Build the NNDescent index (async version). Use this on WASM with GPU where `pollster::block_on` cannot block. On native, prefer [`build`] which wraps

Call a registered function on the worker.

Check if the given rect is occupied, return true if any bit is true

Check if visited and mark as visited in one operation. Returns true if ALREADY visited (before marking).

Push a value into a max-heap, checking for duplicates first. Returns 1 if inserted, 0 if rejected or duplicate.

Returns the alpha value of a circle's pixel at (px, py). * The circle is centered at the center of pixel (0, 0) with the given radius. * The alpha v

UMAP / NNDescent benchmark suite.

Collect the edges that are due to be sampled at `current_epoch`, advancing the per-edge sampling schedule counters in place. Returns `(edge_index, n_

Compute distribution stats for all fields used in a chart.

Compute distances without threshold filtering — all distances are returned. Used for `init_rp_tree` where the heap is initially empty.

Compute global in-degree for all points: how many times each point appears as a neighbor of other points in the KNN graph.

Batch projection: compute margins for multiple nodes in a single GPU dispatch. Each entry in the batch is (hyperplane, offset, point_indices). Return

Dispatch a single chunk of nodes that fits within the pre-allocated hyperplane/offset buffers (at most `max_batch_nodes` nodes).

Compute recall: fraction of true neighbors found.

* A tiny async sequencer that serializes access to a resource which can only be * touched while no async section holds it. * * `step`/`run` are asy

The number of epochs run so far.

cyrb53 (c) 2018 bryc (github.com/bryc) * License: Public domain (or MIT if needed). Attribution appreciated. * * A fast and simple 53-bit string ha

Dispatch a single chunk of pairs (must fit within buffer limits).

Diversify the neighbor graph using relative neighborhood pruning. Removes edges where a shorter path through a retained neighbor exists.

Download the embedding from the GPU into a fresh Vec.

Create an Embedding Atlas widget in Streamlit. Args: data_frame: The data frame to visualize. x: Th

Euclidean hub-based split: uses top-3 highest-degree points to generate hyperplanes between all pairs, picks the most balanced split.

Euclidean random projection split: picks 2 random points, computes their difference as hyperplane and midpoint offset.

Compute exact k-NN using sklearn, including self-matches.

Fetch MNIST (70k images of 28x28 pixels).