\begin{table}%t1
\par
\caption{\label{Error_norms_conv}Variation in the typical orders of magnitude of error norms with
 the convolution length $K$ for an unlensed CMB map simulated using
 NFFT.}  
%\centerline
 {
\begin{tabular}{cccccc}
\hline \hline
Oversampling&
Convolution&
nside&
$l_{\max}$&
Maximum&
rms
\\
factor&
length&
&
&
error&
error
\\
$\left(\sigma\right)$&
$\left(K\right)$&
&
&
$\left(E_{\infty}\right)$&
$\left(E_{2}\right)$
\\
\hline\\[-3.2mm]
2&
4&
1024&
2048&
${\sim} 10^{-8}$&
${\sim} 10^{-8}$
\\
%\hline 
2&
6&
1024&
2048&
${\sim} 10^{-11}$&
${\sim} 10^{-12}$
\\
%\hline
2&
8&
1024&
2048&
${\sim} 10^{-11}$&
${\sim} 10^{-13}$
\\
\hline 
\end{tabular}}
\medskip
\par
 $E_{2}$ and $E_{\infty}$ are the quadratic norm and
 maximum(relative) norm, respectively.
\end{table}