From 43cf1a820cea8d5b3218627d047dd78e4a152dd4 Mon Sep 17 00:00:00 2001
From: Leland McInnes <leland.mcinnes@gmail.com>
Date: Sun, 7 Oct 2018 22:17:58 -0400
Subject: [PATCH] LOBPCG not so good for smaller matrices.

---
 umap/spectral.py | 32 +++++++++++++++++---------------
 1 file changed, 17 insertions(+), 15 deletions(-)

diff --git a/umap/spectral.py b/umap/spectral.py
index 19c5efd6..c82d68de 100644
--- a/umap/spectral.py
+++ b/umap/spectral.py
@@ -254,21 +254,23 @@ def spectral_layout(data, graph, dim, random_state, metric="euclidean", metric_k
     k = dim + 1
     num_lanczos_vectors = max(2 * k + 1, int(np.sqrt(graph.shape[0])))
     try:
-        # eigenvalues, eigenvectors = scipy.sparse.linalg.eigsh(
-        #     L,
-        #     k,
-        #     which="SM",
-        #     ncv=num_lanczos_vectors,
-        #     tol=1e-4,
-        #     v0=np.ones(L.shape[0]),
-        #     maxiter=graph.shape[0] * 5,
-        # )
-        eigenvalues, eigenvectors = scipy.sparse.linalg.lobpcg(
-            L,
-            random_state.normal(size=(L.shape[0], k)),
-            largest=False,
-            tol=1e-4
-        )
+        if L.shape[0] < 2000000:
+            eigenvalues, eigenvectors = scipy.sparse.linalg.eigsh(
+                L,
+                k,
+                which="SM",
+                ncv=num_lanczos_vectors,
+                tol=1e-4,
+                v0=np.ones(L.shape[0]),
+                maxiter=graph.shape[0] * 5,
+            )
+        else:
+            eigenvalues, eigenvectors = scipy.sparse.linalg.lobpcg(
+                L,
+                random_state.normal(size=(L.shape[0], k)),
+                largest=False,
+                tol=1e-8
+            )
         order = np.argsort(eigenvalues)[1:k]
         return eigenvectors[:, order]
     except scipy.sparse.linalg.ArpackError: