Crate roughly 

roughly

Probabilistic data structures for Rust — because sometimes close enough is good enough.

This crate provides space-efficient, approximate data structures for common tasks:

• BloomFilter — membership testing with configurable false positive rate
• HyperLogLog — cardinality estimation (count distinct elements)
• CountMinSketch — frequency estimation (how often items appear)

Quick Start

use roughly::prelude::*;

// Bloom filter: "Have I seen this before?"
let mut bloom = BloomFilter::builder()
    .expected_items(10_000)
    .false_positive_rate(0.01)
    .build();
bloom.insert(&"hello");
assert!(bloom.contains(&"hello"));

// HyperLogLog: "How many unique items?"
let mut hll = HyperLogLog::builder().std_error(0.01).build();
for i in 0..100_000u64 {
    hll.insert(&i);
}
println!("Estimated unique count: {}", hll.count());

// Count-Min Sketch: "How often did this appear?"
let mut cms = CountMinSketch::builder().build();
cms.insert_many(&"popular", 1000);
println!("Estimated frequency: {}", cms.estimate(&"popular"));

Re-exports

pub use bloom::BloomFilter;

pub use countmin::CountMinSketch;

pub use hyperloglog::HyperLogLog;

Modules

bloom

Bloom filter implementation for probabilistic set membership testing.

countmin

Count-Min Sketch implementation for frequency estimation.

hash

Hashing utilities for probabilistic data structures.

hyperloglog

HyperLogLog implementation for cardinality estimation.

prelude

Convenient re-exports of the main types and traits.

traits

Traits defining common interfaces for probabilistic data structures.