← Back to context

Comment by anewhnaccount2

3 years ago

Should the Pareto front not be drawn with line perpendicular to the axes rather than diagonal lines?

9 comments

anewhnaccount2

Reply
JyrkiAlakuijala  3 years ago

Often with this kind of pareto it can be argued that even when continuous decisions are not available, a compression system could keep choosing every second at effort 7 and every second at effort 6 (or any ratio), leading, on the average interpolated results. Naturally such interpolation does not produce straight lines in log space.

penteract  3 years ago

Yes, it should, but it looks like they just added a line to the jxl 0.10 series of data on whatever they used to make the graph, and labelled it the Pareto front. Looking closely at the graphs, they actually miss some points where version 0.9 should be included in the frontier.

  • lifthrasiir  3 years ago

    I think it can be understood as an expected Pareto frontier if enough options are added to make it continuous, which is often implied in this kind of discussions.

    • penteract  3 years ago

      I'm not sure that's reasonable - The effort parameters are integers between 1 and 10, with behavior described here: https://github.com/libjxl/libjxl/blob/main/doc/encode_effort..., the intermediate options don't exist as implemented programs. This is a comparison of concrete programs, not an attempt to analyze the best theoretically achievable.

      Also, the frontier isn't convex, so it's unlikely that if intermediate options could be added then they would all be at least as good as the lines shown; and the use of log(speed) for the y-axis affects what a straight line on the graph means. It's fine for giving a good view of the dataset, but if you're going to make a guess about intermediate possibilities, 'speed' or 'time' should also be considered.

      1 reply →