It is improper to speak of absolute error, as it is dependent on machine error and is practically negligible. The error with respect to the ‘true’ datum is totally dependent on the level of theory used. Disregarding error compensation, to verify the accuracy of a level of theory requires either a comparison with a higher level or with an experimental datum. It is useful to first check the literature for the best level of theory based on the cost-accuracy ratio for the system considered. Instead, to check whether the structure is a minimum structure, it is necessary to do a frequency calculation after optimization (keeping the level of theory constant) and verify that there are no imaginary frequencies.

Fabio Biffoli

Thank you Fabio Biffoli

I approach your time to help me

So, I need to do frequency calculation after optimization, but if I get reasonable results how can be sure on how far that result is accurate? Since the idea here is to do an estimation of Eg for a number of binary and ternary ChGs compounds and finally try to have an empirical relation for Eg. I am quite lost could you refer to similar work or can I contact you?