Research

Week 17: Computing Woes

After my parent’s visit, my trip to Mallorca and the YTF conference in Durham, things finally started to settle down again into some semblance of a meaningful rhythm. Officially, the term had started again, which meant I had to check in at the deanery to confirm that I was still in Cambridge, but otherwise the regular pattern of going to the department, working all day, attending seminars, eating dinner in college had returned. This term, however, I’m not attending any lectures, and I’ve added a new locale to my habitations. Most days now, after dinner, I would go to the GSCR for an extra few hours of studying, that is, if I wasn’t being depressed by any of my results.

This slideshow requires JavaScript.

Anyone who spoke to me all week probably saw my disconnected frustration with what I was working on. I spent all week trying desperately to get my paper arXiv:1610.06737 ready for re-submission to the journal. By way of context, in responding to one of the referee’s comments on the work, I was browsing through some of the code that I had written and noticed something odd: one of the steps in the calculation was missing a negative sign. This discovery had the potential to impact my results in a very non-trivial way, and it meant that all of my simulations had to be run again. That proceeded fine, and I had copied all of the results from the server before Christmas, but one step in plotting the code needed a computer program called IDL installed. IDL is a vastly complex program, but all I needed it for was to read in a particularly formatted file and do some basic calculations on it. I didn’t want to pay for it, but I discovered an open-source replica called GDL. My mission, therefore, was to install GDL. But to install GDL, I needed to install PLPlot, even if I wasn’t actually using the PLPlot features. One of the notorious features of open-source programs is that you can’t just run an interactive installer like you would be familiar with if you are a non-programmer, but instead you have to install things from the command line. There were no clear instructions on how to do this, however, so I was basically troubleshooting my way through. To complicate matters further, I don’t have administrator privileges on my Linux machine in the department, so I was using a Linux emulator on my Windows laptop, which complicated some of the steps which were dependent on the operating system of the computer. By Friday, after a month of zero progress, I had GDL installed and was ready to plot my results.

However, my extreme emotional high came crashing down to an extreme emotional low. In the space of 5 lines of my contemporaneous notes, I went from the exclamation “GDL NOW WORKS” to the simple line “Computer crashed – reboot, all programs lost, no data lost”. Through unrelated activity to my project, on Sunday afternoon, my computer bricked up, refusing to get past the loading screen before crashing and restarting. I was devastated; I couldn’t even boot in Safe Mode. Fortunately, I was able to wipe to factory settings without touching any of my saved data, but that meant I had nothing but the default programs on my computer. I had to go through and reinstall everything, and adjust all of the settings of everything to be how I like them. I can be quite particular in how I like my programs to function, especially my preferred email client, Thunderbird.

It wasn’t until Tuesday before everything was working relatively normally again. But there was still more pain to come. The way I had set up the plotting program meant that it was quite slow, especially compared to what I had before. I wrote a bit of my plotting script poorly, in a way which meant the computer would overload its memory after 80 minutes and crash into a Blue Screen of Death. At the time, I couldn’t be bothered fiddling with the code any more, and just wanted to run the code to its completion, which I was able to do by Thursday.

And there was yet another problem I was working on in the meantime. It turned out that there was a key set of data that wasn’t tracked in the extra simulations I had run. I was racking my brain trying to figure out how to extract the extra data from what I had, but during a break from solving that, there was yet another problem. I was plotting what I had, when I saw that my simulations had done basically nothing; there was absolutely no difference between the negative controls and any of my results. Normally in science, this is fine, since it means the result is negative. But there was a paper which had a broadly comparable scenario which told me that the result had to be positive. When I tried to do a back-of-the-envelope calculation to see where things were different, the numbers suggested exactly where I should see a signal, but I couldn’t see anything at all. I left the topic on Friday with a brief note in my notebook “But there’s nothing there… I don’t get it…” After this, I became so frustrated with it that I decided to leave it there, and put it down for a while.

In other news for the week, I had lunch with my graduate tutor for the first time. He was new to the college, but is supposed to act as my liaison if I ever have any problems. The suggestion was, however, that I won’t see much of him over my time in Cambridge. I also went to the Gates Welcome Back tea in the GSCR. This was a great chance to catch up with people I hadn’t seen in a while, bemoan to multiple people about my computing woes, and hear from multiple people about an amusingly inadvertent but quite hilarious execution of a nationwide practical joke. For the victim’s sanity, I won’t describe any more.

Advertisements

One thought on “Week 17: Computing Woes

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s