FastPriorityQueue.js : a fast, heap-based priority queue in JavaScript

In a priority queue, you can...

query or remove (poll) the smallest element quickly
insert elements quickly

In practice, "quickly" often means in logarithmic time (O(log n)).

A heap can be used to implement a priority queue.

FastPriorityQueue is an attempt to implement a performance-oriented priority queue in JavaScript. It can be several times faster than other similar libraries. It is ideal when performance matters.

License: Apache License 2.0

Usage

var x = new FastPriorityQueue();
x.add(1);
x.add(0);
x.add(5);
x.add(4);
x.add(3);
x.peek(); // should return 0, leaves x unchanged
x.size; // should return 5, leaves x unchanged
while (!x.isEmpty()) {
  console.log(x.poll());
} // will print 0 1 3 4 5
x.trim(); // (optional) optimizes memory usage

You can also provide the constructor with a comparator function.

var x = new FastPriorityQueue(function(a, b) {
  return a > b;
});
x.add(1);
x.add(0);
x.add(5);
x.add(4);
x.add(3);
while (!x.isEmpty()) {
  console.log(x.poll());
} // will print 5 4 3 1 0

If you are using node.js, you need to import the module:

var FastPriorityQueue = require('fastpriorityqueue');
var b = new FastPriorityQueue(); // initially empty
b.add(1); // add the value "1"

Instance methods summary:

add(value): add an element into the queue; runs in O(log n) time.
poll(): remove and return the element on top of the heap (smallest element); runs in O(log n) time. If the priority queue is empty, the function returns undefined.
remove(value): remove an element matching the provided value, if found, from the queue. The item is matched by using the queue's comparator. Returns true if the element is removed, false otherwise.
removeOne(callback): execute the callback function for each item of the queue and remove the first item for which the callback will return true. Returns the removed item, or undefined if nothing is removed. The callback must be a pure function.
removeMany(callback[, limit]): execute the callback function for each item of the queue and remove each item for which the callback will return true, up to a max limit of removed items if specified or no limit if unspecified. Returns an array containing the removed items. The callback must be a pure function.
replaceTop(value): poll() and add(value) in one operation. This is useful for fast, top-k queries. Returns the removed element or undefined, similar to poll().
heapify(array): replace the content of the heap with the provided array, then order it based on the comparator.
peek(): return the top of the queue (smallest element) without removal, or undefined if the queue is empty; runs in O(1) time.
isEmpty(): return true if the the queue has no elements, false otherwise.
clone(): copy the priority queue into another, and return it. Queue items are shallow-copied. Runs in O(n) time.
forEach(callback): iterate over all items in the priority queue from smallest to largest. callback should be a function that accepts two arguments, value (the item), and index, the zero-based index of the item.
trim(): clean up unused memory in the heap; useful after high-churn operations like many add()s then remove()s.

npm install

  $ npm install fastpriorityqueue

Computational complexity

The function calls "add" and "poll" have logarithmic complexity with respect to the size of the data structure (attribute size). Looking at the top value is a constant time operation.

Testing

Using node.js (npm), you can test the code as follows...

  $ npm install mocha
  $ npm test

Is it faster?

It tends to fare well against the competition. In some tests, it can be five times faster than any other JavaScript implementation we could find.

$ node benchmark/test.js
Platform: darwin 20.1.0 x64
Intel(R) Core(TM) i9-9980HK CPU @ 2.40GHz
Node version 14.7.0, v8 version 8.4.371.19-node.12

Comparing against:
js-priority-queue: https://github.com/adamhooper/js-priority-queue 0.1.5
stablepriorityqueue: https://github.com/lemire/StablePriorityQueue.js 0.1.2
heap.js: https://github.com/qiao/heap.js 0.2.6
binaryheapx: https://github.com/xudafeng/BinaryHeap 0.1.1
priority_queue: https://github.com/agnat/js_priority_queue 0.1.3
js-heap: https://github.com/thauburger/js-heap 0.3.1
queue-priority: https://github.com/augustohp/Priority-Queue-NodeJS 1.0.0
priorityqueuejs: https://github.com/janogonzalez/priorityqueuejs 2.0.0
qheap: https://github.com/andrasq/node-qheap 1.4.0
yabh: https://github.com/jmdobry/yabh 1.2.0

starting dynamic queue/enqueue benchmark
FastPriorityQueue x 36,816 ops/sec ±0.74% (92 runs sampled)
FastPriorityQueue---replaceTop x 107,942 ops/sec ±0.71% (91 runs sampled)
sort x 6,240 ops/sec ±1.65% (92 runs sampled)
StablePriorityQueue x 10,333 ops/sec ±4.09% (91 runs sampled)
js-priority-queue x 14,435 ops/sec ±1.97% (91 runs sampled)
heap.js x 6,568 ops/sec ±2.29% (90 runs sampled)
binaryheapx x 8,595 ops/sec ±0.56% (94 runs sampled)
priority_queue x 8,201 ops/sec ±0.74% (94 runs sampled)
js-heap x 557 ops/sec ±1.70% (89 runs sampled)
queue-priority x 291 ops/sec ±2.46% (88 runs sampled)
priorityqueuejs x 13,864 ops/sec ±2.02% (90 runs sampled)
qheap x 26,882 ops/sec ±1.81% (93 runs sampled)
yabh x 10,472 ops/sec ±1.50% (93 runs sampled)
Fastest is FastPriorityQueue

Benchmarks on an Apple M1:

Platform: darwin 20.2.0 arm64
Apple M1
Node version 15.6.0, v8 version 8.6.395.17-node.23

Comparing against:
js-priority-queue: https://github.com/adamhooper/js-priority-queue 0.1.5
stablepriorityqueue: https://github.com/lemire/StablePriorityQueue.js 0.1.2
heap.js: https://github.com/qiao/heap.js 0.2.6
binaryheapx: https://github.com/xudafeng/BinaryHeap 0.1.1
priority_queue: https://github.com/agnat/js_priority_queue 0.1.3
js-heap: https://github.com/thauburger/js-heap 0.3.1
queue-priority: https://github.com/augustohp/Priority-Queue-NodeJS 1.0.0
priorityqueuejs: https://github.com/janogonzalez/priorityqueuejs 2.0.0
qheap: https://github.com/andrasq/node-qheap 1.4.0
yabh: https://github.com/jmdobry/yabh 1.2.0

starting dynamic queue/enqueue benchmark
FastPriorityQueue x 47,894 ops/sec ±0.19% (100 runs sampled)
FastPriorityQueue---replaceTop x 187,809 ops/sec ±0.09% (97 runs sampled)
sort x 9,285 ops/sec ±0.10% (100 runs sampled)
StablePriorityQueue x 19,830 ops/sec ±0.49% (97 runs sampled)
js-priority-queue x 28,382 ops/sec ±0.10% (98 runs sampled)
heap.js x 5,504 ops/sec ±0.22% (100 runs sampled)
binaryheapx x 10,473 ops/sec ±0.11% (98 runs sampled)
priority_queue x 9,041 ops/sec ±0.33% (97 runs sampled)
js-heap x 390 ops/sec ±0.04% (96 runs sampled)
queue-priority x 438 ops/sec ±0.09% (95 runs sampled)
priorityqueuejs x 14,797 ops/sec ±0.07% (101 runs sampled)
qheap x 38,108 ops/sec ±0.12% (99 runs sampled)
yabh x 14,942 ops/sec ±0.24% (99 runs sampled)

Note that qheap has been updated following the introduction of FastPriorityQueue, with a reference to FastPriorityQueue which might explains the fact that its performance is comparable to FastPriorityQueue.

Insertion order

A binary heap does not keep track of the insertion order.