Glyceroletrinitrate "Nitroglycerole" C3H5N3O9

Here's some "explosive" news. We show the optimized structure by Kimball.exe and compare it with the optimized structures obtained by Gaussian94 and the model chemistries RHF/Sto-3G and RHF/6-31G:

 
This is a comparison of the computed overall structure, left for Kimball.exe, right for G94W RHF/6-31G, starting from the Kimball structure, but optimized. The vibrational analysis produces a stable stationary state.
 

The overall geometry does not change much between Kimball and optimized RHF/6-31g. However, there are many changes in bond lengths. Their main characteristics is the exaggerated difference in the lengths of the formal N-O and N=O bonds with Kimball. Since Kimball does not allow partial double bonds, that is what creates the differences in length groups 6-8, 9-11, and 12-14. The yellow triangles are differences between optimized Sto-3G and 6-31G. For two out of the three NO-bonds Sto-3G comes out worse than Kimball in comparison to 6-31G! This is also demonstrated with the energetics (lower left corner), where the difference in total energy between Sto-3G and 6-31G is four times larger than that between Kimball and 6-31G.