about this website

as a chemist i did (of course) not visit the lectures on theoretical or physical chemistry regularely. during my diploma thesis i started with theoretical calculations mainly because i found d-orbitals fascinating and i liked the nice graphics of molecules (especially when in motion and hoping around :-). here i noticed, that i had better listened to these lectures ... well, maybe they could have been better.

moving from parametrizing ligand force fields to DFT in the "learning by doing" way, i had at a certain point the idea (or better dream) to write my own DFT code. i thought, this would be cool, and it would help me to understand how it all works. during my time as a post-doc in tom zieglers group in calgary (thanks tom, it was a great time!) i looked at things like the recurrence formulas for gto integrals or plane waves on the web and in various journals. and i realized that i am basically to stupid (lets say "not skilled enough") to code things like that in a reasonable amount of time. well, that others had done it already (and much much better) was not realy the point. but i gave it up.

one night i came upon an issue of int. j. quant. chem. and a very nice article by thomas beck on "real space methods". i realized that this approach is especially simple and straight forward in the implementation, and i started right away to toy with it at home. at that time i used tcl and C and at some point i got the total energy of the be atom (just one filled MO) reasonbly ok with a very simple pp. then i started to work with Python, Numerical Python and C. serious developement began when i was back in munich in 1999. with the SWIG interface generator the C interface changed. pete joined as a diplomand and got all the gga functionals and the spin polarized version going. the parallelization was the last major change in the whole code. fortunately it is not that many lines (both because of the simplicity of the real space approach and the Python / C combination). after my move to bochum maxim joined the team as a post-doc and we are finally getting into "real calculations". there is still a lot to do and this story is hopefully not at its end, since some things work better as expected but others don't.

at least in one respect i was right: it is definetly cool to write an own DFT code and it helped me a lot to understand how it all works ... not saying that i understand it all ... but it realy helps :-))

rochus (edited 22.07.03)