About Red Green Code Book Matching

[fanluoyyds]

Hello, your job is very good. But I found in the paper that the matching of red and green codebook uses top-k pruning and Blossom algorithm, but the corresponding method in the code is to use greedy algorithm for matching. I hope to get your reply

[yutchina]

Hello, your job is very good. But I found in the paper that the matching of red and green codebook uses top-k pruning and Blossom algorithm, but the corresponding method in the code is to use greedy algorithm for matching. I hope to get your reply

The pairing implementation of HierarchicalCodebook in hc.py is indeed a greedy algorithm. This is because I initially implemented it using a greedy strategy, and later wrote a separate file that performs codebook matching with top‑k pruning plus the Blossom algorithm. I’ve already provided the pre‑matched codebook pairs on GitHub, so the main role of hc.py now is actually to load these precomputed pairs. Because of that, I forgot to update hc.py to use the new top‑k + Blossom implementation.

In fact, updating it would not be difficult: once you have the codebook tensor, both Blossom‑based matching and top‑k pruning are quite straightforward to implement.

I’ve been pretty busy recently, so I haven’t had time to make this change yet, but I plan to update this part when I have time.

[fanluoyyds]

Thank you for your answer, which successfully resolved my confusion

[fanluoyyds]

Hope to update the code as soon as possible, thank you