Fix range of HNSW's M parameter to be >= 2, add tests (#95)

dqii · grubdragon · web-flow · commit 92657501ea0c · 2023-08-27T15:02:02.000-07:00
Co-authored-by: Aryan Agal &lt;20280502+grubdragon@users.noreply.github.com&gt;
diff --git a/src/hnsw/options.c b/src/hnsw/options.c
@@ -188,7 +188,7 @@ void _PG_init(void)
                       "m",
                       "HNSW M hyperparameter",
                       HNSW_DEFAULT_M,
-                      1,
+                      2,
                       HNSW_MAX_M
 #if PG_VERSION_NUM >= 130000
                       ,
diff --git a/test/expected/hnsw_create.out b/test/expected/hnsw_create.out
@@ -53,3 +53,19 @@ EXPLAIN (COSTS FALSE) SELECT * FROM sift_base10k order by v <-> :'v4444' LIMIT 1
          Order By: (v <-> '{55,61,11,4,5,2,13,24,65,49,13,9,23,37,94,38,54,11,14,14,40,31,50,44,53,4,0,0,27,17,8,34,12,10,4,4,22,52,68,53,9,2,0,0,2,116,119,64,119,2,0,0,2,30,119,119,116,5,0,8,47,9,5,60,7,7,10,23,56,50,23,5,28,68,6,18,24,65,50,9,119,75,3,0,1,8,12,85,119,11,4,6,8,9,5,74,25,11,8,20,18,12,2,21,11,90,25,32,33,15,2,9,84,67,8,4,22,31,11,33,119,30,3,6,0,0,0,26}'::real[])
 (3 rows)
 
+--- Validate that M values inside the allowed range [2, 128] do not throw an error
+CREATE INDEX ON small_world USING hnsw (v) WITH (M=2);
+INFO:  done init usearch index
+INFO:  inserted 8 elements
+INFO:  done saving 8 vectors
+CREATE INDEX ON small_world USING hnsw (v) WITH (M=128);
+INFO:  done init usearch index
+INFO:  inserted 8 elements
+INFO:  done saving 8 vectors
+---- Validate that M values outside the allowed range [2, 128] throw an error
+\set ON_ERROR_STOP off
+CREATE INDEX ON small_world USING hnsw (v) WITH (M=1);
+ERROR:  value 1 out of bounds for option "m"
+CREATE INDEX ON small_world USING hnsw (v) WITH (M=129);
+ERROR:  value 129 out of bounds for option "m"
+\set ON_ERROR_STOP on
diff --git a/test/sql/hnsw_create.sql b/test/sql/hnsw_create.sql
@@ -18,4 +18,14 @@ CREATE TABLE sift_base10k (
 \COPY sift_base10k (v) FROM '/tmp/lanterndb/vector_datasets/siftsmall_base_arrays.csv' WITH CSV;
 CREATE INDEX hnsw_idx ON sift_base10k USING hnsw (v dist_l2sq_ops) WITH (M=2, ef_construction=10, ef=4, dims=128);
 SELECT v AS v4444 FROM sift_base10k WHERE id = 4444 \gset
-EXPLAIN (COSTS FALSE) SELECT * FROM sift_base10k order by v <-> :'v4444' LIMIT 10;
+EXPLAIN (COSTS FALSE) SELECT * FROM sift_base10k order by v <-> :'v4444' LIMIT 10;
+
+--- Validate that M values inside the allowed range [2, 128] do not throw an error
+CREATE INDEX ON small_world USING hnsw (v) WITH (M=2);
+CREATE INDEX ON small_world USING hnsw (v) WITH (M=128);
+
+---- Validate that M values outside the allowed range [2, 128] throw an error
+\set ON_ERROR_STOP off
+CREATE INDEX ON small_world USING hnsw (v) WITH (M=1);
+CREATE INDEX ON small_world USING hnsw (v) WITH (M=129);
+\set ON_ERROR_STOP on

Original file line number	Diff line number	Diff line change
`@@ -188,7 +188,7 @@ void _PG_init(void)`
`188`	`188`	`"m",`
`189`	`189`	`"HNSW M hyperparameter",`
`190`	`190`	`HNSW_DEFAULT_M,`
`191`		`- 1,`
	`191`	`+ 2,`
`192`	`192`	`HNSW_MAX_M`
`193`	`193`	`#if PG_VERSION_NUM >= 130000`
`194`	`194`	`,`