acacia 0.1.2

acacia is a spatial tree library written in Rust. It is generic over the dimension of the partitioned space and thus supports binary trees, quadtrees, octrees, etc. The intended goal is to implement these features as fast and covering as many use cases as possible without sacrificing abstraction.

The current state of the project is very experimental. It works and has ample test coverage, but both the API and the internals will likely change in the future to improve interface and performance.

Features

• Tree construction from a simple iterator.
• Associate data to a tree during construction using closures.
• Perform arbitrary computational queries on the trees.
• Faster dimension-specific trees, e.g. an octree (not yet implemented)

Documentation

The documentation is hosted on Rust CI and should be up-to-date with the master branch.

Example: N-body gravity calculation

Gravity calculations are a fairly common example for speeding up calculations with spatial trees. This is a simple example to calculate the gravitational acceleration at a given point between a set of gravitating particles. The code presented here is an excerpt from a complete example you can find in the directory example/gravity.

The tree can be constructed from an iterator over particles and some data about its geometry.

let tree = Tree::new(
    particles.iter(),
    Ncube::new(origin, 11.0),

Note, that the particles implement the Position trait used to define the notion that a type has a position.

Next we need a couple of values and closures to associate data with each node in the tree: a value for empty nodes, a closure that assigns a value to a leaf node given the object stored in it, and a closure that combines two pieces of associated data to compute values for branch nodes.

    (origin, 0.0),
    &|obj| (obj.position, obj.mass),
    &|&(com1, m1), &(com2, m2)|
        if m1 + m2 > 0.0 {(
            origin + (com1.to_vec() * m1 + com2.to_vec() * m2) / (m1 + m2),
            m1 + m2,
        )}
        else {
            (origin, 0.0)
        }
);

The associated data in this example is a tuple made from the center of mass and the total mass of a node.

Now we can issue a computational query to the tree by passing in two more closures to its query_data method: the first one serves as a criterion for recursion. If a branch node passes this, the query continues on its children. The second one collects force terms from each piece of associated data the tree encounters during this recursion.

let mut tree_gravity: Vec3<f64> = zero();
tree.query_data(
    &|node| {
        let &(ref center_of_mass, _) = node.data();
        let d = test_point.dist(center_of_mass);
        let delta: f64 = node.partition().center().dist(center_of_mass);
        d < 2.0 * node.partition().width() + delta
    },

    &mut |&(center_of_mass, mass)| {
        tree_gravity = tree_gravity + newton(mass, center_of_mass, test_point);
    },
);

License

This Source Code Form is subject to the terms of the Mozilla Public License, v. 2.0. If a copy of the MPL was not distributed with this file, You can obtain one at http://mozilla.org/MPL/2.0/.