From d325e3ee9eae7dd161b61b16509a25dbafaa77eb Mon Sep 17 00:00:00 2001
From: ampli <amirpli@gmail.com>
Date: Fri, 17 May 2024 01:42:22 +0300
Subject: [PATCH] eliminate_duplicate_disjuncts(), tracon-set.c: Use Fibonacci
 Hashing

Knuth's Method / Fibonacci Hashing
---
 link-grammar/connectors.h     |  3 ++-
 link-grammar/disjunct-utils.c | 10 +++++++---
 link-grammar/tracon-set.c     |  2 +-
 link-grammar/utilities.h      | 12 ++++++++++++
 4 files changed, 22 insertions(+), 5 deletions(-)

diff --git a/link-grammar/connectors.h b/link-grammar/connectors.h
index 0a0177c475..4b534f16f7 100644
--- a/link-grammar/connectors.h
+++ b/link-grammar/connectors.h
@@ -329,6 +329,7 @@ static inline uint32_t string_hash(const char *s)
 }
 
 typedef uint32_t connector_hash_t;
+static const connector_hash_t FIBONACCI_MULT = 0x9E3779B9;
 
 static inline connector_hash_t connector_hash(const Connector *c)
 {
@@ -351,7 +352,7 @@ static inline connector_hash_t connector_list_hash(const Connector *c)
 
 	for (c = c->next; c != NULL; c = c->next)
 #if FEEDBACK_HASH
-		accum = (accum<<6) + (accum<<16) + (accum >> 16) - connector_hash(c);
+		accum = (accum<<7) + (accum<<14) + (accum >> 16) - connector_hash(c);
 #else
 		// Bad.
 		accum = (19 * accum) + connector_hash(c);
diff --git a/link-grammar/disjunct-utils.c b/link-grammar/disjunct-utils.c
index 3901d4d708..fa80b8c61d 100644
--- a/link-grammar/disjunct-utils.c
+++ b/link-grammar/disjunct-utils.c
@@ -223,7 +223,8 @@ static unsigned int count_connectors(Sentence sent)
 typedef struct disjunct_dup_table_s disjunct_dup_table;
 struct disjunct_dup_table_s
 {
-	size_t dup_table_size;
+	unsigned int dup_table_size;
+	unsigned int log2_size;
 	Disjunct *dup_table[];
 };
 
@@ -244,10 +245,12 @@ static inline unsigned int old_hash_disjunct(disjunct_dup_table *dt,
 		i += 19 * connector_list_hash(d->right);
 	if (string_too)
 		i += string_hash(d->word_string);
-	//i += (i>>10);
 
 	d->dup_hash = i;
-	return (i & (dt->dup_table_size-1));
+
+	i *= FIBONACCI_MULT;
+	// Feed back log2(table_size) MSBs.
+	return ((i ^ (i>>(32-dt->log2_size))) & (dt->dup_table_size-1));
 }
 
 /**
@@ -322,6 +325,7 @@ static disjunct_dup_table * disjunct_dup_table_new(size_t sz)
 
 	dt = malloc(sz * sizeof(Disjunct *) + sizeof(disjunct_dup_table));
 	dt->dup_table_size = sz;
+	dt->log2_size = power_of_2_log2(sz);
 
 	memset(dt->dup_table, 0, sz * sizeof(Disjunct *));
 
diff --git a/link-grammar/tracon-set.c b/link-grammar/tracon-set.c
index ef530b6705..a30afb3c8e 100644
--- a/link-grammar/tracon-set.c
+++ b/link-grammar/tracon-set.c
@@ -52,7 +52,7 @@ static tid_hash_t hash_connectors(const Connector *c, unsigned int shallow)
 {
 	tid_hash_t accum = (shallow && c->shallow) ? 1000003 : 0;
 
-	return accum + connector_list_hash(c);
+	return (accum + connector_list_hash(c)) * FIBONACCI_MULT;
 }
 
 #if 0
diff --git a/link-grammar/utilities.h b/link-grammar/utilities.h
index f03b488292..b5b255f246 100644
--- a/link-grammar/utilities.h
+++ b/link-grammar/utilities.h
@@ -534,4 +534,16 @@ static inline size_t next_power_of_two_up(size_t i)
 	return j;
 }
 
+/**
+ * Return log2 of a given power-of-2 \p i.
+ */
+static inline unsigned int power_of_2_log2(size_t i)
+{
+	unsigned int n = 0;
+	while (i >>= 1)
+		n++;
+	return n;
+}
+
+
 #endif /* _LINK_GRAMMAR_UTILITIES_H_ */