Any text-based information I share with students I can do via the LMS, so I never found a real use for a blog. Being an itinerant visitor, I didn't need a stable home page for my research (which depends on students and available equipment). I do, however, use web-based materials for classes. Since computational tools are now readily available, I use some assignments with simplified versions. To this end, I write a lot of server-side code to access the computational tools so students need not install anything on a local machine.
One such site I have been revisiting in preparation for the fall is a web-based queue system for running Monte Carlo calculations for statistical thermodynamics. Students access the main page which shows all jobs associated with the queue.
A student can submit a new job- either a simple Ising model (magnetic system) or a rare-gas system. A web form lets them input the parameters:
Data from completed jobs can be viewed
Selecting "Get Data" shows a plot of the major results, and provides a link to download all the output data of the calculation as a ZIP archive.
The peak in the heat capacity of the Ising calculation occurs at the Curie temperature of the magnet; melting and boiling points can be determined from the rare gas calculation. The downloadable data contains the spin structures for the Ising calculation and radial distribution function plots for the rare gas calculation.
I have also made sites for visualization of molecular structures; for the thermodynamics class the main idea was to view some molecular dynamics results to understand how energy redistributes in molecules after reactions. This redistribution is significant in calculations of biochemical (enzyme/substrate) systems. The site is here. There are several molecules and animations which can be selected.
This site was done before I learned how to use mouse input to rotate the view; the three viewing angles have to be changed by the sliders at the top of the right-hand side. For the default molecule, theta=2.00, phi=1.80, psi=2.00 provide a nice "dancing teddy bear" view of the vibrations.
I found this to be a reasonably workable system. The whole thing is running on a Raspberry Pi sitting on my living room floor. The biggest problem I had was making sure nobody unplugged the Pi to plug in a phone charger. I don't have a tremendous amount of experience, so there are conditions (certain input data combinations) which can crash the server. If you get an error trying to look at the animation, please let me know so I can restart the server (in particular, I think the animation site may be vulnerable to too many users at once due to the naive implementation- one of the things I'm looking at now).
Kevin, thank you for sharing this process with us. Your "here" link isn't working. I am curious as to how you teach your students to use this platform for Monte Carlo. Do you provide training or is it relatively easy to maneuver once they open the system? As well, is this something they would use beyond this class?
ReplyDeleteThanks. I fixed the link (I'd mistyped the target address). This platform is pretty easy; we discuss the actual calculation in class. There is a general problem in computational chemistry (and physics) right now in that its markedly easier to gain access to high-powered programs than to understand them. The models in this exercise are much simpler than those in the literature, but the Monte Carlo process is the same. For the Ising model, this is a 2-dimensional model, while most real systems are 3-dimensional (2-dimensional magnets are still actively researched, though). It does exhibit all the properties we associate with magnetic systems, though. Similarly, the molecular dynamics calculations they use are real (they're adapted from actual research calculations), but "stripped down" to allow students to understand the process.
DeleteAs far as using them, the second image is the entire input for the Ising model. There are only two parameters: the applied field and the coupling constant. We discuss in class the range of values for these parameters, and students are given a scheme to determine which parameters they use in their assignment (so they don't all run the same calculation). The quantity "beta" is really the temperature in disguise (strictly speaking, the reciprocal of the temperature); it is the fundamental variable in statistical thermodynamics, so we talk about it a lot in class. For the rare gas system, it's similarly simple: a minimal number of parameters.
The techniques are common in research- materials chemistry uses both types of calculation quite frequently, and biochemical systems are often modeled by both molecular dynamics and Monte Carlo processes. Most of my students will never actually use them, but they will definitely interact with people who do use them. My goals are that they be able to understand the descriptions in a paper they read, and be able to have a productive conversation about the methods with a colleague.