如何以tidyverse风格为deSolve包组织一系列模拟（场景）？

huangapple
go评论123阅读模式
英文:

How to organize a series of simulations (scenarios) for the deSolve package in tidyverse style?

问题

以下是翻译好的部分：

在先前的帖子中，原帖作者提出了一个问题，询问如何使用deSolve包来组织一系列具有不同状态和参数的模拟场景。已经提供了一个解决方案，但我想知道是否可以以更适合管道友好风格的方式来实现，但不使用嵌套数据框。目标是将其用作使用清晰的表格数据结构制作最终ggplot的教学示例。

方法1：经典的“基本-R”风格

到目前为止，我通常使用apply函数的列表方法。然后，可以将生成的数据框列表适应于deSolve::plot方法：

  1. library("deSolve")
  3. model <- function(t, y, p) {
  4.   with(as.list(c(y, p)), {
  5.     dN <- r * N * (1 - N/K)
  6.     list(c(dN))
  7.   })
  8. }
  10. times <- 0:10
  11. y0    <- c(N = 0.5)         # 状态变量
  12. p     <- c(r = 0.2, K = 1)  # 模型参数
  14. ## 运行单个模拟
  15. out <- ode(y0, times, model, p)
  16. plot(out)
  18. ## 创建包含一些状态和参数组合的数据框
  19. scenarios <- expand.grid(N = seq(0.5, 1.5, 0.2), K = 1, r = seq(0.2, 1, 0.2))
  21. ## 用于运行数据框的单行模拟的函数
  22. ## 注意 scenarios 和 scenario 之间的差异（复数/单数）
  23. simulate <- function(scenario) {
  24.   ## 拆分场景设置为初始状态（y0）和参数（p）
  25.   y0 <- scenario["N"]
  26.   p  <- scenario[c("r", "K")]
  27.   ode(y0, times, model, p)
  28. }
  30. ## MARGIN = 1：每行是一个场景
  31. ## simplify = FALSE：函数应返回一个列表
  32. outputs <- apply(scenarios, MARGIN = 1, FUN = simulate, simplify = FALSE)
  34. ## plot.deSolve方法可与列表作为第二个参数一起使用
  35. plot(out, outputs)

方法2：迈向管道

基于这个示例，我创建了一个名为simulate_inout的函数，以一种ggplot兼容的方式返回单个场景的输入和输出。然后，应该在管道中为所有场景（所有行）调用它。

以下是示例：

  1. ## 保留输入和输出的simulate版本
  2. simulate_inout <- function(scenario) {
  3.   scenario <- unlist(scenario)
  4.   ## 拆分场景设置为初始状态（y0）和参数（p）
  5.   y0 <- scenario["N"]
  6.   p  <- scenario[c("r", "K")]
  7.   
  8.   ## 集成模型
  9.   output <- ode(y0, times, model, p)
  10.   
  11.   ## 复制输入的行
  12.   input <- do.call("rbind", replicate(length(times), 
  13.     scenario, simplify = FALSE))
  14.   
  15.   ## 返回包含输入和输出的数据框
  16.   cbind(input, output)
  17. }
  19. ## 单个场景
  20. simulate_inout(scenarios[1,])
  22. simulate_all <- function(scenarios) {
  23.   ## 遍历所有行
  24.   ret <- NULL
  25.   for (i in 1:nrow(scenarios)) {
  26.     ret <- rbind(ret, simulate_inout(scenarios[i,]))
  27.   }
  28.   data.frame(ret)
  29. }
  31. ## 使用ggplot绘图
  32. library("ggplot2")
  33. scenarios |> simulate_all() |> ggplot(aes(time, N.1)) + 
  34.   geom_path() + facet_grid(r ~ N)

问题

我希望以一致的tidyverse风格简化此代码，并摆脱simulate_all中的for循环以及其他特定技巧，如do.call。

英文:

In a previous post the original poster asked a question on how to organize a series of scenarios with varying states and parameters for simulations with the deSolve package. A solution was given, but I wonder if it can be made easier in a pipeline-friendly style, but without the use of nested data frames. The goal is to use it as a teaching example using clear tabular data structures that can be fit to a final ggplot.

Approach 1: Classical "base-R" style

Until now, I usually use a list approach with the apply-function. The resulting list of data frames can then be fit to the deSolve::plot-method:

  1. library(&quot;deSolve&quot;)
  3. model &lt;- function(t, y, p) {
  4.   with(as.list(c(y, p)), {
  5.     dN &lt;- r * N * (1 - N/K)
  6.     list(c(dN))
  7.   })
  8. }
  10. times &lt;- 0:10
  11. y0    &lt;- c(N = 0.5)         # state variables
  12. p     &lt;- c(r = 0.2, K = 1)  # model parameters
  14. ## run a single simulation
  15. out &lt;- ode(y0, times, model, p)
  16. plot(out)
  18. ## create a data frame with some combinations of states and parameters
  19. scenarios &lt;- expand.grid(N = seq(0.5, 1.5, 0.2), K = 1, r = seq(0.2, 1, 0.2))
  21. ## a function to run a simulation for a single line of the data frame
  22. ## note difference between scenarios and scenario (plural/singular)
  23. simulate &lt;- function(scenario) {
  24.   ## split scenario settings to initial states (y0) and parameters (p)
  25.   y0 &lt;- scenario[&quot;N&quot;]
  26.   p  &lt;- scenario[c(&quot;r&quot;, &quot;K&quot;)]
  27.   ode(y0, times, model, p)
  28. }
  30. ## MARGIN = 1: each row is a scenario
  31. ## simplify = FALSE: function should return a list
  32. outputs &lt;- apply(scenarios, MARGIN = 1, FUN = simulate, simplify = FALSE)
  34. ## the plot.deSolve method works with lists as second argument
  35. plot(out, outputs)

Approach 2: A step towards a pipeline

Based on this example, I created a function simulate_inout that returns both, inputs and outputs in a ggplot-compatible way for a single scenario. This should then be called for all scenarios (all rows) in a pipeline.

The following works:

  1. ## version of simulate that preserves inputs and outputs
  2. simulate_inout &lt;- function(scenario) {
  3.   scenario &lt;- unlist(scenario)
  4.   ## split scenario settings to initial states (y0) and parameters (p)
  5.   y0 &lt;- scenario[&quot;N&quot;]
  6.   p  &lt;- scenario[c(&quot;r&quot;, &quot;K&quot;)]
  7.   
  8.   ## integrate the model
  9.   output &lt;- ode(y0, times, model, p)
  10.   
  11.   ## replicate rows of inputs
  12.   input &lt;- do.call(&quot;rbind&quot;, replicate(length(times), 
  13.     scenario, simplify = FALSE))
  14.   
  15.   ## return a data frame with inputs and outputs
  16.   cbind(input, output)
  17. }
  19. ## a single scenario
  20. simulate_inout(scenarios[1,])
  22. simulate_all &lt;- function(scenarios) {
  23.   ## iterate over all rows
  24.   ret &lt;- NULL
  25.   for (i in 1:nrow(scenarios)) {
  26.     ret &lt;- rbind(ret, simulate_inout(scenarios[i,]))
  27.   }
  28.   data.frame(ret)
  29. }
  31. ## plot with ggplot
  32. library(&quot;ggplot2&quot;)
  33. scenarios |&gt; simulate_all() |&gt; ggplot(aes(time, N.1)) + 
  34.   geom_path() + facet_grid(r ~ N)

Question

I would like to streamline this code in consistent tidyverse style and to get of the for-loop in simulate_all and other specific tricks like do.call.

答案1

得分: 1

You could use purrr::map_dfr to loop over the rows of your scenarios df. Requires some rewriting of your functions such that it takes the parameters itself as arguments. Additionally, I simplified your code a bit.

EDIT Replaced the superseded pmap_dfr by pmap(...) |&gt; list_rbind(). Use ... to pass the arguments to simulate_inout.

  1. library(deSolve)
  2. library(ggplot2)
  3. library(purrr)
  5. model &lt;- function(t, y, p) {
  6.   N &lt;- y
  7.   r &lt;- p[[1]]
  8.   K &lt;- p[[2]]
  9.   
  10.   dN &lt;- r * N * (1 - N / K)
  11.   
  12.   list(dN)
  13. }
  14. times &lt;- 0:10
  16. simulate_inout &lt;- function(...) {
  17.   args &lt;- list(...)
  18.   
  19.   y0 &lt;- args[[&quot;N&quot;]]
  20.   p &lt;- args[c(&quot;r&quot;, &quot;K&quot;)]
  21.   
  22.   output &lt;- ode(y0, times, model, p)
  23.   
  24.   data.frame(args, output)
  25. }
  27. scenarios &lt;- expand.grid(
  28.   N = seq(0.5, 1.5, 0.2),
  29.   K = 1,
  30.   r = seq(0.2, 1, 0.2)
  31. )
  33. scenarios |&gt;
  34.   purrr::pmap(simulate_inout) |&gt;
  35.   list_rbind() |&gt; # or dplyr::bind_rows()
  36.   ggplot(aes(time, X1)) +
  37.   geom_path() +
  38.   facet_grid(r ~ N)

如何以tidyverse风格为deSolve包组织一系列模拟（场景）？<!-- -->

英文:

You could use purrr::map_dfr to loop over the rows of your scenarios df. Requires some rewriting of your functions such that it takes the parameters itself as arguments. Additionally I simplified your code a bit.

EDIT Replaced the superseded pmap_dfr by pmap(...) |&gt; list_rbind(). Use ... to pass the arguments to simulate_inout.

  1. library(deSolve)
  2. library(ggplot2)
  3. library(purrr)
  5. model &lt;- function(t, y, p) {
  6.   N &lt;- y
  7.   r &lt;- p[[1]]
  8.   K &lt;- p[[2]]
  9.   
  10.   dN &lt;- r * N * (1 - N / K)
  11.   
  12.   list(dN)
  13. }
  14. times &lt;- 0:10
  16. simulate_inout &lt;- function(...) {
  17.   args &lt;- list(...)
  18.   
  19.   y0 &lt;- args[[&quot;N&quot;]]
  20.   p &lt;- args[c(&quot;r&quot;, &quot;K&quot;)]
  21.   
  22.   output &lt;- ode(y0, times, model, p)
  23.   
  24.   data.frame(args, output)
  25. }
  27. scenarios &lt;- expand.grid(
  28.   N = seq(0.5, 1.5, 0.2),
  29.   K = 1,
  30.   r = seq(0.2, 1, 0.2)
  31. )
  33. scenarios |&gt;
  34.   purrr::pmap(simulate_inout) |&gt;
  35.   list_rbind() |&gt; # or dplyr::bind_rows()
  36.   ggplot(aes(time, X1)) +
  37.   geom_path() +
  38.   facet_grid(r ~ N)

如何以tidyverse风格为deSolve包组织一系列模拟（场景）？<!-- -->

huangapple
  • 本文由 发表于 2023年8月5日 00:35:48
  • 转载请务必保留本文链接：https://go.coder-hub.com/76837768.html
  • desolve
  • ggplot2
  • r
  • tidyverse
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定