Applications of SE methods

A good, brief overview of the performance of MNDO, AMI, and PM3 is given by Levine [57], Hehre has compiled an extremely useful book comparing AMI with MM (chapter 3), ab initio (chapter 5) andDFT (chapter ) methods for calculating geometries and other properties [58], and an extensive collection of AMI and PM3 geometries is to be found in Stewart s second PM3 paper [44]. 

02 A of experimental (althongh the AMI C-S bonds are abont 0.06 A too short), and angles are nsnally within 3° of experimental (the worst case is the AMI HOF angle, which is 7.1° too big). 

Of AM 1 and PM3, neither has a clear advantage over the other in predicting geometry, although PM3 C-H and C-X (X-O, N, F, Cl, S) bond lengths appear to be more accurate than AMI. MP2 geometries are considerably better than AMI and PM3, but HF/3-21G and HF/6-31G geometries (Fig. 5.23 and Table 5.7) are only moderately better. 

Errors are given as AM 1/PM3/MP2. In some cases (e.g. MeOH) errors for two bonds are given, on one line and on the line below. A minus sign means that the calculated value is less than the experimental. The numbers of positive, negative, and zero deviations ftom experiment are summarized at the bottom of each coluttm. The averages at the bottom of each column ate arithmetic means of die absolute values of the errors. 

---

Application of SE-methods with mechanic support (e.g. Program inversion, hatch tools)... 

---