|Posted on September 30, 2012 at 3:35 PM|
This september I went back to Donosti. I'm involved in an on-going project with many people from Jesus Ugalde's group, which aims at paralellizing the PNOF program of Mario Piris. So far, we had parallelized the main bottleneck of the program (the utilization of two-electron integrals to construct the pseudo-Fockian matrix), producing very high parallelization performance yields. However, as the number of cores increases, other processes in the code gain importance and these yields drop. The aim on this three-week incursion in Donosti has been the parallelization of the second most important process in the code. The Donosti InternationalPhysics Center (DIPC), which had employed me in the past, paid for my expenses.
In addition, I have also been busy writing the second Barcelona Supercomputer Centre (BSC) application on PNOF5 calculations. This time, we turn our sight to transition metal (TM) complexes, in a project which would be the very first one to address TMs from a density-matrix functional theory (DMFT) perpective (actually PNOF5 it is not a pure DMFT approach, it is rather an effective-pairing theory but the different is subtle). Little by little PNOF5 it is making its own way among other computational chemistry tools.
Finally, together with Pedro Salvador and Eloy Ramos we have also been working on the local spin decomposition of a few PNOF functionals. Interestingly, the local-spin decomposition has reaveled itself as an stringent test for the cumulant part of the second-order density matrices, and this decomposition has been used to unveil some limitations of natural orbital functionals.
Now I'm back in Girona, and full-time focused on writing the Starting Grant. Results are ready. The idea is clear. Some people have give his opinion on the topic. Some writing has been done. I just need to polish the whole thing in the next few weeks.