the HEP 
requirements will be discussed followed by the structure of \zfit{} and its 
implementation. Finally, examples which 
quantify the performance and demonstrate the feasibility of 
\zfit{} for a whole range of real world applications are shown and an 
additional library for phasespace generation, \zphasespace, is introduced.

  
\end{abstract}

%\vspace*{2.0cm}



%\vspace{\fill}

%{\footnotesize 
% Edit macro in main.tex to keep metadata correct
% \centerline{\copyright~\papercopyright. \href{\paperlicenceurl}{\paperlicence}.}}
%\vspace*{2mm}

\end{titlepage}


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%  EOD OF TITLE PAGE  %%%%%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%  empty page follows the title page ----
\newpage
\setcounter{page}{2}
\mbox{~}



\cleardoublepage