I have devised (automatically, I have just let it grow) an algorithm, which enlists all the leap years in the Gregorian calendar using the Cosine function. Scraping the ugly constants of 100 and 400.
I’m having difficulty envisioning what problem this solves. Leap years are already defined by a very simple function, and subbing in a cosine for a discrete periodicity adds complexity, does it not?
I think (although Thomas leaves it frustratingly unclear) the point is that this algorithm was discovered by some kind of automatic process—genetic programming or something. (If Thomas is seriously suggesting that his algorithm is an improvement on the usual one containing the “ugly constants” then I agree that that’s misguided.)
Having an algorithm fit a model to some very simple data is not noteworthy either. It’s possible that the means by which the “pure mechanical invention” was obtained are interesting, but they are not elaborated on in the slightest.
I have devised (automatically, I have just let it grow) an algorithm, which enlists all the leap years in the Gregorian calendar using the Cosine function. Scraping the ugly constants of 100 and 400.
Here
I’m having difficulty envisioning what problem this solves. Leap years are already defined by a very simple function, and subbing in a cosine for a discrete periodicity adds complexity, does it not?
I think (although Thomas leaves it frustratingly unclear) the point is that this algorithm was discovered by some kind of automatic process—genetic programming or something. (If Thomas is seriously suggesting that his algorithm is an improvement on the usual one containing the “ugly constants” then I agree that that’s misguided.)
Last line of the article explains the motivation:
Having an algorithm fit a model to some very simple data is not noteworthy either. It’s possible that the means by which the “pure mechanical invention” was obtained are interesting, but they are not elaborated on in the slightest.