I was in a state of intellectual bliss since Monday as I have arrived in Imperial college to attend a five-days summer school on DFT with CRYSTAL code. In the past few years (may be 2 years or so) I have had many doubts and uncertainties in DFT calculation. This summer school is so far the most satisfying short course I ever attended, and I think I have absorbed the knowledge like a sponge soaking fully in water. After talking to many people in the field during the summer school I gain a much improved idea what's going on, and how people are doing research with DFT. A whole lot of research ideas have emerged in my mind of how to proceed with my proposed research projects, as well as generating new ones. I realise what are the knowledge I had insufficiently. On the other hand I also have realised that I am not as weak as I have imagined. Computational physics is really a very powerful method, and its not actually that hard as particle physics.
The most serious insufficiency I had is my very weak basic in quantum mechanics, solid state physics and advanced statistical physics. Over the last one week I really wish that I could have done better for my basic quantum mechanics and statistical physics, which are tremendously important for any serious research in such fields. Fortunately my few months in Delft have allow me to read a bit more on quantum mechanics, and I think I have improved much on my technical knowledge in the computational aspect of quantum mechanics. In the past I don't even know how to solve a slightly more advanced quantum system other than hydrogen atom or finite step wells. A few years back I don't even know what variational method and perturbation methods are. But now I have learned up who to write diffusion Monte Carlo code to solve 3D harmonic oscillators, and am in the process of extending the code to calculate a 9D Josephson junction array that can act as a qubit. If the code is a success, I would be tremendously happy because it proves that I have at least understood some specific technique to solve complicated quantum mechanics.
It seems that I am really hooked to computational physics, with a simple reason: I can understand them and solve them by writting Fortran code. Thijssen is a very dedicated teacher who has won a major teaching prize. The prize reflects the fact the he is very patient and kind heart-ed person. I learn slowly but also steadily from him and from my own effort.
I have read the Prof. Wang of NUS web page
at http://staff.science.nus.edu.sg/~phywjs/cttpnotes.html, bookmarked it, and downloaded some of his material for future reference. When I went thru his webpages and course material I can't resist my excitement. He is just like another Thijssen, perhaps 'deeper'.
The field of computational physics is so vast and full of opportunities. There are many real problems that can be solved using computational methods. In the last few months, I have been thinking of what I can do with it. I realise that one must have hands-on practical knowledge in making real complicated code using advanced programing techniques. I think I have learned many such skill which were very mysterious to me, e.g. Makefile, or calling Netlib library with Fortran. My successful coding experience of DMC for the textbook-standard 3D quantum oscillator has provided much impetus to strengthen my confident in code writing.
Another research topic that have aroused much of my interest is Monte Carlo (MC) and molecular dynamics (MD). There are many things these kind of codes can do, and it is conceptually simple. An American colleague in TU Delft commented that molecular dynamics are simply complicated but not difficult. One just needs to have a good book keeping on the particle's track. I recall that it took me only about a night to code my prototype 3D MD code for a simple gas system. Usually I am slow and take many days to write a code. Yet I manage to code it over a night. =b
I am now thinking of many different research topics to do: Diffusion Monte Carlo to solve high-dimensional quantum mechanical problems, modelling of phase transition using Monte Carlo and MD, modelling of phase transition using DFT and first principle thermodynamics, modelling of defects using DFT and first principle thermodynamics.
I recall that when I lasts visited NUS sometimes ago, I hear for the first time the word 'molecular dynamics'. Then I did not even understand what they meant by this terminology, and what is its relation to DFT. So are the jargons like `supercell', 'modelling of defects' etc. When I look back, I am indeed pleased that all those very remote concepts are now sufficiently (at least subjective) understood. =b
The summer school ends in Friday, and I began my lone London exploration beginning Saturday. The Museum of Natural History and the Albert and Victoria museum are just located really 'next door' to Imperial College. These are great museums, but they are totally no comparison to the famed British museum (at least for my personally). I am glad that I didn't skip the museum visit as I originally plan. I walked around the city alone using a cheap tourist guide since 9 am this morning. Now is already nearly 12 am and I am still not exhausted, except a bit hungry.
21 September 2009 6:59
沒有留言:
張貼留言