Almost there...

Somewhat Disturbing Moments In Public Transport #53: This morning, I looked to my right to see the uni student next to me reading through what looked like a presentation to be delivered. Something about children's eating habits in China and Japan. But it began with "Good morning, my name is Haster".

Somewhat Disturbing Moments In Public Radio #28: Discovering that there are *two* songs (and quite different ones) based on the "it puts the lotion on its skin" segment of Silence of the Lambs.

( Final house inspection last night... )

Some days, even the failures are beautiful.

The smoothing algorithm I'm using is an iterative process. If you think of the initial function value as 'Signal' + 'Noise', the idea is to apply a transformation that preserves Signal and multiplies Noise by something less than one. Repeat this enough times, and the noise is damped out leaving me with a nice smooth function.

For some data sets, however, it ends up multiplying by more than one. (I'm still trying to figure out exactly when this occurs.) When this happens, I get a sort of positive-feedback effect.

Today, I input a very simple data set: 100 x 100 grid, fix some points along the boundary at values of 1 or -1. The result of this positive feedback is a rather beautiful sort of resonance. At first it looks like a smooth set of curves. In fact, if you look closely, it oscillates wildly between top and bottom, with the extremes of the oscillation marking those 'curves':

http://www.flickr.com/photos/59012110@N00/1987010/
http://www.flickr.com/photos/59012110@N00/1987011/
http://www.flickr.com/photos/59012110@N00/1987009/
http://www.flickr.com/photos/59012110@N00/1987008/
http://www.flickr.com/photos/59012110@N00/1987017/

Courage

Hats off to Jeremy Marks at the very aptly-named 'Courage' ministry. It takes a lot of guts to rethink what you've been teaching people for fourteen years, acknowledge that it isn't working, and look for a better answer.

(Ganked from RoO.)