mathr / blog / #

I've been hacking on gmndl lately, working out how best to extend it with more high-level exploration features. Meanwhile I wrote a little code that turns angled internal addresses into pairs of external angles (by the braindead method of enumerating all external angles that have the required period and turning them into angled internal addresses, and seeing which match). I used these external angles as input to the external ray drawing in Wolf Jung's Mandel program, then found rough coordinates for their endpoints. I copied these coordinates into some code to refine them using Newton's method to the true location of the nucleus. I got up to period 5 before getting bored and thinking there must be a better (non-interactive) way to do this...

Having the locations, periods, and island-hood status of each of these mu-atoms of periods up to 5, I wrote a small function to find good view ports for them, and stuck the resulting list into gmndl:

I modified gmndl to stop rendering when 10 seconds passed with no points escaping, save the high resolution image, and move to rendering the next view port; so that it could render a sequence of images unattended. The files are named according to their angled internal addresses.