Skip to content

Commit 1ac61dd

Browse files
amejiarosarioamejiarosario
committed
stats library
1 parent 692bd18 commit 1ac61dd

File tree

5 files changed

+317
-216
lines changed

5 files changed

+317
-216
lines changed

‎src/data-structures/hash-maps/hash-map.js

Lines changed: 40 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -1,3 +1,5 @@
1+
/* global BigInt */
2+
13
/**
24
* Hash Map data structure implementation
35
*
@@ -24,30 +26,54 @@ class HashMap {
2426
}
2527

2628
/**
27-
* Decent hash function where each char ascii code is added with an offset depending on the possition
29+
* Use the binary represenation (IEEE 754) as a hashCode
2830
* @param {any} key
2931
*/
30-
static hashCode(key) {
31-
let hashValue = 0;
32-
const stringTypeKey = `${key}${typeof key}`;
33-
34-
for (let index = 0; index < stringTypeKey.length; index++) {
35-
const charCode = stringTypeKey.charCodeAt(index);
36-
hashValue += charCode << (index * 8);
37-
}
32+
static hashCodeForNumber(number) {
33+
const buffer = new ArrayBuffer(8); // 8 bytes for float64
34+
const dataView = new DataView(buffer);
35+
dataView.setFloat64(0, number); // set as float64
36+
const longBits = dataView.getBigInt64(0); // read as long int (BigInt)
37+
// gurantee only positive (big) integers
38+
return longBits > 0 ? longBits : BigInt(2 ** 63) + (longBits * BigInt(-1));
39+
}
3840

39-
return hashValue;
41+
/**
42+
* Polynomial hsah codes are used to hash String typed keys.
43+
* A string is a sequence of characters encoded
44+
* using Unicode's numerical value.
45+
* @param {any} key converted to string
46+
*/
47+
hashFunctionForString(key) {
48+
return Array
49+
.from(key.toString())
50+
.reduce((hashIndex, char) => this.hashCodeToIndex((41 * hashIndex) + char.codePointAt(0)), 0);
4051
}
4152

4253
/**
4354
* A hash function converts keys into array indices
4455
* @param {any} key
45-
* @returns {Number} array index given the bucket size
56+
* @returns {BigInt} array index given the bucket size
4657
*/
4758
hashFunction(key) {
48-
const hashValue = HashMap.hashCode(key);
49-
const bucketIndex = hashValue % this.buckets.length;
50-
return bucketIndex;
59+
if (typeof key === 'number') {
60+
return this.hashCodeToIndex(HashMap.hashCodeForNumber(key));
61+
}
62+
return this.hashFunctionForString(key);
63+
}
64+
65+
/**
66+
* Multiply-Add-Divide (MAD) compression
67+
* @param {number} hash hashCode
68+
* @param {number} size bucket array size
69+
* @returns {number} array bucket index
70+
*/
71+
hashCodeToIndex(hash, size = this.buckets.length) {
72+
const prime = BigInt(6700417); // prime number > size
73+
const a = prime - BigInt(2); // integer from range [1..(p-1)]
74+
const b = prime - BigInt(1); // integer from range [0..(p-1)]
75+
const hashIndex = (((a * BigInt(hash)) + b) % prime) % BigInt(size);
76+
return parseInt(hashIndex, 10);
5177
}
5278

5379
/**

‎src/data-structures/hash-maps/hash-map.spec.js

Lines changed: 124 additions & 122 deletions
Original file line numberDiff line numberDiff line change
@@ -2,131 +2,133 @@ const HashMap = require('./hash-map');
22
// import HashMap from './hash-map';
33

44

5-
describe('without collisions', () => {
6-
let hashMap;
7-
8-
beforeEach(() => {
9-
hashMap = new HashMap();
10-
});
11-
12-
it('gets values', () => {
13-
hashMap.set('test', 'one');
14-
expect(hashMap.get('test')).toBe('one');
15-
});
16-
17-
it('should increase load factor and size', () => {
18-
expect(hashMap.getLoadFactor()).toBe(0);
19-
expect(hashMap.size).toBe(0);
20-
hashMap.set('test', 'one');
21-
expect(hashMap.getLoadFactor()).toBe(1 / 16);
22-
expect(hashMap.size).toBe(1);
23-
});
24-
25-
it('should overwrite values and keep same size', () => {
26-
hashMap.set('test', 'uno');
27-
expect(hashMap.get('test')).toBe('uno');
28-
hashMap.set('test', 'dos');
29-
expect(hashMap.get('test')).toBe('dos');
30-
expect(hashMap.size).toBe(1);
31-
});
32-
33-
it('should return with has', () => {
34-
expect(hashMap.has('test')).toBe(false);
35-
hashMap.set('test', 'uno');
36-
expect(hashMap.has('test')).toBe(true);
37-
});
38-
39-
it('should update keys on deletes', () => {
40-
hashMap.set('Despacito', 'Luis Fonsi');
41-
hashMap.set('Bailando', 'Enrique Iglesias');
42-
hashMap.set('Dura', 'Daddy Yankee');
43-
44-
expect(hashMap.delete('Bailando')).toBe(true);
45-
expect(hashMap.delete('Bailando')).toBe(false);
46-
expect(hashMap.get('Bailando')).toBe(undefined);
47-
48-
expect(hashMap.keys()).toEqual(['Despacito', 'Dura']);
49-
});
50-
});
51-
52-
describe('with many values (and collisions)', () => {
53-
let hashMap;
54-
55-
beforeEach(() => {
56-
hashMap = new HashMap(2, 0);
57-
58-
hashMap.set('Pineapple', 'Pen Pineapple Apple Pen');
59-
hashMap.set('Despacito', 'Luis Fonsi');
60-
hashMap.set('Bailando', 'Enrique Iglesias');
61-
hashMap.set('Dura', 'Daddy Yankee');
62-
hashMap.set('Lean On', 'Major Lazer');
63-
hashMap.set('Hello', 'Adele');
64-
hashMap.set('All About That Bass', 'Meghan Trainor');
65-
hashMap.set('This Is What You Came For', 'Calvin Harris ');
66-
});
67-
68-
it('gets values', () => {
69-
hashMap.set('test', 'one');
70-
expect(hashMap.get('test')).toBe('one');
71-
expect(hashMap.get('Dura')).toBe('Daddy Yankee');
72-
expect(hashMap.get('Bailando')).toBe('Enrique Iglesias');
73-
});
74-
75-
it('should increase load factor and size', () => {
76-
expect(hashMap.getLoadFactor()).toBe(4);
77-
expect(hashMap.size).toBe(8);
78-
hashMap.set('test', 'one');
79-
expect(hashMap.getLoadFactor()).toBe(9 / 2);
80-
expect(hashMap.size).toBe(9);
5+
describe('HashMap Tests', () => {
6+
describe('without collisions', () => {
7+
let hashMap;
8+
9+
beforeEach(() => {
10+
hashMap = new HashMap();
11+
});
12+
13+
it('gets values', () => {
14+
hashMap.set('test', 'one');
15+
expect(hashMap.get('test')).toBe('one');
16+
});
17+
18+
it('should increase load factor and size', () => {
19+
expect(hashMap.getLoadFactor()).toBe(0);
20+
expect(hashMap.size).toBe(0);
21+
hashMap.set('test', 'one');
22+
expect(hashMap.getLoadFactor()).toBe(1 / 16);
23+
expect(hashMap.size).toBe(1);
24+
});
25+
26+
it('should overwrite values and keep same size', () => {
27+
hashMap.set('test', 'uno');
28+
expect(hashMap.get('test')).toBe('uno');
29+
hashMap.set('test', 'dos');
30+
expect(hashMap.get('test')).toBe('dos');
31+
expect(hashMap.size).toBe(1);
32+
});
33+
34+
it('should return with has', () => {
35+
expect(hashMap.has('test')).toBe(false);
36+
hashMap.set('test', 'uno');
37+
expect(hashMap.has('test')).toBe(true);
38+
});
39+
40+
it('should update keys on deletes', () => {
41+
hashMap.set('Despacito', 'Luis Fonsi');
42+
hashMap.set('Bailando', 'Enrique Iglesias');
43+
hashMap.set('Dura', 'Daddy Yankee');
44+
45+
expect(hashMap.delete('Bailando')).toBe(true);
46+
expect(hashMap.delete('Bailando')).toBe(false);
47+
expect(hashMap.get('Bailando')).toBe(undefined);
48+
49+
expect(hashMap.keys()).toEqual(['Despacito', 'Dura']);
50+
});
8151
});
8252

83-
it('should overwrite values and keep same size', () => {
84-
hashMap.set('test', 'uno');
85-
expect(hashMap.get('test')).toBe('uno');
86-
hashMap.set('test', 'dos');
87-
expect(hashMap.get('test')).toBe('dos');
88-
expect(hashMap.size).toBe(9);
53+
describe('with many values (and collisions)', () => {
54+
let hashMap;
55+
56+
beforeEach(() => {
57+
hashMap = new HashMap(2, 0);
58+
59+
hashMap.set('Pineapple', 'Pen Pineapple Apple Pen');
60+
hashMap.set('Despacito', 'Luis Fonsi');
61+
hashMap.set('Bailando', 'Enrique Iglesias');
62+
hashMap.set('Dura', 'Daddy Yankee');
63+
hashMap.set('Lean On', 'Major Lazer');
64+
hashMap.set('Hello', 'Adele');
65+
hashMap.set('All About That Bass', 'Meghan Trainor');
66+
hashMap.set('This Is What You Came For', 'Calvin Harris ');
67+
});
68+
69+
it('gets values', () => {
70+
hashMap.set('test', 'one');
71+
expect(hashMap.get('test')).toBe('one');
72+
expect(hashMap.get('Dura')).toBe('Daddy Yankee');
73+
expect(hashMap.get('Bailando')).toBe('Enrique Iglesias');
74+
});
75+
76+
it('should increase load factor and size', () => {
77+
expect(hashMap.getLoadFactor()).toBe(4);
78+
expect(hashMap.size).toBe(8);
79+
hashMap.set('test', 'one');
80+
expect(hashMap.getLoadFactor()).toBe(9 / 2);
81+
expect(hashMap.size).toBe(9);
82+
});
83+
84+
it('should overwrite values and keep same size', () => {
85+
hashMap.set('test', 'uno');
86+
expect(hashMap.get('test')).toBe('uno');
87+
hashMap.set('test', 'dos');
88+
expect(hashMap.get('test')).toBe('dos');
89+
expect(hashMap.size).toBe(9);
90+
});
91+
92+
it('should return with has', () => {
93+
expect(hashMap.has('test')).toBe(false);
94+
hashMap.set('test', 'uno');
95+
expect(hashMap.has('test')).toBe(true);
96+
});
8997
});
9098

91-
it('should return with has', () => {
92-
expect(hashMap.has('test')).toBe(false);
93-
hashMap.set('test', 'uno');
94-
expect(hashMap.has('test')).toBe(true);
95-
});
96-
});
97-
98-
describe('#rehash', () => {
99-
let hashMap;
100-
101-
beforeEach(() => {
102-
hashMap = new HashMap();
103-
104-
hashMap.set('Pineapple', 'Pen Pineapple Apple Pen');
105-
hashMap.set('Despacito', 'Luis Fonsi');
106-
hashMap.set('Bailando', 'Enrique Iglesias');
107-
hashMap.set('Dura', 'Daddy Yankee');
108-
hashMap.set('Lean On', 'Major Lazer');
109-
hashMap.set('Hello', 'Adele');
110-
hashMap.set('All About That Bass', 'Meghan Trainor');
111-
hashMap.set('Wake Me Up', 'Avicii');
112-
hashMap.set('Brother', 'Avicii');
113-
hashMap.set('Faded', 'Alan Walker');
114-
hashMap.set('The Spectre', 'Alan Walker');
115-
});
116-
117-
it('should rehash after 12 items by default', () => {
118-
expect(hashMap.getLoadFactor()).toBe(11 / 16);
119-
expect(hashMap.buckets.length).toBe(16);
120-
hashMap.set('Alone', 'Alan Walker');
121-
expect(hashMap.getLoadFactor()).toBe(0.75);
122-
123-
hashMap.set('Levels', 'Avicii');
124-
125-
expect(hashMap.getLoadFactor()).toBe(13 / 32);
126-
expect(hashMap.buckets.length).toBe(32);
127-
128-
expect(hashMap.get('Dura')).toBe('Daddy Yankee');
129-
expect(hashMap.get('Bailando')).toBe('Enrique Iglesias');
130-
expect(hashMap.get('Levels')).toBe('Avicii');
99+
describe('#rehash', () => {
100+
let hashMap;
101+
102+
beforeEach(() => {
103+
hashMap = new HashMap();
104+
105+
hashMap.set('Pineapple', 'Pen Pineapple Apple Pen');
106+
hashMap.set('Despacito', 'Luis Fonsi');
107+
hashMap.set('Bailando', 'Enrique Iglesias');
108+
hashMap.set('Dura', 'Daddy Yankee');
109+
hashMap.set('Lean On', 'Major Lazer');
110+
hashMap.set('Hello', 'Adele');
111+
hashMap.set('All About That Bass', 'Meghan Trainor');
112+
hashMap.set('Wake Me Up', 'Avicii');
113+
hashMap.set('Brother', 'Avicii');
114+
hashMap.set('Faded', 'Alan Walker');
115+
hashMap.set('The Spectre', 'Alan Walker');
116+
});
117+
118+
it('should rehash after 12 items by default', () => {
119+
expect(hashMap.getLoadFactor()).toBe(11 / 16);
120+
expect(hashMap.buckets.length).toBe(16);
121+
hashMap.set('Alone', 'Alan Walker');
122+
expect(hashMap.getLoadFactor()).toBe(0.75);
123+
124+
hashMap.set('Levels', 'Avicii');
125+
126+
expect(hashMap.getLoadFactor()).toBe(13 / 32);
127+
expect(hashMap.buckets.length).toBe(32);
128+
129+
expect(hashMap.get('Dura')).toBe('Daddy Yankee');
130+
expect(hashMap.get('Bailando')).toBe('Enrique Iglesias');
131+
expect(hashMap.get('Levels')).toBe('Avicii');
132+
});
131133
});
132134
});

‎src/data-structures/hash-maps/hashmap.perf.js

Lines changed: 9 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -1,12 +1,14 @@
1+
// nodemon src/data-structures/hash-maps/hashmap.perf.js
12
const HashTable = require('./hash-map');
23
// const asciichart = require ('asciichart');
34
// const { Series } = require('pandas-js');
5+
const Series = require('./stats');
46

57
const dict = new HashTable();
68

79
const keys = [];
810
const values = [];
9-
for (let i = 0; i < 1e1; i++) {
11+
for (let i = 0; i < 1e5; i++) {
1012
keys.push(makeid());
1113
values.push(parseInt(Math.random()*1e2));
1214
}
@@ -29,18 +31,11 @@ const usage = dict.buckets.reduce((array, el, index) => {
2931
return array;
3032
}, Array(dict.buckets.length).fill(0));
3133

32-
console.log(usage, dict.buckets);
34+
// console.log(usage, dict.buckets);
3335

34-
const ds = new Series(usage);
35-
console.log({
36-
length: usage.length,
37-
max: Math.max.apply(null, usage),
38-
min: Math.min.apply(null, usage),
39-
count: ds.length,
40-
mean: ds.mean(),
41-
std: ds.std(),
42-
variance: ds.variance(),
43-
});
36+
const s = new Series(usage);
37+
console.log('Bucket array distribution', s.describe());
38+
console.log('Bucket array frequencies', s.frequencies);
4439

4540
// console.log({elementsByBucket});
4641
// console.log (asciichart.plot(elementsByBucket));
@@ -49,9 +44,9 @@ console.log({
4944

5045
function makeid() {
5146
const text = [];
52-
const possible = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789ñåé😁🖖1234567890!@#$%^&*()_+{}';<>?";
47+
const possible = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789ñåé😁🖖1234567890!@#$%^&*()_+{}';<>? ";
5348

54-
for (let i = 0; i < 5; i++) {
49+
for (let i = 0; i < parseInt((Math.random() * 20) + 1, 10); i += 1) {
5550
text.push(possible.charAt(Math.floor(Math.random() * possible.length)));
5651
}
5752

0 commit comments

Comments
 (0)