Changes in version 4.0.2                        

The countsplit package now includes the est_overdispersions() function
which returns a vector of overdispersion parameter estimates for each
row of the expression matrix. This functions builds off of the
sctransform package to obtain reasonable overdispersion estimates while
also accounting for zero rows and other data imperfections.

                        Changes in version 4.0.1                        

The countsplit package will now throw a helpful error message if the
parameter "epsilon" is not passed in correctly.

                 Changes in version 4.0.0 (2023-08-24)                  

Thanks to some Github contributions from Mischko Heming, this version of
the package has been substantially sped up compared to version 3.0.0.
This speedup added a dependency on Rcpp, as the main function is now
implemented in C++.

                        Changes in version 3.0.0                        

This is the initial CRAN submission for countsplit. Compared to previous
versions we have:

1.  Moved the data and tutorials to a new package to save on build time
    and package size.
2.  Fully implemented Poisson and negative binomial count splitting,
    both in their 2-fold and their multi-fold versions.
3.  Streamlined so that all types of count splitting are included in the
    same countsplit() function.