Wednesday, September 17, 2003

I left the house late this morning expecting that I would miss the delivery of the last part of my book order. When I returned, though, there was no sticker on my door to say that UPS tried to deliver while I was out. I'm not sure whether to be glad, because I didn't miss the delivery after all; or impatient, because I want the book to come soon. I think I'll settle for glad.

I finished the proof of a theorem that I conjectured was true nearly three weeks ago. Ultimately, the proof was short and simple, but I'm more inclined to pat myself on the back for finally finding it than to chastise myself for not finding it sooner. Still, I expect this result to be useful in my evaluation of a source of (mercifully rare) error growth in the structured eigenvalue solver I've looked at for the past few months. Yesterday, I did some calculations based on the sensitivity of the coefficients of a polynomial to perturbations in the roots. Tomorrow, perhaps, I will spend some time actually devising test code based on these results.

Actually, I expect I will spend much of tomorrow wearing my mechanics hat. There is a homework set due on Friday in my fluids class, and with the IEEE meeting on Thursday, I expect little time to finish the homework after tomorrow, and I'm a little confused by it right now. I hope that I'll find out in office hours tomorrow that I'm confused on something simple and that the problem is, in fact, as simple as it sounds like it ought to be. I'm also supposed to meet with Sanjay tomorrow afternoon to compare notes on our respective numerical solutions of a simple 1D thermoelastic damping problem. I'd like to review my code before we meet, since he mentioned that the solution has a boundary layer that I may not have resolved. I don't expect that any refinements I make will substantially change my results, but I should still check.

I worked from home and from a local coffee shop for most of today. I was in the office just long enough to have a conversation with Kahan regarding evaluation of accuracy for polynomial root finders. Then I went to the cafe and sat with my cookie and coffee and pad of paper for a while to think thoughtful thoughts. Anant came by a little later, and we had an interesting conversation. Conversations with Anant are always interesting; he works on nuclear magnetic resonance and I work on numerical computations and linear algebra (and other odds and ends), but we both have at least a little understanding what the other does, and we both like to explain what we're working on to anyone who looks even remotely interested. So I tell him about eigenvalue problems and he tells me about nuclear spins and we both walk away happy and a little more informed about the world.

My days have been full and interesting recently. It keeps me happy, but it's exhausting, too.

  • Currently drinking: Red tea