>I saw on GAMESS manual that three different algorithms are available to calculate the partial charges of the molecules. GEODESIC algorithm is as default. Why is it better than Connoly and CHELPG?
I think GEODESIC is not "better" than Connoly or CHELPG. Basically it was found ESP fitting atomic charges depend a lot on the locations of chosen fitting points. As a results there are several versions of ESP fitting charges, if I remember correctly, MK, CHELP, CHELPG, ..., they are just different from each other for the fitting points' location. ESP fitting charges are also known for their conformational variablity. Thy manual states GEODESIC is less rotation dependent (less conformational variability), in this sense one might claim it is "better". Hope it help.