From 6e05a7aab2c436132f863cf797a7f1bc2beadf17 Mon Sep 17 00:00:00 2001
From: Jakob Nybo Nissen <jakobnybonissen@gmail.com>
Date: Tue, 27 Aug 2024 09:19:43 +0200
Subject: [PATCH] Fixup: Add comment

---
 vamb/reclustering.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/vamb/reclustering.py b/vamb/reclustering.py
index 914e6b36..63f9c94d 100644
--- a/vamb/reclustering.py
+++ b/vamb/reclustering.py
@@ -23,6 +23,11 @@
 EPS_VALUES = np.arange(0.01, 0.35, 0.02)
 
 
+# TODO: This might be slightly algorithmically inefficient. The big problem is that it re-heapifies
+# the heap whenever a bin is emitted, which makes it O(N^2).
+# To solve this, I need some datastructure which is like a heap, but which allows me to update
+# arbitrary elements in the heap.
+# This can be solved with a MutableBinaryMaxHeap in Julia's DataStructures.jl, for inspiration.
 def deduplicate(
     scoring: Callable[[set[ContigId]], float],
     bins: dict[BinId, set[ContigId]],