屏蔽R中的最高设置位?

时间:2016-08-10 16:45:34

标签: r bit-manipulation

我在R.工作。我需要找到一个快速函数,它将屏蔽整数的最高设置位。例如:

# 6 binary is 110, this should turn into 010 which is 2
function_mask(6) = 2

# 8 in binary is 1000, this should turn into 0000
function_mask(8) = 0

这相当于减去最接近的2的较低功率。如果我能找到一个能够找到最接近的2的较低功率的快速功能,我将很高兴。例如:

# 6 in binary is 110, the MSB is 100
function_power_two(6) = 4
function_mask(6) = 6 - function_power_two(6) = 2

# 8 in binary is 1000, the MSB is 1000 which is 8 in base 10
function_power_two(8) = 8
function_mask(8) = 8 - function_power_two(8) = 0

我在R中找到了按位运算:例如,bitwShiftL和bitwShiftR。但是,我不知道如何在R中实现解决方案。

我见过其他语言的解决方案:JavaCC++。但是,我不知道如何在R中实现这些解决方案。

C ++中有使用Rcpp的解决方案,但Rcpp不支持大于32位的整数。我需要更大的整数。

4 个答案:

答案 0 :(得分:1)

你可以这样做:

function_mask <- function(x) {
    bits = intToBits(x)                 # Convert integer to binary vector
    ii = tail(which(bits > 0), n=1)     # Identify most significant bit
    bits[ii] = as.raw(0)                # Mask the most significant bit
    out = packBits(bits,type='integer') # Convert binary back to integer
    return(out)
}

测试:

function_mask(6) = 2
function_mask(8) = 0

答案 1 :(得分:1)

不确定它是否快速,但这是另一种可能性:

maskHighBit <- function(x){strtoi(sub("^1", "", R.utils::intToBin(x)), base=2)}

答案 2 :(得分:1)

此功能比the answer I posted earlier更快(4倍)。

pow2 <- c(0,1,2,4,8,16,32,64,128,256,512,1024)
function_mask <- function(x) x - pow2[findInterval(x, pow2)]

您可以根据需要制作pow2矢量,以应对更大的整数

答案 3 :(得分:0)

对于R解决方案:

function_mask <- function(x) {
  xb <-intToBits(x)
  highestbit <- length(xb) - match("01", rev(xb)) 
  x-2L^highestbit
}

将速度与其他答案进行比较,到目前为止我们看到这个速度最快。

function_mask1 <- function(x) {
  bits = intToBits(x)                 # Convert integer to binary vector
  ii = tail(which(bits > 0), n=1)     # Identify most significant bit
  bits[ii] = as.raw(0)                # Mask the most significant bit
  out = packBits(bits,type='integer') # Convert binary back to integer
  return(out)
}

maskHighBit <- function(x){
  strtoi(gsub("^1", "", R.utils::intToBin(as.integer(x))), base=2)}

library(microbenchmark)
microbenchmark(function_mask(112L), function_mask1(112L), maskHighBit(112L), times=1000)
#Unit: microseconds
#expr     min      lq      mean  median      uq      max neval cld
#function_mask(112L)  17.961  20.014  23.65080  23.092  24.632   57.475  1000 a  
#function_mask1(112L)  39.514  44.132  49.79744  47.724  49.777  107.765  1000  b 
#maskHighBit(112L) 108.791 114.435 127.53792 118.540 133.422 2054.189  1000   c