Reducing the Computational Complexity of Electron Correlation Techniques

This Small Business Innovation Phase I project aims to demonstrate that it is possible to reduce the scaling of computational complexity iwth molecular size from quintic to quadratic for the widely used dsecond order moller-plesset method (MP2). MP2 theory is the simplest class of quantum mechanical methods for predicting structure and reactivity of molecules with the proper inclusion of correlations between electrons, as is essential for obtaining useful accuracy. The reduction in the compuational complexity will permit applications of the MP2 method to much larger moleucles that must at present be treated the even more accurate Coupled Cluster Method in Phase II. The reduction is made possible by exploiting natural length scales on which electrons in molecules are localized, and reformulating the defining equations of MP2 theory in local terms that permit rigorous elimination of numerically small quantities. Numerical results on a range of prototypical model systems will be obtained to demonstrate feasibility and to quantify the potential advantages relative to conventional implementations of the MP2 method for treating large molecules.