1: \begin{definition}\label{D:Diagram}

2: A \emph{quiver} is a quadruple $\xI=(V,E,s,t)$ where~$V$ and~$E$ (the ``vertices'' and the ``edges'', respectively) are sets and $s,t\colon E\to V$ (the ``source'' and ``target'' map, respectively). An element $e\in E$ is thus viewed as an ``arrow'' from~$s(e)$ to~$t(e)$.

3: 

4: A \emph{diagram} in a category~$\xC$, \emph{indexed by~$\xI$}, is a functor from~$\xI$ to~$\xC$. Formally, it is given by a pair $(\Phi_{\mathrm{o}},\Phi_{\mathrm{m}})$, where~$\Phi_{\mathrm{o}}$ (resp., $\Phi_{\mathrm{m}}$) is a map from~$V$ (resp., $E$) to the object class (resp., morphism class) of~$\xC$, such that $\Phi_{\mathrm{m}}(e)$ is a morphism from~$\Phi_{\mathrm{o}}(s(e))$ to~$\Phi_{\mathrm{o}}(t(e))$ for any $e\in E$. There is no composition of arrows in~$\xI$, thus nothing to worry about at that level.

5: 

6: The usual definitions of a \emph{natural transformation} and a \emph{natural equivalence} carry over to diagrams without modification. For diagrams $\Phi,\Psi\colon\xI\to\xC$, we denote by $\Phi\cong\Psi$ the natural equivalence of~$\Phi$ and~$\Psi$. For categories~$\xA$ and~$\xB$, we say that a diagram $\lift\colon\xI\to\xA$ \emph{lifts} a diagram $\Psi\colon\xI\to\xB$ with respect to a functor $\Phi\colon\xA\to\xB$ if $\Psi\cong\Phi\lift$ (cf. Figure~\ref{Fig:liftdiag}). In all our lifting problems, we will need to specify additional composition relations among the arrows in the range of~$\Gamma$. (The first such instance in this paper is Theorem~\ref{T:FirstNonLift}: the additional requirement is $h\circ s=h$.)

7: \begin{figure}[htb]

8:  \[

9: \xymatrixrowsep{2pc}\xymatrixcolsep{1.5pc}

10: \def\labelstyle{\displaystyle}

11: \xymatrix{

12: \xI\ar[rr]^{\lift}\ar[rrd]_{\Psi} && \xA\ar[d]^{\Phi}\\

13: && \xB

14: }

15:  \]

16: \caption{The diagram $\lift$ lifts the diagram $\Psi$ with respect to $\Phi$}

17: \label{Fig:liftdiag}

18: \end{figure}

19: \end{definition}

20: