November 13th

Moose comes to the rescue! 

I faced a nasty problem last night involving oscillatory behavior in the simulation. It was showing a tendency to generate high-frequency oscillations that grew to destructive levels. There were a variety of other problems with the overall behavior of the system. I was pondering the problem and couldn’t come up with a general solution. I kept imagining single-factor solutions, but I was wary that these would simply pass the problem onto another factor. Realizing the depth of the problem, I decided to walk away from the computer and mull over the problem downstairs, where we had a fire going in the fireplace. The dogs were camped around the fireplace, soaking up the heat, and I settled down on the couch with a pad of paper and started scribbling random ideas. But no good answer came to me. My frustration steadily grew as the minutes ticked by without a solution. 

I looked up in frustration. Moose was staring at me intently. I was taken aback by the intensity of his stare; Moose is a laid-back dog who just enjoys life as it comes, but here he was staring at me as if to say, “I have something important to tell you.” And just then, the solution flashed into my mind: a relaxation algorithm. By relaxing solutions towards their ideal values, I smoothed out high-frequency oscillations. Of course! Why didn’t I think of that earlier?

Moose laid his head down on the carpet.