Virtual Molecules

Here at this illustrious University, researchers have access to some pretty hefty supercomputing capabilities. It was actually the second most powerful academic supercomputer in Europe when it was initially set up. As a matter of fact, there are three disciplines which use more of the cluster’s computing power than anyone else — astronomers, chemists and computer scientists. Normally, I’d put in a joke here about clusters computing clusters (atom clusters, galaxy clusters…), but I’m not in a particularly nerdy mood, so I’ll let you make one up yourself.

Realising that I represent two of these three groups, it would be rude of me not to do some CPU-intensive uber-calculations, now wouldn’t it?

Well, in all seriousness, it’s mainly the fact that a big part of astrochemistry is spectroscopy, and a big part of theoretical spectroscopy is Density Functional Theory (DFT). DFT calculations let you map out the electron density of a molecule (as shown in this image of benzyne — redder means higher electron density). It’s a rather nicer method than having to use Schrodinger’s wave equation too. The Hartree-Fock (HF) method uses that, and it only really works for atomic hydrogen. Add any more electrons and it breaks (which is where the Born-Oppenheimer approximation comes in).

Anyway, DFT (and it’s younger sibling, time dependent DFT) approach the problem from the opposite perspective, building molecular orbitals from the top down, as opposed to the bottom up approach of HF calculations. Actually, DFT is pretty damn versatile. As well as calculating spectroscopic frequencies, you can use it to determine information on transition states, activation energies, potential energy surfaces…

But I should stop ranting now, because few people find theoretical chemistry that interesting (and most people find it rather scary). Incidentally, one of the most used computations astronomers like to play with are Monte Carlo simulations. Chances are, I’ll have to learn how to use them too at some point… But all in due time.

Suffice to say that my laptop now contains even more scientific software (namely Gabedit and GAMESS) for doing DFT work with. Gabedit, incidentally, is a GUI for most major computational chemistry software, and it’s pretty damn good. It’s nice to be able to run Linux apps through X11, though if I’m honest, most command line interfaces still intimidate me somewhat…

About Invader Xan

Molecular astrophysicist, usually found writing frenziedly, staring at the sky, or drinking mojitos.
This entry was posted in chemistry, Imported from Livejournal and tagged , . Bookmark the permalink.

5 Responses to Virtual Molecules

  1. Your Dad helped develop Spartan!? How cool is that!
    It is pretty cool. Generally I remember he’d be working at home as a kid, come across some problem, call up the folks at Wavefunction and tell then, “Hey, I was working with such and such protein, and I tried to do this and it gave me this error message…” Basically, my father was (and still is) pushing software to the edge, and it’s those sorts of applications that the Wavefunction people appreciated feedback.
    *cough* *cough* *haorganicck* *hachemistsck* :P
    (Probably not entirely true, but they’re fun to tease…)

    Very true. Although I guess the only exception I know is my dad. Back before there were Physical Organic degrees, there were just Physical and Organic, so my dad, being the crazy guy he is, did both.
    DFT overestimates the frequencies? That I didn’t know… How so exactly? And is it the same for TDDFT?
    I don’t remember exactly how it overestimates the frequencies (I think I learned about this in Quantum Chem, but I don’t feel like digging out my notes this second), I just know if you read most of the DFT literature, the reported frequencies are higher than the experimental values that are reported. I actually briefly mentioned this back in 2003 in this paper. I can’t say anything about TDDFT calcs, mostly because I’m pretty sure my exposure to doing DFT work was all time independent.

  2. invaderxan says:

    Always lovely to know there are other happy spectroscopic people out there, thank you! :)
    Your Dad helped develop Spartan!? How cool is that!
    Must be nice to be able to get a free copy. I’d probably be using it myself if it didn’t cost me lots of money. Scientific software, it seems, is always either free or extortionate!
    Uni machines do have Spartan, admittedly, but it’s annoying having to sit in a computer room… :P
    If people don’t have an appreciation for theoretical chemistry, then they don’t have a prayer understanding experimental chemistry to it’s fullest potential.
    *cough* *cough* *haorganicck* *hachemistsck* :P
    (Probably not entirely true, but they’re fun to tease…)
    DFT overestimates the frequencies? That I didn’t know… How so exactly? And is it the same for TDDFT?

  3. invaderxan says:

    Heheh… Po-tay-to, po-tah-to. ;)
    I’m sure there are a lot of people who’d argue for hours about the differences between physical chemistry and molecular physics — but I’m not one of them. It’s the physics of chemicals… Though you must admit, a lot of physics simply ignores chemistry altogether!

  4. Theoretical chemistry is very physics-y!
    And quantum chemistry? That’s just physics with a different name.

  5. But I should stop ranting now, because few people find theoretical chemistry that interesting (and most people find it rather scary).
    I would hardly call that ranting. If people don’t have an appreciation for theoretical chemistry, then they don’t have a prayer understanding experimental chemistry to it’s fullest potential.
    I’ll have to look into Gabedit… that sounds pretty interesting. Personally I have Spartan ’04 on my Mac, and I just love it. Then again I tend to get Spartan for free since my father is a long-time user of the program (being a computational chemist himself), and has worked pretty closely with the developers on improving each new addition. :)
    Of course the only problem is that DFT is still always over-estimating the frequencies of vibrational modes. Although if you know the vibrational modes you’re pairing up with DFT calcs, it does make life a bit easier assigning frequencies with experimental results.
    [/computational love from a spectroscopist]

Comments are closed.