This page will look much nicer in a browser that supports CSS, or with CSS turned on.

Uncertain Principles

Physics, Politics, Pop Culture

Friday, July 11, 2003

Work Habits of the Easily Frustrated

Back when I was a post-doc, I used to have the occasional arguments with my boss regarding the proper approach to intractable problems. He was firmly of the opinion that there was no problem that couldn't be solved by just staying in the lab continuously until it was solved, while I tended to disagree. After a few hours of futility, my brain just shuts down, and I stop being able to think of new approaches to the problem. Once that happens, unless the problem in question can be solved by rote, mechanical manipulations, I'm useless.

I've always found that when I'm stuck on some problem that just won't give, the best course of action is to set it aside for a few hours, and start fresh after a break. At least half of the major lab problems that got solved in my post-doc days were solved when the solution occurred to me on the walk home after shutting down for the night, or on the walk back in the next morning. Most of the time, I wanted to kick myself for having missed something that seemed so incredibly obvious after a few hours of sleep.

I was reminded of this this morning, when I thought of the solutions to two different problems that were plaguing me yesterday over my morning blogroll. One of them was experimental, having to do with the strange grounding habits of the data acquisition board one of my students is setting up. The other was theoretical, having to do with some number crunching I've been doing (at home, due to software licensing issues), and involved simulated Fabry-Perot transmission peaks ending up in strange places. Both were nagging me all last evening, and continued to bug me this morning. Then, somewhere between Crooked Timber and Making Light, the two solutions hit me almost simulataneously: The experimental problem could be easily resolved by grounding one of the other inputs, while the theoretical problem was the result of two factor-of-two errors that nearly offset one another.

(Vaguely Relevant Geek Joke: A colleague mentioned seeing an oral exam in which a student worked through a derivation on the chalkboard, and ended up with has answer having the wrong sign. "I seem to have made a sign error," he said. "No," corrected one of his professors, "you seem to have made an odd number of sign errors.")

A quick trip into the lab, and a strategically placed 50-ohm terminator fixed the one, and a half-hour of number crunching at home resolved the other. Which leaves me the afternoon free, to spend kicking myself for not seeing the answers sooner...

Posted at 12:27 PM | link | follow-ups |