(All solids - Periodic Table - Home - PM7 Accuracy - Manual)
A frequently asked question is, "Which method in MOPAC is the most accurate?" Although apparently a simple question, there is no simple answer. Several years of work went into developing the PM7 method, and that work made extensive use of the experience gained during the development of the earlier methods. So my answer would be, "PM7." But if you were to accept that answer, you would also need to trust that what I said was also accurate. When there is the opportunity to see an unbroken chain of logic from raw experimental data to a summary of a statistical analysis of the errors in various properties for various methods, that is almost infinitely preferable to hearing someone say, "trust me." The data in the PM7 Accuracy web-page is intended to provide enough information to allow users of MOPAC to form an educated impression of the accuracy of PM7.
The methods in MOPAC were designed to provide practical chemists with a tool for modeling chemical systems. To this end, the focus has been on main-stream chemistry - think sulfuric acid - and on systems with common but interesting structures, for example hydrogen bonds in enzymes. Modeling methods are of little practical use unless they can be used without much effort, i.e., in a reasonable time on readily available hardware. These objectives can be contrasted with the philosophy behind many high-level methods, where the objective is to generate accurate data on small systems, or to model exotic systems or exotic phenomena. Such methods are often very time consuming. This is not an attempt to disparage such methods: if semiempirical methods are to be improved, they will have to depend on reference data generated using high level methods. Indeed, that is exactly what was done in parameterizing the non-covalent interactions using the S22 set.