Lib.rs

Slighly over-engineered Fenwick Tree implmentation.

Allows efficient prefix sum calculation.

Created for training purposes to test:

1. rust typesystem, default trait implmentation, enums as a way for polymorphism
2. memory management and consumption of value
3. cargo tools, docs, tests, clippy and benchmarks, build and publish.

Code is free to do whatever you feel like.

Provides abstraction for Fenwick tree data structure and 2 implmentations:

• prelude::FixedSizeFenwickTree
• prelude::GrowingFenwickTree

Key space for a tree lies within usize range. Tree support any value that implements FenwickTreeValue trait. FenwickTreeValue is automatically implmented for all primitive numeric types that support std::ops::AddAssign, std::ops::Sub, core::cmp::PartialEq and Copy traits.

Installation

cargo add fenwick-bit-tree

Test

cargo test

Benchmarks

cargo bench --features benchmarks

Basic usage:

use fenwick_bit_tree::prelude::*;

// Create the tree with capacity for 32 aggregated [`i32`] data points.
// One can use whole usize range to store datapoints for unicode timestamps
let mut tree = FixedSizeFenwickTree::<i32>::new(32);

// Add values

tree.update(0, 1);
tree.update(0, 4); // Will aggregate value at index 0 so it would be 5
tree.update(10, 10);
tree.update(20, 10);
tree.update(30, 10);

// Now you can query data.
// NOTE: FixedSizeFenwickTree will raise error when query goes out of bounds.
//       GrowingFenwickTree will automatically truncate the range to the rightmost index.

assert_eq!(tree.query(4).unwrap(), 5);
assert_eq!(tree.query(15).unwrap(), 15);
assert_eq!(tree.query(31).unwrap(), 35);

// Also allows making range queries

let val = tree.range_query(2, 16).unwrap(); // Will return aggregated sum of all values between those keys.
assert_eq!(val, 10);

Current version: 2.0.2

License: MIT OR Apache-2.0