I developed some RcppParallel code to calculate large distance matrices
quickly.
I later discovered https://github.com/hypertidy/geodist and thought
that a more aligned parallel version of that package could possibly have
some use.
Also see: hypertidy/geodist#16
# install.packages("devtools")
devtools::install_github("mkuehn10/geodistpar")library(geodistpar)
library(geodist)
library(microbenchmark)
library(ggplot2)
set.seed(10)
n <- 1000
x <- cbind (-10 + 20 * stats::runif (n), -10 + 20 * stats::runif (n))
y <- cbind (-10 + 20 * stats::runif (2 * n), -10 + 20 * stats::runif (2 * n))
colnames (x) <- colnames (y) <- c ("x", "y")
a <- geodist::geodist(x, y, measure = "haversine")
b <- geodistpar::geodistpar(x, y, measure = "haversine")
head(a[, 1:5])
#> [,1] [,2] [,3] [,4] [,5]
#> [1,] 1225210.6 1788146.1 243415.8 941220.4 1184245.2
#> [2,] 1421649.3 1443725.5 693979.4 1436373.5 1146991.8
#> [3,] 990672.7 670982.9 1244972.4 1948938.0 381055.2
#> [4,] 459949.5 1496824.5 618702.2 1158123.6 549935.1
#> [5,] 1760882.6 1120247.4 1259339.6 2018837.7 1300613.0
#> [6,] 1399721.5 581798.6 1311268.4 2083635.0 828119.6
head(b[, 1:5])
#> [,1] [,2] [,3] [,4] [,5]
#> [1,] 1225210.6 1788146.1 243415.8 941220.4 1184245.2
#> [2,] 1421649.3 1443725.5 693979.4 1436373.5 1146991.8
#> [3,] 990672.7 670982.9 1244972.4 1948938.0 381055.2
#> [4,] 459949.5 1496824.5 618702.2 1158123.6 549935.1
#> [5,] 1760882.6 1120247.4 1259339.6 2018837.7 1300613.0
#> [6,] 1399721.5 581798.6 1311268.4 2083635.0 828119.6
identical(a, b)
#> [1] TRUE
mb1 <- microbenchmark::microbenchmark(
geodist = geodist::geodist(x, y, measure = "haversine"),
geodistpar = geodistpar::geodistpar(x, y, measure = "haversine"),
times = 25
)
mb1
#> Unit: milliseconds
#> expr min lq mean median uq max neval
#> geodist 325.3630 334.3536 341.89431 337.8233 343.5093 387.3065 25
#> geodistpar 63.8012 74.3011 88.97711 86.6924 95.6527 141.7862 25
#> cld
#> b
#> a
autoplot(mb1)
#> Coordinate system already present. Adding new coordinate system, which will replace the existing one.
a <- geodist::geodist(x, y, measure = "vincenty")
b <- geodistpar::geodistpar(x, y, measure = "vincenty")
head(a[, 1:5])
#> [,1] [,2] [,3] [,4] [,5]
#> [1,] 1225210.6 1788146.1 243415.8 941220.4 1184245.2
#> [2,] 1421649.3 1443725.5 693979.4 1436373.5 1146991.8
#> [3,] 990672.7 670982.9 1244972.4 1948938.0 381055.2
#> [4,] 459949.5 1496824.5 618702.2 1158123.6 549935.1
#> [5,] 1760882.6 1120247.4 1259339.6 2018837.7 1300613.0
#> [6,] 1399721.5 581798.6 1311268.4 2083635.0 828119.6
head(b[, 1:5])
#> [,1] [,2] [,3] [,4] [,5]
#> [1,] 1225210.6 1788146.1 243415.8 941220.4 1184245.2
#> [2,] 1421649.3 1443725.5 693979.4 1436373.5 1146991.8
#> [3,] 990672.7 670982.9 1244972.4 1948938.0 381055.2
#> [4,] 459949.5 1496824.5 618702.2 1158123.6 549935.1
#> [5,] 1760882.6 1120247.4 1259339.6 2018837.7 1300613.0
#> [6,] 1399721.5 581798.6 1311268.4 2083635.0 828119.6
identical(a, b)
#> [1] TRUE
mb2 <- microbenchmark::microbenchmark(
geodist = geodist::geodist(x, y, measure = "vincenty"),
geodistpar = geodistpar::geodistpar(x, y, measure = "vincenty"),
times = 25
)
mb2
#> Unit: milliseconds
#> expr min lq mean median uq max neval
#> geodist 410.6076 418.1190 429.3395 424.9092 429.1759 552.5144 25
#> geodistpar 95.5457 97.0618 111.3339 104.8388 111.7687 234.4770 25
#> cld
#> b
#> a
autoplot(mb2)
#> Coordinate system already present. Adding new coordinate system, which will replace the existing one.