Large--Scale Angular Correlations in CDM Models

We generate artificial Lick maps using $N$--body simulations and compare the angular correlation function, $w(\vartheta)$, measured from the simulations with the APM correlation. For the Gaussian CDM model, neither the standard biassed model nor a more evolved model (as suggested by the COBE data), reproduce the correlations on large angular scales. We come to a similar conclusion about CDM models with positively skewed initial fluctuation distributions. In contrast, models with initially negatively skewed fluctuations produce a $w(\vartheta)$ that declines much more gently on large scales. Such models are therefore in principle capable of reconciling the lack of large--scale power of the CDM spectrum with the observed clustering of APM galaxies.