|Posted on January 21, 2013 at 7:00 PM|
Update: You can download the talks from this link.
In 2011 I did my third postdoc stay in Prof. Ugalde's group at the Euskal Herriko Unibertsitatea (EHU). Last week three researchers from this group visited our institute. Prof. Mario Piris, Dr. Jon M. Matxain and Dr. Fernando Ruipérez spent the whole week in Girona working with Dr. Pedro Salvador, Eloy Ramos and myself.
Since I left Donosti I pay a visit to the group once in a while (I visited the group on Spring and Autumn 2012), so now I was happy to host some basque visitors in Girona. The topic of the Career Integration Grant (CIG) I was granted in 2011 focused on the study of density matrix functional theory (DMFT) and its applications, and thus the collaboration with Basque country is particularly relevant. Since Prof. Piris joint the EHU, the group of Prof. Ugalde is actively working on a particular form of DMFT known as the natural orbital functional theory (NOFT) through the development and application of Piris-NOFs (PNOF). Recently, I programmed a parallel (MPI) version of PNOFid code which yields very promising efficiencies. Hence, together with the people at the EHU we obtained computer time at the Mare Nostrum (MN) supercomputer center (more than one million hours were granted on different PNOF projects thus far). These hours have been used to prove PNOF5 capabilities in small biological systems and transition metal complexes. This year three publications will be submitted concerning these topics.
Now, together with Dr. Pedro Salvador and Eloy Ramos, we are analyzing the performance of PNOF5-PT2 (the ontop second-order perturbation correction to PNOF5, recently implemented by Prof. Piris) as a benchmarking tool. To this aim, we have requested more computer time in MN in order to test PNOF5-PT2 in a series of difficult molecular systems.
In addition, also with Pedro, Eloy and Mario, we are also improving the convergence speed with PNOF5, which is the main weakness of the method. And, last but not least, we have also applied our recently developed spin decomposition scheme to PNOF5. Interestingly, this scheme does not only unveils the limitations of PNOF5 to attain the local atomic spin value upon dissociation, but it also points out which part of PNOF5 functional needs to be improved to fulfill this property. Therefore, local spin values can be used as an stringent test for DMFT and DFT functionals. The reduced number of such tests makes this finding particularly relevant for the development of new functional expressions.