Algorithm 2 Search algorithm of parallel hierarchical clustering tree
Input: T = {Ti|i = 1, 2,⋯, M}: hierarchical clustering trees;
1:	Q: query point;
2:	Lmax: maximum leaf size
Output: K nearest approximate neighbors of query point Q
3:	L = 0 ⊲ L is the number of points searched
4:	generate two empty priority queues Pf and Ps
5:	for i = 1 → M do
6:	TraverseTree(Ti, Pf, Ps)
7:	end for
8:	while Pf is not empty and L < Lmax do
9:	N = top of Pf
10:	TraverseTree(N, Pf, Ps)
11:	end while
12:
13:	function TraverseTree(N, Pf, Ps)
14:	if node N is a leaf node then
15:	Search all the points in N and add them to Ps
16:	L = L + |N|
17:	else
18:	C = child nodes of N
19:	Cq = closest node of C to query Q
20:	Cp = C\Cq ⊲ delete Cq from C
21:	Add all nodes in Cp to Pf
22:	TraverseTree(Cq, Pf, Ps)
23:	end if
24:	end function