?

Log in

No account? Create an account
Research Update - Nick [entries|archive|friends|userinfo]
Nick

[ website | gagravarr.org ]
[ userinfo | livejournal userinfo ]
[ archive | journal archive ]

Research Update [Mar. 10th, 2003|12:50 pm]
Nick
[mood |productive]
[music |Catatonia - Mulder And Scully]

Haven't written much about my research of late, other than as part of why I don't want to keep doing it for the next 3 years. So, since it's about time for me to start writing lots about it, I thought I should wibble a bit about it.

In case you didn't know, I'm in the Madden Group, doing my 4th year masters project in computational chemistry. My thesis title is "Structure Of Polarisable Ionic Fluids".

I'm told that I need to start the process of writing up in the next 3-4 weeks. Supposedly, first third of writing up is spent half writing, half doing more simulations / analysis on existing data. Then, the closer to handing in (12 noon on Friday 7th week) you get, the more time is spent writing, and the less time is spent on getting new data.

Owing to my supervisor being quite busy early on in the year, and my ability to work quite well with only minimal supervision, I did a very large amount of work on liquid ScCl3 before being given my main project. This means that my thesis is now going to have two parts, firstly the "main bit" on MD (molecular dynamics, newtonian simulations) simulations of polarisable ions to try to match up with the "Restricted Primative Model", and then a load of stuff on ScCl3. The others aren't going to be doing much on their MX3 systems for their thesis, as they didn't get as much done on them, and got given their main projects sooner.

Now, a few small issues. Firstly, the first word of my thesis title, Structure. For ScCl3, getting the approx structure out was easy. Getting a good one would be really hard, and I just never got round to reading up on the main technique to improve it, "Reverse Monti-Carlo fitting of structural data". Instead, I've been doing lots of work on the Dynamics in ScCl3, escpecially looking at how they change with Temperature and Density (i.e. mapping their changes across part of the phase space). So, looks like I'm going to have to talk lots about how the "interesting" structure of liquid ScCl3 (almost molecular like at times, it has amazing long range order) plays into the dynamics, and how the dynamics is the best way to look at it. Probably now done about half the final simulations I need for ScCl3, just need 2 more different temperature sets, and then some different polarisability runs, to tie in with the other section.

The "main" bit of my project is done with Gus. His PHD thesis was on modifying the Restrictive Primative Model to cope with polarisability. The RPM does OK but not stunningly at explaining ionic systems, and work in our group shows that introducing polarisability greatly helps things, so Gus did some work (others have also done some) on including polarisability.

My work is coming at the problem from the MD side, not the theory side. So, stuffing Gus's model into the MD code, churning it out without worrying too much about the theory side (only caring about the system parameters etc), and seeing what comes out. Impressivly, we get a got match on some of the structure data. Since the MD code can do dynamics and Gus's can't, I've been working mostly on looking at how the dynamics of this theoretical system change with polarisability and temperature (going to start on density changes mid week). Again, notice dynamics not structures.

One hope in sight for structures is that I've been plotting dynamics against Epsilon Infinity, which is related to the polarisability by a funky formula who's name I ought to learn. However, we see a temperature dependence that Gus can't currently explain. Luckily, we can use the short P part of the Szz plot (that's a structure thingy) to find Epsilon Zero, by fitting it to a "a + bx^2 + cx^4" curve. Hopefully, plotting dynamics against Epsilon Zero will remove the pesky temperature dependence, but either way will include some structures stuff.

Right, I'm off to go write a program to fit graphs to quadratic functions, such fun. Then I just need to work out how to free enough space to get some of my ScCl3 data (30 gigs and counting) from the supercomputer centre to my machine for post processing, while still keeping hold of the polarisability work data (25 gigs), all on a 35gb disk....
linkReply