#cardinality #approximation #intersection #union #operation #port #constant-memory

bin+lib hyperminhash

Fast, constant-memory cardinality approximation

4 releases

0.1.3 Feb 28, 2024
0.1.2 Feb 25, 2024
0.1.1 May 5, 2019
0.1.0 May 5, 2019

#6 in #intersection

Download history 244/week @ 2024-02-19 320/week @ 2024-02-26 27/week @ 2024-03-04 11/week @ 2024-03-11 66/week @ 2024-04-01

79 downloads per month

MIT license

30KB
226 lines

Hyperminhash for Rust

Crates.io Version Docs

A straight port of Hyperminhash for Rust. Very fast, constant memory-footprint cardinality approximation, including intersection and union operation.

use std::{io, io::Bufread, fs};

let reader = io::BufReader::new(fs::File::open(fname)?).lines();
let sketch = reader.collect::<io::Result<hyperminhash::Sketch>>()?;
println!("{}", sketch.cardinality());

Two files of 10,000,000 random strings each:

Operation Runtime Result
Cardinality via sort strings1.txt | uniq | wc -l 13.57 secs 9,774,970
Union via cat strings1.txt string2.txt | sort | uniq | wc -l 84.4 secs 19,122,087
Intersection via comm -12 <(sort string1.txt) <(sort strings2.txt) | wc -l 25.3 secs 428,370
Cardinality via Hyperminhash 0.69 secs 9,861,113
Cardinality via Hyperminhash (multithreaded) 0.15 secs 9,971,928
Union via Hyperminhash 1.59 secs 19,042,941
Intersection via Hyperminhash 1.52 secs 430,977

Dependencies

~125KB