D’oh. Missed the exploding meteor last night. From the news reports it sounds like a pretty impressive sight.  (N.B. I like the comment on the stuff.co.nz article that says "Faster than a plane = definitely over 10000 km an hour. I don’t know how many planes this guy has travelled in, but doing 10000 km an hour would certainly make a trip from Auckland to London a lot less stress.)

Actually, last night was spent trying to teach our most adorable catty-puss some finer points of mouse-catching etiquette:

1. When one is a cat it is perfectly acceptable to catch small rodents. However, all felines should note the following:

2. All games with one’s catch should be undertaken outside. Taking one’s toy into the house to play with is considered bad manners.

3. After use, it is polite to kill one’s mouse. Bringing it inside in a deceased form and presenting it to one’s owner as a gift is acceptable; leaving it half-alive scrabbling about the kitchen floor for one’s owner to discover later is poor form.

Somehow I suspect the message hasn’t got through.

Back to astronomy. A couple of nights ago I glanced out of the window to see Orion (upside down of course) looking back at me. At least, I think it was Orion; it was hard to tell because I didn’t have my glasses on. What I actually saw was a few glowing splodges in the dark sky, roughly making out the shape of the constellation, with a brigher splodge up and to the right, which I assumed to be Sirius.  My eyes are certainly getting worse as I get older. (Though, Waikato drivers should be pleased to here that on putting my glasses on the stars return to their point-like form.)

Each splodge can be considered the ‘point-spread function’ for my eye. This optics terminology describes how a point source of light, like a star, is mapped imperfectly onto a sensor element (my retina). The broader the point-spread function, the worse the eyesight. It’s a useful thing to know, because you can then model how any picture would appear to your optics. What you’d do is a convolution of the unblurry image with the point-spread function, that is, apply the point-spread function at every point. 

The reverse process ‘de-convolution’ is possibly more useful – if you can work out exactly how an image is being blurred, you can work out what to change about its optics to make the image sharp (i.e. design an appropriate ‘lens’ for it.) 

For those who know about such things, convolutions are easy to do numerically with Fourier Transforms.