Struct HyperBall

pub struct HyperBall<'a, G1: RandomAccessGraph + Sync, G2: RandomAccessGraph + Sync, D: Succ<Input = usize, Output = usize>, L: MergeEstimationLogic<Item = G1::Label> + Sync, A: EstimatorArrayMut<L>> { /* private fields */ }

An algorithm that computes an approximation of the neighborhood function, of the size of the reachable sets, and of (discounted) positive geometric centralities of a graph.

Implementations

impl<G1: RandomAccessGraph + Sync, G2: RandomAccessGraph + Sync, D: Succ<Input = usize, Output = usize> + Sync, L: MergeEstimationLogic<Item = usize> + Sync, A: EstimatorArrayMut<L> + Sync + AsSyncArray<L>> HyperBall<'_, G1, G2, D, L, A>

where L::Backend: PartialEq,

pub fn run( &mut self, upper_bound: usize, threshold: Option<f64>, thread_pool: &ThreadPool, rng: impl Rng, pl: &mut impl ConcurrentProgressLog, ) -> Result<()>

Runs HyperBall.

Arguments

• upper_bound: an upper bound to the number of iterations.

• threshold: a value that will be used to stop the computation by relative increment if the neighborhood function is being computed. If None the computation will stop when no estimators are modified.

• thread_pool: The thread pool to use for parallel computation.

• pl: A progress logger.

pub fn run_until_stable( &mut self, upper_bound: usize, thread_pool: &ThreadPool, rng: impl Rng, pl: &mut impl ConcurrentProgressLog, ) -> Result<()>

Runs HyperBall until no estimators are modified.

Arguments

• upper_bound: an upper bound to the number of iterations.

• thread_pool: The thread pool to use for parallel computation.

• pl: A progress logger.

pub fn run_until_done( &mut self, thread_pool: &ThreadPool, rng: impl Rng, pl: &mut impl ConcurrentProgressLog, ) -> Result<()>

Runs HyperBall until no estimators are modified with no upper bound on the number of iterations.

Arguments

• thread_pool: The thread pool to use for parallel computation.

• pl: A progress logger.

pub fn neighborhood_function(&self) -> Result<&[f64]>

Returns the neighborhood function computed by this instance.

pub fn sum_of_distances(&self) -> Result<&[f32]>

Returns the sum of distances computed by this instance if requested.

pub fn harmonic_centralities(&self) -> Result<&[f32]>

Returns the harmonic centralities (sum of inverse distances) computed by this instance if requested.

pub fn discounted_centrality(&self, index: usize) -> Result<&[f32]>

Returns the discounted centralities of the specified index computed by this instance.

Arguments

• index: the index of the requested discounted centrality.

pub fn closeness_centrality(&self) -> Result<Box<[f32]>>

Computes and returns the closeness centralities from the sum of distances computed by this instance.

pub fn lin_centrality(&self) -> Result<Box<[f32]>>

Computes and returns the lin centralities from the sum of distances computed by this instance.

Note that lin’s index for isolated nodes is by (our) definition one (it’s smaller than any other node).

pub fn nieminen_centrality(&self) -> Result<Box<[f32]>>

Computes and returns the Nieminen centralities from the sum of distances computed by this instance.

pub fn reachable_nodes_from(&self, node: usize) -> Result<f64>

Reads from the internal estimator array and estimates the number of nodes reachable from the specified node.

Arguments

• node: the index of the node to compute reachable nodes from.

pub fn reachable_nodes(&self) -> Result<Box<[f32]>>

Reads from the internal estimator array and estimates the number of nodes reachable from every node of the graph.

hyperball.reachable_nodes().unwrap()[i] is equal to hyperball.reachable_nodes_from(i).unwrap().