delvingbitcoin
Merging incomparable linearizations
Posted on: November 30, 2023 19:30 UTC
The email discussion delves into the construction of a feerate diagram for the combined lists $L_G + L_B$, focusing on the comparison between different subsets and their corresponding diagrams.
The core idea presented is to evaluate the feerate efficiency of a combined list by comparing it to the original list $L$. To do this, the sender suggests an approach where the initial segments, characterized by having a feerate of $f$ or lower, are collectively considered, followed by the remaining portion of $L_G$ not included in these segments, which has a higher feerate.
A key concept introduced is the 'kinking' up of the diagram when including the suffix of $L_G$. This kinking indicates that the subset being considered is more efficient than the former chunks, including up to $c_j$. A visual aid is mentioned as being added to better illustrate this point.
Furthermore, the discussion evolves towards an alternative representation method using a ray plot. Here, the suggestion is to plot points $P(\gamma_i)$, representing different segments for specific ranges along the x-axis, with their corresponding $R(\gamma_i)$ values plotted on the y-axis. The goal is to compare two diagrams by checking if one lies entirely below the other, implying a more optimal chunking reflected by a monotonically decreasing line.
Finally, there appears to be a bit of confusion regarding whether to plot the feerate $R(\gamma_i)$ or the fee $F(\gamma_i)$. It's clarified that plotting the fee would result in a monotonically increasing line since fees cannot be negative, contrasting with the earlier implication of plotting the feerate.