LANL*

Hey everybody!

Sorry it’s been so long… We had quite the experience trying to get a steady internet connection set up in our apartments. Anyway, this would be the much awaited ‘physics post’ for my research experience (although I’ll probably take up more space writing about non-physics stuff).

For those of you who hate condensed versions and like the drawn out version with way too many technical terms, you can see what my project is all about here. This is the article that my mentor (Josef) and his colleagues have already published. Here is the condensed version:

My project’s goal is to develop a new method for calculation the magnetic drift invarient L* used in many space weather applications.  I’m sure many of you know what L* is, but here’s a little refresher straight from Josef’s paper: ‘L* is dircctly proportional to the integral of the magnetic flux contained within the surface defined by a charged particle moving in the Earth’s geomagnetic field.’ The problem that we face when calculating L* is that, typically, an L* calculation can require ~10^5 calls to a model. This takes a LONG TIME! To put it in perspective, each call takes about a minute on the computers here (wait ’til I tell you about the cluster I’m going to be working with!). Doing some simple math (1*10^5), that would take 100000 minutes! I don’t know about you, but I just don’t have the time to sit around and wait for that. What Josef has been working on (and what I’m now working on), is to develop a neural network surrogate model that can calculate L* millions of times faster than the standard direct integration techniques. The neat thing about a charged particle’s motion through Earth’s magnetic field is that it remains ‘trapped’ in certain places (provided it does not get lost in the Earth’s atmosphere or travel past the magnetopause). The particle exhibits three types of motion - gyro, bounce, and drift. The gyromotion is the partcle circling around a magnetic field line. The bounce is the particle ‘bouncing’ up and down along a magnetic field line. This motion depends on the particle’s pitch angle with respect to the magnetic field line. The last motion is the particle ‘drifting’ around the Earth, parallel to the magnetic field lines.

That’s about it for the physics part - my main job here at LANL is to train a neural network (a big-@$$ computer network) to learn what results come from what input parameters. I think about it (not sure how correct this is) as giving the computer examples and building up its ability to instinctively spit out an answer. Someone who is more computer-savvy might want to explain the exact workings… So basically, my job is to provide the network with many real examples (from data collected the past four decades) and then validate the results with other sets of data that the network hasn’t seen. LANL came up with a clever name for their model of L* - LANL*!

So far, I’ve been stretching my brain to learn Python (yes, named after Monty Python!) and understand what Josef already has written. I have already written a couple of scripts that call upon the 40 year data table and organized that data into dictionaries in Python (basically a list with many key-value pairs. ex: key-value=’Year’-1963). Then I also wrote a script that transforms TAI (I think) time value, which is basically seconds since the start of 1958, into a human-understandable reading of day-month-year-hour-minute-second. So far it’s been a lot of fun using Python and troubleshooting on my own. I highly suggest all computer language illiterate people to at least try some simple stuff with a language. I’m not claiming I’m an expert, but it has been a very enjoyable experience so far, not to mention it will be very useful in the future.

This week was very exciting for me because I was finally told to log in to the computer cluster that I will be working with. I don’t know how much I can tell you about the cluster because I haven’t been able to find public info on the web. It is only described on LANL’s internal network. But I can tell you that it is monstrous. I get nervous everytime I log in to the cluster through my terminal. Oh well, maybe I’ll get used to the idea of having tons of computing power at my fingertips…

Well, enough talk about work. This past week we went hiking to three different places. We went up Pajarito Ski Area, down Hamilton Camp Trail, and up Cerro Grande in the Jemez Mountains. Tons of fun. I just realized I don’t know how to put a picture in here… That’s sad. A little help, Isaac!!

So, I guess I lied when I said there would be more non-work writing. I’ll get some pictures up and expand on my social activities once I figure out how…

Hope everyone is enjoying their summers!

-Martín