“If we had enough computational power and mathematical knowledge, we could use wave equations to solve all the open questions in chemistry,” enthused Brad Ganoe, postdoctoral fellow in James Shee’s lab at Rice University.
However, since those capabilities are far off – if even achievable – Dr. Ganoe is working to create new, easy-to-use theoretical tools to develop better approximations that can then help focus experimental research. His research draws upon his expertise in many-body quantum mechanics, the field that underpins the description of chemistry from first principles and the basis of theoretical chemistry.
While much of chemistry is well-described by traditional orbital models, many interesting systems involve strong correlations, which present difficult cases for traditional wavefunction and density functional theory methods. In many-body physics, with two or more quantum particles, the math becomes too difficult to solve so you need a good approximation of many-body waves, starting from a trial wave function.
“For example, we can’t accurately solve around half of transition metal complexes today, important in the development of catalysts, drugs and other materials,” he said. “Improved theoretical techniques could prove valuable to many current problems in chemistry.”
He is combining two relatively new techniques, auxiliary-field quantum Monte Carlo (AFQMC) and correction of spin contamination via polynomial-scaling projection, in clever ways to provide better information for experimental research.
AFQMC-based methods, which Dr. Shee was instrumental in developing, have a stochastic framework using iterations to converge on an answer for quantum mechanical systems. Their major drawback, however, is that if the starting point isn’t accurate, the errors compound and the convergences become too imprecise to be useful. And while spin projection can provide good analytical approximations, there is room to further explore its usefulness for the stochastic method in AFQMC.
Dr. Ganoe is developing novel spin projection techniques by extending imaginary-time evolution, the technique traditional AFQMC and some quantum algorithms are based on, to the spin domain. He computes a spin-projected state “ansatz,” or educated guess, to provide a good trial wavefunction before proceeding to the energy calculation. This then can significantly reduce the precision error in the iterative calculations in the stochastic process.
“What I do is very basic research,” Dr. Ganoe said. “The Welch grant is amazing. It gives me time to sit down and think. We are tool makers, creating what applied chemists need to solve the questions they find fascinating.”
Oxygenase, the process by which plants turn carbon dioxide into oxygen, is one important strongly correlated problem which traditional methodologies have not been able to answer. Dr. Ganoe hopes improved theoretical techniques could help researchers unravel how biology does it.