Archive for the ‘Quantum chemistry’ Category

ACS now offering RSS feeds for journals

11 January, 2006

RSS has really changed the way I use the Internet, updates of my favourite sites are delivered at customisable intervals and it makes catching up with all the news (most decent forums offer RSS feeds now too) easy.

The good news is that the ACS are now offering RSS feeds of articles ASAP and the TOC alerts for all their journals.

If you’ve never used RSS before, there are some basic capabilites built into some browsers such as Firefox, but I personally find it much easier to use a dedicated aggregator (more like an email client than a web browser) – the following free (as in no money-cost) aggregators seem good to me:

Update: The ACS just don’t seem to get RSS. The feeds tend to get a couple of hundred entries per day, many of which are simple refreshes from the days previous – that’s very likely to cause me to hit the “mark” all as read button!

Webmo as a teaching tool

20 December, 2005

A new version of webmo has been recently released. For those of you unfamiliar with this piece of software it is a web-based front-end for several QC codes such as Gaussian and Molpro.

As part of the my teaching responsibilities in my new post I’ve had to “demonstrate” in the BSc and MChem practical sessions for molecular modelling where webmo is used. I must say that I was very impressed with it, as it is accessed with a web browser it’s OS agnostic and allows students to start running some calculations quickly. Whilst a student undertaking a final year project may benefit more from learning some UNIX and actually looking at the input and output files from QC codes, webmo seems ideal for one-off practical sessions or as part of a short course in QC – the graphical output allows for an easy intrepretation of results, and there’s no need to explain z-matrices as the GUI does a reasonably good job of allowing you to draw structures.

There’s a demo version (free to use) on the webmo website, and the software itself is available in both free (reduced features) and paid for (as Webmo-Pro) versions. I’d recommend anyone undertaking any QC work to check it out.

Schrödinger Maestro free for academic use

13 December, 2005

Schrödinger Inc are offering their molecular mechanics program maestro free for academic use. Although I personally don’t carry out any molecular mechanics calculcations as part of my research (yet), I thought it was probably worth grabbing a copy and giving it a go – it may be handy for producing starting geometries for QC calculations.

First impressions are that it’s quite difficult to use and the demo from the help menu is almost pointless, I was expecting some kind of tutorial to show me how to run calculations and instead I got the program doing various things on it’s own that you couldn’t interact with. Still, it seems capable of producing some really nice images so it may be worth a longer trial.

Density functional repository

29 November, 2005

I stumbled across this site today that acts as a repository for Fortran 77 subroutines that describe density functionals for use in electronic structure programs. The site is associated with the CCLRC so I assume these are the subroutines (or very close to them) that are used in GAMESS-UK.

Looking at the source code it seems that the subroutines have been generated automatically with Maple and the dfauto script that has been described in Comp. Phys. Comm. It may just be the computer/maths geek in me, but something about that seems very cool.

GOTO BLAS libraries make slashdot

28 November, 2005

The GOTO BLAS libraries have been mentioned on slashdot (and on the new york time website). I’ve been using this library for quite a while now and I’ve seen massive improvements in the speed of several codes (including the venerable Gaussian). I can really recommend them to anyone trying to save a little time with code that relies on BLAS or ATLAS.

Glad to see the author is getting some much deserved credit for his good work.

What is an orbital?

22 November, 2005

I thought of this interesting question today – “what is an orbital?”

I asked it to the various PhD students that I share an office with and it made some of them squirm, the answers ranged from the (hopefully) humorous – “they’re a bit like a donut”, to the not bad – “area of space within which there is a certain probability that an electron can be found”. I should probably state that none of these match what my answer would be.

The responses made me wonder if some of the frankly bizarre posts on CCL about how nothing meaningful can be obtained from orbitals (a purely mathematical construct) and only the electron density should be used stem from not properly grasping what an orbital (in ab initio theories) is?? Is this a deficiency in how QC is taught / what’s written in texts?

It would be interesting to get a few comments about how readers would answer “what is an orbital?”

Intel Powerbooks

14 November, 2005

Those of you that know me will probably realise I’m a bit of an Apple fan – owning an iMac for almost a year now. Whilst everyone and their dog seems to have had an opinion about Apple’s switch to using Intel x86 based chips from IBM PPC’s I’ve never really sat down and wrote anything about it…

The event that’s prompted me to do this is several rumour sites suggesting that Intel based Powerbooks and iMacs could well be announced/available in January. I’ve been toying with the idea of replacing my Linux/Windows Toshiba laptop for a while now, with a Powerbook being one of the top candidates – moving to Intel just seems to sweeten the deal! Faster processors, possibility of dual booting with windows (for playing the odd game), and the availability of the Intel compilers all make for an attractive platform, especially as Apple laptops generally seem to be well engineered and actually look quite nice!

 From a quantum chemistry point of view the availability of the Intel compilers should mean software such as GAMESS and Gaussian should be quickly available for the platform (although perhaps not supported – hand-editing of the makefiles will probably be necessary).  I wonder how Intel will make the compilers available though, few people in the academic community seem to realise that the current Intel compilers on the “free” non-commercial license are not allowed to be used to research if you recieve a payment for your work.

Open Source QC packages

10 November, 2005

As a follow up to my post regarding calls for the Gaussian code to be released (for free) into the community I had a look around for any other OSS QC packages – I’ve found two:

  • MPQC (Massively Parrallel Quantum Chemistry) – it has energies and gradients for HF, DFT and MP2. It claims to have a good geometry optimiser but I haven’t personally tested it.
  • PyQuante – python (with some C for speed) quantum chemistry package. The author aknowledges that the package is slow compared with Gaussian, Gamess etc but claims the advantage that all the routines are easily understood from looking at the source code. Although the program can be used as a single package the main goal is to provide routines that can be incorporated into other codes. Features include HF, LDA DFT and CIS

PyQuante looks promising to me, especially the ability to grab what you need from the codebase and use it in your own applications. Although the author doesn’t ask for any help, a couple of extra people working on this could really take it places.

Open Sourcing Gaussian

28 October, 2005

There’s been a lot of noise on the CCL (Computational Chemistry List) recently regarding the  gaussian program package. Gaussian is a general quantum chemistry package that allows you to perform many different types of calculations in pretty much a black-box fashion. A couple of people on the list were getting a bit upset about the number of gaussian techincal support questions arising, which should really be handled by Gaussian Inc themselves.

This discussion lead to “arguments” about whether Gaussian and computational chemistry programs in general should be open source, and that as US government funding goes into Gaussian then US taxpayers should be able to use the program for free (Mike Frisch later posted to the list effectively stating that there is very little US taxpayers money going into Gaussian (except via the purchase of licenses of course) and convinced at least me of their current funding situation). What some people failed to realise is that Gaussian is in fact open source (if you purchase the correct license) and they are making a fudamental mistake of confusing free and open source software!

Whilst having free access to the Gaussian source code with submit access to a version control system would be of benefit (at the very least in money terms) to many people working in this field, what benefits are there for Gaussian Inc in doing this? If the FOSS model is such a good idea for a computational chemistry package, why hasn’t there been a collaboration of theoretical chemists across the globe on such a project from scratch? Perhaps we’re too busy deriving and implementing new features (or applying them to new situations) to bother with writing code that needs to do everything (and needs to do it efficiently). Is there enough people out there able to code well enough (and to be interested in technical support, writing documentation, fixing bugs, etc.) for such a project to happen? Would anyone be able to get funding to create such a project?

I see a lot of people moaning about Gaussian, and asking for a GPL or BSD licensed code. What I don’t see is anyone bothering to invest their time and effort into creating such a package, “put your money where your mouth is”.