From b07e03a171359c44b7e4fc2e40e318bcf501efa8 Mon Sep 17 00:00:00 2001
From: Boris Stefanovic <owldev@bluewin.ch>
Date: Tue, 7 Jun 2022 16:04:08 +0200
Subject: [PATCH] EDIT: points do not need to know the cluster they belong to
---
src/cluster.c | 14 ++++++++------
src/cluster.h | 5 ++---
src/kmeans.c | 22 +++++++++++-----------
src/kmeans.h | 8 ++++----
src/linkedlist.c | 43 ++++++++++++++++++++-----------------------
src/linkedlist.h | 13 ++++++-------
src/point.c | 33 ---------------------------------
src/point.h | 28 ----------------------------
8 files changed, 51 insertions(+), 115 deletions(-)
delete mode 100644 src/point.c
delete mode 100644 src/point.h
diff --git a/src/cluster.c b/src/cluster.c
index 84e6beb..939ae5c 100644
--- a/src/cluster.c
+++ b/src/cluster.c
@@ -32,26 +32,28 @@ void cluster_destroy_fpt(cluster_fpt_t* cluster) {
}
-void cluster_add_point_int(cluster_int_t* cluster, point_int_t* point) {
+void cluster_add_point_int(cluster_int_t* cluster, vector_int_t* point) {
//TODO
if (NULL == cluster || NULL == point) return;
+ list_points_append_int(cluster->points, point);
if (NULL == cluster->center) {
- cluster->center = vector_copy_int(point->vector);
+ cluster->center = vector_copy_int(point);
} else {
- vector_int_t* delta = vector_copy_int(point->vector);
+ vector_int_t* delta = vector_copy_int(point);
vector_div_inplace_int(delta, cluster->points->size);
vector_add_inplace_int(cluster->center, *delta);
vector_destroy_int(delta);
}
}
-void cluster_add_point_fpt(cluster_fpt_t* cluster, point_fpt_t* point) {
+void cluster_add_point_fpt(cluster_fpt_t* cluster, vector_fpt_t* point) {
//TODO
if (NULL == cluster || NULL == point) return;
+ list_points_append_fpt(cluster->points, point);
if (NULL == cluster->center) {
- cluster->center = vector_copy_fpt(point->vector);
+ cluster->center = vector_copy_fpt(point);
} else {
- vector_fpt_t* delta = vector_copy_fpt(point->vector);
+ vector_fpt_t* delta = vector_copy_fpt(point);
vector_div_inplace_fpt(delta, cluster->points->size);
vector_add_inplace_fpt(cluster->center, *delta);
vector_destroy_fpt(delta);
diff --git a/src/cluster.h b/src/cluster.h
index 963e417..0e2993a 100644
--- a/src/cluster.h
+++ b/src/cluster.h
@@ -3,7 +3,6 @@
#include <stdlib.h>
#include "linkedlist.h"
-#include "point.h"
#include "vector.h"
@@ -28,9 +27,9 @@ void cluster_destroy_int(cluster_int_t* center);
void cluster_destroy_fpt(cluster_fpt_t* center);
-void cluster_add_point_int(cluster_int_t* cluster, point_int_t* point);
+void cluster_add_point_int(cluster_int_t* cluster, vector_int_t* point);
-void cluster_add_point_fpt(cluster_fpt_t* cluster, point_fpt_t* point);
+void cluster_add_point_fpt(cluster_fpt_t* cluster, vector_fpt_t* point);
void cluster_update_center_int(cluster_int_t* cluster);
diff --git a/src/kmeans.c b/src/kmeans.c
index 55ddccd..7ee9705 100644
--- a/src/kmeans.c
+++ b/src/kmeans.c
@@ -1,18 +1,16 @@
#include "kmeans.h"
-#include "point.h"
-vector_int_t** kmeans_init_clusters_int(const point_int_t** points, const size_t point_count, const size_t nclusters) {
- if (nclusters < 2) return NULL;
- if (NULL == points) return NULL;
- vector_int_t** clusters = calloc(nclusters, sizeof(vector_int_t*));
+cluster_int_t** kmeans_init_clusters_int(const vector_int_t** points, const size_t point_count, const size_t nclusters) {
+ if (NULL == points || point_count < 2 || nclusters < 2) return NULL;
+ cluster_int_t** clusters = calloc(nclusters, sizeof(vector_int_t*));
if (NULL == clusters) return NULL;
// determine range in which we are working
- vector_int_t* min = vector_copy_int(points[0]->vector);
- vector_int_t* max = vector_copy_int(points[0]->vector);
+ vector_int_t* min = vector_copy_int(points[0]);
+ vector_int_t* max = vector_copy_int(points[0]);
for (size_t i = 0; i < point_count; ++i) {
for (size_t p = 0; p < max->dim; ++p) {
- const int_t value = points[i]->vector->data[p];
+ const int_t value = points[i]->data[p];
if (value < min->data[p]) min->data[p] = value;
if (value > max->data[p]) max->data[p] = value;
}
@@ -24,15 +22,17 @@ vector_int_t** kmeans_init_clusters_int(const point_int_t** points, const size_t
center->data[p] = rand_int_range(min->data[p], max->data[p]);
}
// TODO: maybe check center is not already in clusters, although probability is extremely low...
- clusters[i] = center;
+ clusters[i]->center = center;
}
return clusters;
}
void kmeans_int(
- point_int_t** points, const size_t point_count,
- vector_int_t** clusters, const size_t nb_clusters,
+ vector_int_t** points, const size_t point_count,
+ cluster_int_t** clusters, const size_t nb_clusters,
fpt_t (* distance_function)(const vector_int_t*, const vector_int_t*)) {
//TODO
+ bool changed = true;
+ while (changed) {}
}
diff --git a/src/kmeans.h b/src/kmeans.h
index c1c5ff7..34a83c8 100644
--- a/src/kmeans.h
+++ b/src/kmeans.h
@@ -1,15 +1,15 @@
#ifndef PROG_KMEANS_KMEANS_H
#define PROG_KMEANS_KMEANS_H
-#include "point.h"
+#include "cluster.h"
#include "linkedlist.h"
-vector_int_t** kmeans_init_clusters_int(const point_int_t** points, const size_t point_count, const size_t nclusters);
+cluster_int_t** kmeans_init_clusters_int(const vector_int_t** points, const size_t point_count, const size_t nclusters);
void kmeans_int(
- point_int_t** points, const size_t point_count,
- vector_int_t** clusters, const size_t nb_clusters,
+ vector_int_t** points, const size_t point_count,
+ cluster_int_t** clusters, const size_t nb_clusters,
fpt_t (* distance_function)(const vector_int_t*, const vector_int_t*));
diff --git a/src/linkedlist.c b/src/linkedlist.c
index 898c7d3..3ea5994 100644
--- a/src/linkedlist.c
+++ b/src/linkedlist.c
@@ -1,24 +1,21 @@
#include "linkedlist.h"
#include <assert.h>
#include <stdbool.h>
-#include "point.h"
-list_points_node_int_t* list_points_node_create_int(vector_int_t* vec) {
+list_points_node_int_t* list_points_node_create_int(vector_int_t* point) {
+ if (NULL == point) return NULL;
list_points_node_int_t* node = malloc(sizeof(list_points_node_int_t));
if (NULL == node) return NULL;
- point_int_t* point = point_create_int(vec);
- if (NULL == point) return NULL;
node->point = point;
node->next = NULL;
return node;
}
-list_points_node_fpt_t* list_points_node_create_fpt(vector_fpt_t* vec) {
+list_points_node_fpt_t* list_points_node_create_fpt(vector_fpt_t* point) {
+ if (NULL == point) return NULL;
list_points_node_fpt_t* node = malloc(sizeof(list_points_node_fpt_t));
if (NULL == node) return NULL;
- point_fpt_t* point = point_create_fpt(vec);
- if (NULL == point) return NULL;
node->point = point;
node->next = NULL;
return node;
@@ -27,20 +24,19 @@ list_points_node_fpt_t* list_points_node_create_fpt(vector_fpt_t* vec) {
void list_points_node_destroy_int(list_points_node_int_t* node, const bool full) {
if (NULL == node) return;
- if (full) point_destroy_int(node->point);
+ if (full) vector_destroy_int(node->point);
free(node);
}
void list_points_node_destroy_fpt(list_points_node_fpt_t* node, const bool full) {
if (NULL == node) return;
- if (full) point_destroy_fpt(node->point);
+ if (full) vector_destroy_fpt(node->point);
free(node);
}
list_points_int_t* list_points_create_int() {
- list_points_int_t* list = NULL;
- list = malloc(sizeof(list_points_int_t));
+ list_points_int_t* list = malloc(sizeof(list_points_int_t));
if (NULL == list) return NULL;
list->head = NULL;
list->tail = NULL;
@@ -49,8 +45,7 @@ list_points_int_t* list_points_create_int() {
}
list_points_fpt_t* list_points_create_fpt() {
- list_points_fpt_t* list = NULL;
- list = malloc(sizeof(list_points_fpt_t));
+ list_points_fpt_t* list = malloc(sizeof(list_points_fpt_t));
if (NULL == list) return NULL;
list->head = NULL;
list->tail = NULL;
@@ -80,9 +75,9 @@ void list_points_destroy_fpt(list_points_fpt_t* list, const bool full) {
}
-void list_points_append_int(list_points_int_t* list, vector_int_t* vector) {
- if (NULL == vector) return;
- list_points_node_int_t* node = list_points_node_create_int(vector);
+void list_points_append_int(list_points_int_t* list, vector_int_t* point) {
+ if (NULL == list || NULL == point) return;
+ list_points_node_int_t* node = list_points_node_create_int(point);
if (NULL == list->head) { // if list is empty
list->head = node;
list->tail = list->head;
@@ -93,9 +88,9 @@ void list_points_append_int(list_points_int_t* list, vector_int_t* vector) {
++list->size;
}
-void list_points_append_fpt(list_points_fpt_t* list, vector_fpt_t* vector) {
- if (NULL == vector) return;
- list_points_node_fpt_t* node = list_points_node_create_fpt(vector);
+void list_points_append_fpt(list_points_fpt_t* list, vector_fpt_t* point) {
+ if (NULL == list || NULL == point) return;
+ list_points_node_fpt_t* node = list_points_node_create_fpt(point);
if (NULL == list->head) { // if list is empty
list->head = node;
list->tail = list->head;
@@ -107,8 +102,9 @@ void list_points_append_fpt(list_points_fpt_t* list, vector_fpt_t* vector) {
}
-point_int_t** list_points_to_array_int(const list_points_int_t* list) {
- point_int_t** a = calloc(list->size, sizeof(point_int_t*));
+vector_int_t** list_points_to_array_int(const list_points_int_t* list) {
+ if (NULL == list) return NULL;
+ vector_int_t** a = calloc(list->size, sizeof(vector_int_t*));
if (NULL == a) return NULL;
list_points_node_int_t* cur = list->head;
size_t idx = 0;
@@ -121,8 +117,9 @@ point_int_t** list_points_to_array_int(const list_points_int_t* list) {
return a;
}
-point_fpt_t** list_points_to_array_fpt(const list_points_fpt_t* list) {
- point_fpt_t** a = calloc(list->size, sizeof(point_fpt_t*));
+vector_fpt_t** list_points_to_array_fpt(const list_points_fpt_t* list) {
+ if (NULL == list) return NULL;
+ vector_fpt_t** a = calloc(list->size, sizeof(vector_fpt_t*));
if (NULL == a) return NULL;
list_points_node_fpt_t* cur = list->head;
size_t idx = 0;
diff --git a/src/linkedlist.h b/src/linkedlist.h
index adfdeb9..de857ef 100644
--- a/src/linkedlist.h
+++ b/src/linkedlist.h
@@ -2,24 +2,23 @@
#define PROG_KMEANS_LINKEDLIST_H
#include <stdbool.h>
-#include "point.h"
#include "vector.h"
typedef struct list_points_node_int {
- point_int_t* point;
+ vector_int_t* point;
struct list_points_node_int* next;
} list_points_node_int_t;
typedef struct list_points_node_fpt {
- point_fpt_t* point;
+ vector_fpt_t* point;
struct list_points_node_fpt* next;
} list_points_node_fpt_t;
-list_points_node_int_t* list_points_node_create_int(vector_int_t* vec);
+list_points_node_int_t* list_points_node_create_int(vector_int_t* point);
-list_points_node_fpt_t* list_points_node_create_fpt(vector_fpt_t* vec);
+list_points_node_fpt_t* list_points_node_create_fpt(vector_fpt_t* point);
void list_points_node_destroy_int(list_points_node_int_t* node, const bool full);
@@ -55,9 +54,9 @@ void list_points_append_int(list_points_int_t* list, vector_int_t* vector);
void list_points_append_fpt(list_points_fpt_t* list, vector_fpt_t* vector);
-point_int_t** list_points_to_array_int(const list_points_int_t* list);
+vector_int_t** list_points_to_array_int(const list_points_int_t* list);
-point_fpt_t** list_points_to_array_fpt(const list_points_fpt_t* list);
+vector_fpt_t** list_points_to_array_fpt(const list_points_fpt_t* list);
#endif //PROG_KMEANS_LINKEDLIST_H
diff --git a/src/point.c b/src/point.c
deleted file mode 100644
index 3dcb58f..0000000
--- a/src/point.c
+++ /dev/null
@@ -1,33 +0,0 @@
-#include "point.h"
-#include <stdlib.h>
-#include "vector.h"
-
-
-point_int_t* point_create_int(vector_int_t* vector) {
- point_int_t* point = malloc(sizeof(point_int_t));
- if (NULL == point) return NULL;
- point->vector = vector;
- point->cluster = NULL;
- return point;
-}
-
-point_fpt_t* point_create_fpt(vector_fpt_t* vector) {
- point_fpt_t* point = malloc(sizeof(point_fpt_t));
- if (NULL == point) return NULL;
- point->vector = vector;
- point->cluster = NULL;
- return point;
-}
-
-
-void point_destroy_int(point_int_t* cp) {
- if (NULL == cp) return;
- vector_destroy_int(cp->vector);
- free(cp);
-}
-
-void point_destroy_fpt(point_fpt_t* cp) {
- if (NULL == cp) return;
- vector_destroy_fpt(cp->vector);
- free(cp);
-}
diff --git a/src/point.h b/src/point.h
deleted file mode 100644
index 381dabe..0000000
--- a/src/point.h
+++ /dev/null
@@ -1,28 +0,0 @@
-#ifndef PROG_KMEANS_POINT_H
-#define PROG_KMEANS_POINT_H
-
-#include "vector.h"
-
-
-typedef struct point_int {
- vector_int_t* vector;
- vector_int_t* cluster;
-} point_int_t;
-
-typedef struct point_fpt {
- vector_fpt_t* vector;
- vector_fpt_t* cluster;
-} point_fpt_t;
-
-
-point_int_t* point_create_int(vector_int_t* vector);
-
-point_fpt_t* point_create_fpt(vector_fpt_t* vector);
-
-
-void point_destroy_int(point_int_t* cp);
-
-void point_destroy_fpt(point_fpt_t* cp);
-
-
-#endif //PROG_KMEANS_POINT_H
--
GitLab