Given these conflicting results, Raghavachari did what any other good scientist would do -- he tried other theoretical techniques such as perturbation theory and several different gradient corrected density functional methods. However, instead of a trend emerging, his graph of energies increasingly resembled the swing of a teeter-totter. Faced with the limited predictive power of these methods, Raghavachari turned to QMC.

NCSA's Mitas and Grossman applied the quantum Monte Carlo methods developed by David Ceperley at NCSA. (Ceperley was among the handful of physicists to first apply stochastic QMC approaches in the 1970s.) Their results identified the bowl as the lowest energy configuration for C20, which supports the experiments in which the C20 cage was not observed. What's more, QMC predicted the differences in total energies among the three isomers with an uncertainty of the order of 0.2 eV -- close to kcal/mole range that is considered the level of chemical accuracy.

"We were very pleased with the results," says Mitas. "In fact, some experimentalists told us that had they known what our calculation showed, they would have designed the experiments in a different way, saving a lot of time and effort."