Friday, July 22, 2016

Friday: Solving problems.

Problem 1: It turns out I had poor data sampling in one thing, resulting in a bunch of low signal-to-noise points corrupting a lot of measurements.  Culling those down to a more reasonable number helps a lot.  It also allows me to see that for one population of points, the comparison paper is probably wrong.

Problem 2: The annoying problem that is supposed to be solved by an eight year old program (that doesn't work anymore because it hasn't been used in eight years) is probably not as hard as I initially thought.  Because the distortion solution I want probably has to be continuous (and differentiable), I can probably write something reasonably quick to do iterative solutions starting from the center of the frame and working outwards.

Also, there's a tropical storm coming this weekend, so it's probably going to be impossible to do any shopping until it's over, because everyone panic buys and then the storm turns out to not be a big deal.

This screenshot is relevant when you read the links.

No comments:

Post a Comment