2020年CUG大气科学系
R、Python、Julia、Fortran、Rcpp混合编程模板
选择任意一个你擅长的语言，在该repository的基础上，编写滑动平均函数，测试通过后github推送给我。
主要步骤如下：
- 登陆github账户，fork此repository
- git clone 克隆的仓库链接
- 编写函数，编写R语言接口
- devtools::load_all()配合browser测试
- git commit提交到github，commit message中写明学号和姓名
- Pull request提交给我
如何修改remote-url:
# change remote url
git remote set-url origin https://github.com/your_username/movmean
git pushdevtools::install_github("kongdd/movmean")library(movmean)
x = c(NA, 1, 2, 3, 4, 5, Inf)
# answer = c(1, 1.5, 2, 3, 4, 4.5, 5)
x = c(rnorm(1000))
r_jl <- movmean_julia(x, halfwin = 5)
#> Julia version 1.5.3 at location C:\PROGRA~1\Julia\JULIA1~1.3\bin will be used.
#> Loading setup script for JuliaCall...
#> Finish loading setup script for JuliaCall.
# 欲用此版本需安装Julia，Julia版本暂时不能考虑NA, Inf
# r_jl <- movmean_julia(x, halfwin = 1)
microbenchmark::microbenchmark(
r_jl <- movmean_julia(x, halfwin = 5),
r_cpp <- movmean_rcpp(x, halfwin = 5),
r_f90 <- movmean_f90(x, halfwin = 5L),
r_r <- movmean_r(x, halfwin = 5, na.rm = FALSE), # the version of wyxz
r_py <- movmean:::movmean_py(x, halfwin = 5L),
r_np <- movmean_np(x, halfwin = 5L)
)
#> Unit: microseconds
#> expr min lq mean
#> r_jl <- movmean_julia(x, halfwin = 5) 81.8 129.05 175.402
#> r_cpp <- movmean_rcpp(x, halfwin = 5) 80.1 94.10 100.265
#> r_f90 <- movmean_f90(x, halfwin = 5L) 21.5 36.70 51.808
#> r_r <- movmean_r(x, halfwin = 5, na.rm = FALSE) 4553.6 4646.65 7528.842
#> r_py <- movmean:::movmean_py(x, halfwin = 5L) 1350.7 1445.00 1542.813
#> r_np <- movmean_np(x, halfwin = 5L) 10074.3 10578.60 11028.487
#> median uq max neval
#> 191.00 210.70 309.9 100
#> 100.35 105.45 238.5 100
#> 50.95 55.70 641.4 100
#> 4715.00 4804.05 210040.2 100
#> 1530.35 1611.70 2164.7 100
#> 11047.55 11531.85 11913.7 100Keep in mind that this repository is released under a GPL3 license, which permits commercial use but requires that the source code (of derivatives) is always open even if hosted as a web service.