dogsort v0.1.0

Dogsort 🐶

Status: Work in Progress

Dogsort is an in-place sorting algorithm (additional required memory is O(1)) with an average complexity of O(N*logN) (the worst complexity is currently under investigation but is guaranteed to be not greater than O(N^2)). It combines Quicksort and Average Sort (https://www.cscjournals.org/manuscript/Journals/IJEA/Volume2/Issue2/IJEA-16.pdf).

⚠️ This project has Work in Progress status. It passes all tests the author has written but it is not guaranteed to pass all edge cases. Please use with caution.

Comparison with other algorithms

Dogsort is more than twice faster than V8 implementation of Array.sort in average for arrays that have size more than 1024 elements (see Roadmap). It also beats Timsort on random arrays but still dramatically loses on sorted arrays just yet, which is expected (see Roadmap). Here are some crude results on arrays of 10^6 size:

Install

npm install dogsort --save

Include

NodeJS

const dogsort = require('dogsort');

Browser

<script src='node_modules/dogsort/build/dogsort.js' type='text/javascript'></script>

...or use ES6 import or EMD

Use

const array = [2, 8.88, -7, 90, 0, 120000, -3.455555]

dogsort(array)
console.log(array)

Roadmap

• Improve performance for small arrays
• Improve performance for sorted arrays
• Add a possibility for a custom comparison function
• Change recursion to stack to save memory if performance is not sacrificed