Monday, October 28, 2013

Monday: Four days

Today I spent a bunch of time discovering that cfitsio still sucks.  I eventually sorted out that "new algorithm X" is at least 20% better than the old one, and some rather large percentage of the remaining 80% are due to a known error in the algorithm.  Tomorrow I need to parallelize the test I'm runnning, and go through 6000 images instead of just one.  You know, because central limit theorem.

Except my pencil is made out of steel.

No comments:

Post a Comment