Structure generation computational requirements

The generation of a structure from spectroscopic data alone (a true unknown where there is no proposed synthetic route nor an a priori postulated structure) is a much more significant problem. The amount of computation required is clearly much higher, but even in 2003 unless ab initio calculations are required, this does not constitute a significant constraint on how rapidly this can be achieved. However, the acquisition of the pre-requisite manifold spectroscopic data does constitute a significant constraint and it is less clear where the applications of such an approach lie. Whereas the automated confirmation of postulated structure can be done rapidly and frees experienced spectroscopists from a repetitive and tedious task, the generation of an unknown structure from various spectroscopic data is a challenge in which most experienced spectroscopists revel. More particularly, because the interactive manner in... 

Molecular orbital descriptions offer a number of significant advantages over conventional resonance structures. For one, they often provide more compact descriptions, e.g., the LUMO in planar benzyl cation conveys the same information as four resonance structures. Second, orbital descriptions are quantitative, compared to resonance structures which are strictly qualitative. Finally, molecular orbital descriptions may be applied much more widely than resonance descriptions. Of course, molecular orbital descriptions cannot be generated using a pencil as can resonance structures, but rather require a computer. It can be argued that this does not constitute a disadvantage, but rather merely reflects a natural evolution of the tools available to chemists. 

An unbiased simulation may use a truncated basis set that represents the lowest complex surface harmonics of the atomic valence shell on a Born-Oppenheimer framework with the correct relative atomic masses. For small molecules, of less than about fifteen atoms, the nuclear framework could perhaps even be generated computationally without assumption. The required criterion is the optimal quenching of angular momentum vectors. The derivation of molecular structure by the angular-momentum criterion will be demonstrated qualitatively for some small molecules. 

Currently the synthesis program at the University of Toronto, running on an IBM 3033 computer, requires about four milliseconds to generate the reactant B , let us say that will give rise to a product A. If Bf consists of two coreactants the time will be slightly more. If the molecules have simple structures, such as those seen in elementary organic chemistry textbooks, then the time could be as little as half a millisecond. 

Among the advantages with 2D-based descriptors are their rapid speed of computation for large sets of compounds and that they do not require 3D structures. Thus, these descriptors avoid the problem and compute times associated with 3D structure generation and conformational analysis, even though there are programs available that generate reliable 3D structures, for example, CORINA [5]. 

The differences of the initial structure generation procedure will influence the quality of NMR structures, because of the initial structure dependency of the calculations. The ability to cross the potential barrier (i.e., conformational space exploration efficiency) definitely depends on the molecular dynamics conditions (typically simulation time). The conditions, however, are usually not validated, because of limitations by computational resources. The computing power required is linearly dependent on the simulation time, and thus has been a strong limitation of the calculations. Furthermore, temperature and steric repulsion weighing schedules as well as the force field will influence the structural quality. 

---