| Title: | Fast Computation of Pairwise Hamming Distances |
| Type: | Package |
| Version: | 1.2 |
| Depends: | R (≥ 4.0.0) |
| Description: | Pairwise Hamming distances are computed between the rows of a binary (0/1) matrix using highly optimized 'C' code. The input is an integer matrix where each row represents a binary feature vector and returns a symmetric integer matrix of pairwise distances. Internally, rows are bit-packed into 64-bit words for fast XOR-based comparisons, with hardware-accelerated popcount operations to count differences. 'OpenMP' parallelization ensures efficient performance for large matrices. |
| License: | GPL-3 |
| Encoding: | UTF-8 |
| RoxygenNote: | 7.3.2 |
| SystemRequirements: | C compiler (C99), OpenMP |
| NeedsCompilation: | yes |
| Packaged: | 2025-04-26 21:44:01 UTC; raviv |
| Author: | Ravi Varadhan [aut, cre] |
| Maintainer: | Ravi Varadhan <ravi.varadhan@jhu.edu> |
| Repository: | CRAN |
| Date/Publication: | 2025-04-27 02:00:02 UTC |
Pairwise Hamming distances
Description
Computes the pairwise Hamming distances between rows of a binary matrix.
Usage
hamming_distance(X, nthreads = NULL)
Arguments
X |
A binary (0/1) numeric matrix. |
nthreads |
Integer; number of OpenMP threads to use. If |
Value
An integer matrix of pairwise Hamming distances.
Examples
n <- 10000
m <- 1000
set.seed(2468)
X <- matrix(sample(0:1, n * m, replace = TRUE), nrow = n)
# Use all available threads
system.time(result <- hamming_distance(X))
# limit to 2 threads
system.time(hamming_distance(X, nthreads = 2))