find out the medoid, the vertex in the dataset that is closest to the centroid

result_set_indices and truth_set_indices correspond by row index. the columns in each row contain the indices of the nearest neighbors, with

* @brief Checks if the given offset is valid. * * This function checks if the given offset is within the bounds of the IO object. * If

* Delete all temporary artifacts. * In the process of creating the stitched index, some temporary artifacts are * created: * 1. the separate bin fi

Create disk layout Sector #1: disk_layout_meta Sector #n: num_nodes_per_sector nodes Each node's layout: {full precision vector:[T; DIM]}{num_nbrs: u3

Get the path for a binary or exit

fill buffer with below and return a wrappered StreamReader object: ['1' '1' ... repeats 1024 times] ['2' '2' ... repeats 1024 times] ['3' '3' ... repe

Test search index

Test search index

Test search index

Test search index

Generate random float32 vectors for testing

get_bin_metadata functions START

Gets a command line for clang-tidy.

Returns a vector of unique neighbors for the given vertex, along with their distances. # Arguments `neighbors` - A vector of neighbor id index for t

GreedySearch against query node Returns visited nodes # Arguments `query` - query vertex `scratch` - in-memory query scratch TODO: use_filter, filter_

* We need to check for popcount support at runtime on Windows with __cpuid * See https://msdn.microsoft.com/en-us/library/bb385231.aspx */

Initialize graph for point # Arguments `query` - query vertex `init_ids` - initial nodes from which search starts `scratch` - in-memory query scratch

Initialize HNSW index

Initialize HGraph index

Initialize Bruteforce index

Initialize IVF index

* Checks the first 16 bits of the memory to see if the element is marked deleted. */

Select pivots in k-means++ algorithm Points that are farther away from the already chosen centroids have a higher probability of being selected as the