Computational Chemistry and Biochemistry

Research

High-level QM/MM applications

PHBHWe are interested in the study of biochemical systems through the application of high-level quantum chemical methods. Chemical reactions when catalyzed by enzymes or when taking place in solution, are hard to handle through computation due to the large number of atoms involved. However, even in such systems, the reactions are usually confined to a small region which is usually termed as active site. Through the use of QM/MM approaches, this active site can be treated at a different level from the surrounding environment. Parameterized force fields are used in the description of the latter, while in the chemically relevant region, where quantum effects are significant, electronic structure approaches are used. We have been applying local correlation methods in the description of the reaction energetics with some success. Research is now focused in improving the way the systems are sampled in their reactive path, as well as in improving the reliability of the local correlation approaches.


Many-body approximations in correlated methods

MBThe application of ab initio correlated computational methods is severely limited by the exponential scaling of computational cost relative to the system size. With this problem in mind, we are developing lower-order scaling approaches through the use of many-body approximations. Instead of treating the system as a whole, one divides the system into fragments, and calculates contributions through a N-body expansion. Many properties can be accurately described in this way, particularly solvation effects. We have developed a suite of tools for many-body expansions, with interface to the Molpro and Gaussian quantum chemistry programs. The Python tools for Many-Body Approaches (PyMBA) suite is written in Python and makes use of the embarrassingly parallel features of these approximations through MPI commands.


Chemical software for embedded systems

The current trend in hardware seems to be the further development of heterogeneous architectures. For example, the use of CPU-GPU hybrid computational systems, where the graphic accelerators card works together with the host system to increase the speed of computation. For the full use of such emerging hardware structures, one need to rethink the algorithms, and develop suitable software. In collaboration with the group of Prof. Leonel Sousa (Technical University of Lisbon), we are exploring the use of GPUs for accelerating computational chemistry applications.