R中的滑动窗口，但步长小于采样率

Question

我正在尝试将滑动窗口应用于数据框，类似于rollapply和SlidingWindow的工作方式。但是，我想知道这些功能或其他功能中的任何一个是否可以使用（1）为小数，而（2）可以小于采样率的步骤（请参见以下示例）。这样的功能可能会更有效地执行以下操作：

library(dplyr)
beaver1.1 <- beaver1 %>% mutate(activ = 0)
beaver1.2 <- beaver1 %>% mutate(activ = 1,
                                temp = temp + 1)
beaver_combined <- rbind(beaver1.1, beaver1.2)
beaver_combined <- beaver_combined %>%
  mutate(time = time *.001)

window_freq <- .001
window_size <- .049
first_window <- min(beaver_combined$time)
last_window <- max(beaver_combined$time)
window_vector <- seq(first_window, last_window, window_freq)

windowing <- function(window_start){
  window_end <- window_start + window_size
  pullfrom <- beaver_combined %>%
    filter(between(time, window_start, window_end)) %>%
    group_by(activ) %>%
    summarise(temp = mean(temp)) %>%
    spread(activ, temp)
}

needsbinding <- mapply(windowing, window_vector)
do.call(rbind, needsbinding)

非常感谢！