问题描述
一些背景:
我必须经常使用 HMR 包中的函数 HMR.不幸的是，这个功能很慢.(HMR本质上是一个拟合函数，设计得尽可能的健壮，这也是效率不高的原因之一.)HMR函数调用函数HMR::.HMR.fit1，进行实际拟合.使用 Rprof 我知道关于效率的主要问题是使用 lsfit，它被调用了很多.因此，我修改了.HMR.fit1的代码，直接调用lsfit使用的C函数，没有lsfit的所有开销，应该导致显着的速度增益.
I have to use function HMR from package HMR a lot. Unfortunately, this function is very slow. (HMR is essentially a fitting function, which is designed to be as robust as possible, which is one reason for the lack of efficiency.) Function HMR calls function HMR::.HMR.fit1, which does the actual fitting.Using Rprof I know that the main problem regarding efficiency is the use of lsfit, which gets called a lot. Therefore, I modified the code of .HMR.fit1 to call the C function used by lsfit directly without all the overhead of lsfit, which should result in a substantial speed gain.
现在我想用我修改的函数替换 HMR::.HMR.fit1 并测试 HMR 是否给出相同的结果以及我获得了多少速度.
Now I would like to substitute HMR::.HMR.fit1 with my modified function and test HMR if it gives the same results and how much speed I gain.
我尝试过这样做:
mod.fun <- function(<many args>) {
<a lot of code>
}
environment(mod.fun) <- environment(.HMR.fit1)
.HMR.fit1 <- mod.fun
但是，HMR::.HMR.fit1 并没有因此而改变，显然 HMR::HMR 没有使用我修改后的拟合函数.有没有办法在不从源代码构建包的情况下实现我想要的，由于我的(Windows)计算机上的用户权限限制，我无法做到这一点?
However, HMR::.HMR.fit1 is not changed by doing this and apparently HMR::HMR does not use my modified fitting function. Is there a way to achieve what I want without building the package from source, which I cannot do due to user rights restrictions on my (windows) computer?
目前我的解决方案是复制HMR::HMR的代码，希望有更方便的解决方案.
Right now, my solution would be to copy the code of HMR::HMR, but I hope there is a more convenient solution.
推荐答案
试试
?assignInNamespace
将 HMR 包中的 .HMR.fit1 替换为您的版本.
to replace .HMR.fit1 in the HMR package with your version.
可能重复:
这篇关于更改包的内部功能的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持！