Subtree-entropy proof note: credit Goubko (2016), record prior art

[ms609]

Records the prior art for the subtree-entropy table (.MinSubtreeEntropy()) in dev/min_subtree_entropy_proof.md.

• Theorem 2 (heavy-leaf Huffman form) is a special case of Goubko 2016, MATCH Commun. Math. Comput. Chem. 75:3-27 (arXiv:1502.01216), Thm 3 with Lemmas 19-20. Theorem 3 (tie-completeness) follows from his equality clause and Lemma 8. Neither result is new.
• What remains ours: the heavy-leaf reduction of the recurrence, the short capping + Glassey & Karp proof, the independent flip proof, and the O(log s) enumeration of optimal splits that gives the O(n log n) table.
• Adds the general statement (any symmetric concave split cost; strict concavity needed for Thm 3; linear h fails at n = 8, s = 6) and drops the hypothesis s > n/2 from Thm 3.
• Corrects the Forst & Thorup Lemma 5.2 quote and adds the explicit n = 7 counterexample.
• Records the exact-arithmetic re-verification for n <= 300, adds DOIs, and adds a summary table of attributions.
• The only change in src/ is a comment in nni_max_step.cpp crediting the sources. User-facing docs make no novelty claim and are unchanged.

Still unread: Cai et al. 2018 (AMC 333:435, doi:10.1016/j.amc.2018.03.117), which claims a stronger result than Goubko's.

[ms609]

Waiting until you've incorporated Cai& 2018.

[ms609]

Cai et al. (2018) is now read. It covers only the vertex-weighted Wiener index, dropping Goubko's degree-monotonicity hypothesis; general concave costs appear only as future work in its section 4. Goubko (2016) therefore remains the source. The doc is updated.

[github-actions]

Performance benchmark results

Call	Status	Change	Time (ms)
ClusteringInfoDistance(tr280)	⚪ NSD	-0.29%	9.75, 9.62 → 9.67, 9.88
ClusteringInfoDistance(tr50)	⚪ NSD	1.8%	6.77, 6.52 → 6.55, 6.58
LAPJV(test2000)	⚪ NSD	6.8%	116, 114 → 110, 106
LAPJV(test40)	⚪ NSD	0.16%	0.0184, 0.0185 → 0.0185, 0.0183
LAPJV(test400)	⚪ NSD	0.47%	3.35, 3.3 → 3.31, 3.31
MutualClusteringInfo(tr280)	⚪ NSD	-2.02%	19.1, 20.1 → 20.5, 19.5
MutualClusteringInfo(tr50)	⚪ NSD	-2.66%	18.7, 19.1 → 19.4, 19.8
NNIMaxStep(2000, normalize = TRUE)	⚪ NSD	-0.03%	0.184, 0.184 → 0.184, 0.184
NNIMaxStep(300, normalize = TRUE)	⚪ NSD	-0.13%	0.0301, 0.03 → 0.03, 0.0301
NNIMaxStep(300)	⚪ NSD	1.31%	0.00844, 0.00838 → 0.00826, 0.00835
PathDist(postTrees)	⚪ NSD	1.11%	2.62, 2.57 → 2.56, 2.57
PhylogeneticInfoDistance(tr180)	⚪ NSD	-0.35%	60.2, 60.2 → 60.4, 60.5
PhylogeneticInfoDistance(tr50)	⚪ NSD	1.02%	42.6, 42.7 → 42.3, 42
RobinsonFoulds(tr280)	⚪ NSD	0.19%	4.71, 4.65 → 4.65, 4.68
RobinsonFoulds(tr280)	⚪ NSD	0.21%	4.5, 4.49 → 4.48, 4.48
RobinsonFoulds(tr50)	⚪ NSD	-0.03%	4.43, 4.42 → 4.4, 4.44

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.57%. Comparing base (dfa69b0) to head (bf58543).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #217   +/-   ##
=======================================
  Coverage   97.57%   97.57%           
=======================================
  Files          60       60           
  Lines        6065     6065           
=======================================
  Hits         5918     5918           
  Misses        147      147           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.