Querying when tree_init=False

[jlmelville]

Is there an easy way to query an index built when tree_init=False (I haven't looked at the latest source, but I am assuming it should default to random initialization for the nearest neighbor descent part).

I hoped the following would work (using random data as input):

import numpy as np
from numpy.random import default_rng
import pynndescent

rng = default_rng()

nrows = 1000
ncols = 2
data = np.array(rng.standard_normal(nrows * ncols)).reshape(nrows, ncols)

nn_index = pynndescent.NNDescent(data, n_neighbors=15, tree_init=False)
neighbors = nn_index.query(data, k=15)

but I get:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-40-16461a0356f0> in <module>
----> 1 neighbors = nn_index.query(data, k=15)

~/dev/pynndescent/venv/lib/python3.8/site-packages/pynndescent/pynndescent_.py in query(self, query_data, k, epsilon)
   1584         """
   1585         if not hasattr(self, "_search_graph"):
-> 1586             self._init_search_graph()
   1587 
   1588         if not self._is_sparse:

~/dev/pynndescent/venv/lib/python3.8/site-packages/pynndescent/pynndescent_.py in _init_search_graph(self)
    953 
    954         if not hasattr(self, "_search_forest"):
--> 955             tree_scores = [
    956                 score_linked_tree(tree, self._neighbor_graph[0])
    957                 for tree in self._rp_forest

TypeError: 'NoneType' object is not iterable

This is with version 0.5.1.

[jlmelville]

This no longer causes an error with 0.5.2.